Quantcast

Page 3 of 7 FirstFirst 1234567 LastLast
Results 31 to 45 of 92

Thread: More Alternate Reality Stuff or Did the 32x Make Sense?

  1. #31
    Hero of Algol kool kitty89's Avatar
    Join Date
    Mar 2009
    Location
    San Jose, CA
    Age
    30
    Posts
    9,725
    Rep Power
    64

    Default

    Small update: I just noticed that Virtua Racing Deluxe and Shadow Squadron both do sample playback almost exclusively on the Genesis end, with the sole exception being the stereo panned zoom by on the Sega title screen of VR Deluxe. Everything else, including speech, is played back by the Z80 through the YM2612 DAC.
    Both use hybrid 68k+Z80 drivers, dedicating Z80 to PCM playback, with VR using the "32X standard J" driver and Shadow Squadron using good ol' SMPS 68k. VRD is only using 1 channel mixed to the DAC, but Shadow Squadron has one for music (drums) and one for SFX. (thus, SFX will only cut out other SFX and not music samples)
    Metal Head uses SMPS 68k with the DAC for music and PWM for SFX and speech (really terrible quality ), though there's also PCM effects in the sound test (seem to be variants of the PWM SFX actually), so I'm not sure what that's about.
    Sound engines according to: http://gdri.smspower.org/wiki/index....nd_Engine_List

    Some games do use PWM extensively, but honestly I can't say they really sound "better" for it. (maybe Knuckles Chaotix, but I doubt that wouldn't have sounded similarly good arranged for 1 less FM channel and 1 or 2 sample channels on the DAC )

    So, there you go, a good chunk of 32x games didn't need added sound hardware to do what they already do. Granted, lack of any info on the DMA sound crippled the usefulness of PWM anyway.

    On another note, this also implies that the only reasons for the difference in SVP VR and Deluxe sound-wise is up to ROM space (3 vs 2 MB) and programming. Virtua Racing SVP uses SMPS Z80, though there's a note about being banked too.
    Last edited by kool kitty89; 01-29-2013 at 04:32 AM.
    6 days older than SEGA Genesis
    -------------
    Quote Originally Posted by evilevoix View Post
    Dude it’s the bios that marries the 16 bit and the 8 bit that makes it 24 bit. If SNK released their double speed bios revision SNK would have had the world’s first 48 bit machine, IDK how you keep ignoring this.
    Quote Originally Posted by evilevoix View Post
    the PCE, that system has no extra silicone for music, how many resources are used to make music and it has less sprites than the MD on screen at once but a larger sprite area?

  2. #32
    I remain nonsequitur Shining Hero sheath's Avatar
    Join Date
    Jul 2010
    Location
    Texas
    Age
    42
    Posts
    13,313
    Rep Power
    130

    Default

    Well that is interesting. Virtua Racing Deluxe has a ton of samples playing in the music, the cars passing, the announcer, that all sound higher quality than the Genesis could be expected to put out alone. I just assumed it was using the 32X for the samples and the Genesis for the music and FM sound effects. Shadow Squadron is what I would have expected VR Deluxe to sound like with a purely Genesis driver. I'm not sure if there is a sample in the entire game, and the Genesis backgrounds for the stars and sun is quite bland. Somebody got fancy with Virtua Racing Deluxe.

    -edit-

    Also, I think my "Neptune" // "Jupiter" price comparison got lost in the mix up in the last page full of arbitrary assignments to what is and is not an add-on. One thing I think I missed in my own point is that no 32X means no 1994 launch in the West, end of story. The Saturn had all of the hardware allocated to its Japanese launch and a Jupiter launch alongside that would have just made the schedule tighter. For whatever reason the 32X launching in the US alongside the Saturn in Japan wasn't as big of a strain on component makers. Otherwise why didn't Sega launch the Saturn worldwide in 1994, we know they were desperate to and in fact did rush the Western launch as a result. So, no 1994 US Saturn = no 1994 US Jupiter and no 1994 US Neptune either.

    Now for the price comparison schedule:

    1994:
    32X
    $160 total cost for existing Genesis owners.

    "Neptune"
    N/A practically speaking, $260 combined Genesis 32X cost that Fall.

    "Jupiter"
    N/A practically speaking, $250 postulated, add cost of extra controller at minimum to actual cost.

    Spring/Summer 1995:
    32X
    $100

    "Neptune"
    $150 - no added controller costs for existing Genesis owners

    "Jupiter"
    $250-200 - plus Saturn controller cost at minimum

    By September the Saturn was $300 in the real world US retail, so:
    Fall 1995
    32X
    $80

    "Neptune"
    $120-100

    "Jupiter"
    $200-150?

    As you can see, I don't see the Jupiter solution as practical for the low end as the real world 32X plus Neptune would have been. Jupiter would be in a totally different class of hardware, and I'm not even considering the eventual cost of adding the CD-ROM, which historically always costs more than just buying the full blown system alone. I also don't see the infatuation with stand alone hardware versus add-ons as entirely logical. It seems hung up in now antiquated journalistic opinions of add-ons and less practical and cost effective in general.

    For another comparison, here is what I spent on console hardware since starting my lawn mowing job at age 11:
    1988
    Sega Master System - $100 (my dad pitched in about $100 extra for a controller and a couple of games)
    1989
    NES - $100
    Genesis $189 (It wasn't long before I had the 3-button Joystick)
    1990
    TurboGrafx-16 $150 (I sold my NES and SMS and all games and controllers to make it)
    1992
    Sega CD $300 (Sold my TG16 and all games to get it down to about $100 my cost I think)
    1994
    32X $160
    1995

    Saturn $300 (Bought my Hori Joystick alongside it, total with VF2 was $400)
    1998
    PS1 $200
    1999
    Dreamcast $200 (I had traded my PS1 by then for store credit on games)
    2001
    PS2 $300
    2003
    Xbox $200

    Okay, so that is fourteen years of hardware purchases for an average cost of $157 per year. If Sega hadn't made the Sega CD in 1992 and the 32X in 1994 I almost definitely would have bought a SNES somewhere in there. I also wasn't counting PC hardware, which I bought and expanded for about $300 every few years starting in 1996. At any rate, I have never been in an exceptionally high income household but I always found the money to buy hardware I wanted that had games I wanted, and then either rented or bought the games over time.

    With this breakdown I just do_not see how two add -ns in five years for the Genesis was a bad idea from a marketing perspective.
    Last edited by sheath; 01-29-2013 at 04:16 PM.
    "... If Sony reduced the price of the Playstation, Sega would have to follow suit in order to stay competitive, but Saturn's high manufacturing cost would then translate into huge losses for the company." p170 Revolutionaries at Sony.

    "We ... put Sega out of the hardware business ..." Peter Dille senior vice president of marketing at Sony Computer Entertainment

  3. #33
    Master of Shinobi
    Join Date
    Sep 2012
    Posts
    1,180
    Rep Power
    31

    Default

    Ideally, the 32x would've been never made, and instead they should've dedicated resources for a backwards compatibility adapter for the Saturn (same concept as the Power Base Converter, just actually containing MD hardware).

  4. #34
    Bring on the noise! WCPO Agent Bones Justice's Avatar
    Join Date
    Jul 2011
    Posts
    946
    Rep Power
    34

    Default

    Quote Originally Posted by zyrobs View Post
    Ideally, the 32x would've been never made, and instead they should've dedicated resources for a backwards compatibility adapter for the Saturn (same concept as the Power Base Converter, just actually containing MD hardware).
    That wouldn't have been ideal for me. I would never give up the time that I spent playing Virtua Fighter or Virtua Racing on 32X with my friends during that time, or Doom, Star Wars, and Shadow Squadron. Sure, I could have waited another year for something "better" (and more expensive, by the way) but I'd rather be playing games now than waiting for something later.
    Slayer-1: "Minus 210 degrees...minus 215...minus 220. Why are you still standing?"

    Bones: "My will...will not die."

    https://m.youtube.com/watch?v=AJ-iQfP3mFs

  5. #35
    Outrunner Greg2600's Avatar
    Join Date
    Apr 2010
    Posts
    689
    Rep Power
    18

    Default

    Backwards compatibility, at least I feel, is a red herring. As long as you go enough years in between new systems, softening the $$$ burden on mom's and dad's, you will be fine. The first bwc system I ever had was the PS2 (other than super gameboy) and never really cared. But I was someone who didn't throw or trade in systems. I kept them. The worse thing is to cut the system off from people who aren't ready to upgrade!!! That's what SEGA did.

    Second point is that the 32X was needed because it provided capabilities the Genesis could not. Well, the SNES couldn't either, until they started putting the chips in the carts. That was huge, because I was one of the people who switched to SNES and stayed with it for years until the N64. I left SEGA behind. SEGA should have spent the 32X/Neptune development money on those add-on chips.

    Lastly, as sheath detailed, the 32X was just another large price tag SEGA expected parents to pay. Utter foolishness. Atari Pac-Man in your sky lodge elitist stupidity.

    So as I've written many times, the 32X was a giant gaffe by SEGA. It steered more people away from them. Plus their subsequent dropping of major Genesis game support with the arrival of the Saturn lost more people, not to mention easy money. SofA just simply made a mess of it all. Had they done what Nintendo did, they would have had a chance against Sony. Maybe.

  6. #36
    Hero of Algol kool kitty89's Avatar
    Join Date
    Mar 2009
    Location
    San Jose, CA
    Age
    30
    Posts
    9,725
    Rep Power
    64

    Default

    Quote Originally Posted by sheath View Post
    Well that is interesting. Virtua Racing Deluxe has a ton of samples playing in the music, the cars passing, the announcer, that all sound higher quality than the Genesis could be expected to put out alone. I just assumed it was using the 32X for the samples and the Genesis for the music and FM sound effects. Shadow Squadron is what I would have expected VR Deluxe to sound like with a purely Genesis driver. I'm not sure if there is a sample in the entire game, and the Genesis backgrounds for the stars and sun is quite bland. Somebody got fancy with Virtua Racing Deluxe.
    Remember, the Genesis alone is capable of higher quality PCM sound capabilities than seen in pretty much all 32x games in general . . . it's only a matter of programming and ROM space. The few games making heavy use of sample synth on 4+ channels for the main music would be the exceptions (most of which also are less SH2 intensive 2D games).

    VR Deluxe actually uses simpler PCM sound than Shadow Squadron. It's just plain 1 channel PCM (or maybe DPCM) played back by the Z80 (it cuts out to play other samples), and not even at sample rates that uncommon for the system (not sure, but it doesn't sound far off from 11-16 KHz). The main issue is that very few games used lots of speech samples of that quality due to ROM space, especially if using uncompressed 8-bit PCM. (4-bit DPCM a la SMPS or uncompressed 4-bit linear PCM a la GEMS would sound considerably worse at the same sample rates . . . granted, taking half the space and at least better than 8-bit at 1/2 the rate -depending on the bitrate and sample sound in question)

    Also remember that any distortion caused by uneven timing (including that from the 68k accessing the Z80 bus -really nasty in SFII) is greatly diminished at high sample rates (the error is spread out, diffused), so as long as the driver itself is capable of those sample rates without choking the CPU, then you'll get progressively less distorted playback on top of the higher quality samples in general. (albeit, this also begs the question of why games didn't tend to just scale up lower rate samples too, or at least done so more often -there's several games with very obviously higher sample rate drums with very low rate -and distorted- speech)


    Anyway, with the resource of the SVP available, the only disadvantages of using the YM2612 DAC would be mono (or hard pan -all PCM channels panning hard L/R/center) and using up an FM channel. There's the potential for higher res and higher max sample rates for mixing on PWM, but without DMA being used there's very limited practical use of that, and 9 or 10-bit resolution on PWM still won't sound far off from 8-bit output (especially with the other quality trade-offs of using PWM output in general, especially below ~18 kHz where the squeal/aliasing becomes obvious).
    The MD DAC can capably handle up to 26 kHz (above that and you start missing writes), and a 22-26 kHz PCM playback driver ding a simple 8-bit stream via Z80 to complement the SVP's own sound mixing (mixing to SVP DRAM and having Z80 read from that) should have been fairly realistic and a decent workaround for not having built-in DMA sound on the SVP. Granted, games maxing out the SVP DSP for 3D computation or such would be limited to more typical genesis driven sound. (which, gain, VR Deluxe and several other 32x games already do)

    Darxide does all its samples with PWM, but it certainly doesn't sound impressive . . . and no more so than Metal head for that matter. (some of the best samples on the 32x are played by the MD alone )
    Blackthorne is probably the only example that would have been impractical to do on the MD alone music-wise . . . but only then because it totally relied on a (still low quality) MOD player of sorts without any use of the onboard synth hardware. Techically, it could do that quality music using the 68k in a demo (mono), but not in-game. (like Atari ST demos, but using a real DAC rather than the YM2149 PSG) Only 1 MD game even attempted that (Toy Story) and did it quite poorly considering what the ST managed.

    -So if I didn't imply this well enough: SVP sound mixing played by the Z80 ~22 kHz could have sounded significantly BETTER than anything actually done on the 32x.

    Also, I think my "Neptune" // "Jupiter" price comparison got lost in the mix up in the last page full of arbitrary assignments to what is and is not an add-on. One thing I think I missed in my own point is that no 32X means no 1994 launch in the West, end of story. The Saturn had all of the hardware allocated to its Japanese launch and a Jupiter launch alongside that would have just made the schedule tighter. For whatever reason the 32X launching in the US alongside the Saturn in Japan wasn't as big of a strain on component makers. Otherwise why didn't Sega launch the Saturn worldwide in 1994, we know they were desperate to and in fact did rush the Western launch as a result. So, no 1994 US Saturn = no 1994 US Jupiter and no 1994 US Neptune either.
    Assuming Sega's $200 price projection for the Neptune was a genuine reflection for selling at cost, then the Jupiter really shouldn't have been much more than that anyway. The custom chips were bigger (SCU+VDP1+VDP2 vs MD VDP/ASIC+SuperVDP+32xASIC), but the really big difference would have been RAM, and motherboard complexity would have been fairly similar too.
    Given the pricing of the Saturn in Japan, I still think $250 should have been do-able for "Jupiter" (especially with the low-RAM removed -so 1MB SDRAM main, 512k sound, 512k+512k+512k video, 2 MB SDRAM and 512kB cheap DRAM total). Given Sega's manufacturing volume and industry negotiation/connection advantages as well as them selling at cost, compared to the Jaguar's $250 price point (low volume, poorer negotiation/connections, and above cost), then this should have been very realistic.

    This is on top of the many other advantages the Jupiter provided. (and monetary/resource advantages -especially software development logistics)

    However, if retail/consumer cost was the more important factor, than SVP was the way to go by far. The add-on would have been very cheap and potentially get a head-start as a direct replacement for standalone VR (bundled launch at same price), and an integrated (with MD) version not only would have been much cheaper than Neptune, but should have been cheap enough to integrate as a standard feature of all late model MD/Genesis. (again, once consumer/developer interest merited it -otherwise keeping the slightly cheaper to build, more profitable base model would have been better)

    As you can see, I don't see the Jupiter solution as practical for the low end as the real world 32X plus Neptune would have been. Jupiter would be in a totally different class of hardware, and I'm not even considering the eventual cost of adding the CD-ROM, which historically always costs more than just buying the full blown system alone. I also don't see the infatuation with stand alone hardware versus add-ons as entirely logical. It seems hung up in now antiquated journalistic opinions of add-ons and less practical and cost effective in general.
    You are ONLY looking at the consumer pricing context though, and on that front is where the SVP wins by miles anyway.

    Compared to the Jupiter, it's cheaper, but it also sucks away software development resources, is way too weak to competently compete in the long run, and has far poorer consumer value in general than the Jupiter. 32x is just a final boost to an aging platform, while Jupiter is the introduction of a progressive (semi-modular) new platform, a real investment.

    Targeting the low-end isn't the point here, but targeting the lower-end of the "next gen" new market segment is. For the low-end, the 32x was too costly for consumers and Sega/3rd parties to invest in with too many compromises and conflicts to really be practical. SVP would have been a slight conflict with Saturn by comparison, but cheap enough all around to at least make sense for the low-end.

    This is also the reason I don't really favor SVP either: making preparations to facilitate the Saturn's mass-market success was more important than trying to extend the breadth of capabilities of the MD/CD. (keep supporting the "16-bit" market as the primary interest for the time being in either case, of course, but aim at focusing on the full next-gen hardware for the future, and push projects onto the new platform that would be overly compromised by the old -along with doing a few cross-platform releases where desirable)
    Jupiter introduces both consumers and developers to the Saturn architecture sooner, and provides a lower initial investment for consumers to get onboard (buy CD later, especially after prices drop) rather than introducing a totally separate platform needing its own specific programming and learning curve, and the same investment for consumers and developers to make when it came time for the Saturn in general.

    Nintendo's claims of cutting back on certain things (including Star Fox 2) for a "fresh start" with 3D on the N64 has merit, though there was probably more middle ground to be found than Nintendo exploited. (and this is where you could argue having both SVP and Jupiter could have worked out well . . . and IMO it could have, but there'd have been more potential with screw-ups without the right marketing/management to back it up -still nowhere near what the 32x did on that front)


    This is aside from the mind-bogglingly stupid decision to launch the Saturn in May 1995 . . . that was just weird all around. (grated, even that is something Jupiter could have smoothed out somewhat, since you'd have all those cart games the Saturn could play -confusing developers and retailers along with Sony's price war would still have sucked though)

    If Sega hadn't made the Sega CD in 1992 and the 32X in 1994 I almost definitely would have bought a SNES somewhere in there. I also wasn't counting PC hardware, which I bought and expanded for about $300 every few years starting in 1996. At any rate, I have never been in an exceptionally high income household but I always found the money to buy hardware I wanted that had games I wanted, and then either rented or bought the games over time.
    OK, but how about this: did you refrain from buying an SNES because you spent too much on Sega stuff, or because the Sega stuff you bought made the SNES superfluous?

    TBH, I can't really see how it could have been the latter, since the main reason to get an SNES would have been for the unique/exclusive software available, and the games offered by MCD/32x hardly addressed that better than the MD alone . . . buying a system "for the hardware" just doesn't make sense --sure, some idiots who fall for marketing might do it (just as easily could fall for software-specific marketing), or the much less common but more legitimate real tech-heads who take the risk based on platform technical potential, but the latter is so small that there's really no marketing catering to that for consoles.
    6 days older than SEGA Genesis
    -------------
    Quote Originally Posted by evilevoix View Post
    Dude it’s the bios that marries the 16 bit and the 8 bit that makes it 24 bit. If SNK released their double speed bios revision SNK would have had the world’s first 48 bit machine, IDK how you keep ignoring this.
    Quote Originally Posted by evilevoix View Post
    the PCE, that system has no extra silicone for music, how many resources are used to make music and it has less sprites than the MD on screen at once but a larger sprite area?

  7. #37
    Hero of Algol kool kitty89's Avatar
    Join Date
    Mar 2009
    Location
    San Jose, CA
    Age
    30
    Posts
    9,725
    Rep Power
    64

    Default

    Quote Originally Posted by Greg2600 View Post
    Backwards compatibility, at least I feel, is a red herring. As long as you go enough years in between new systems, softening the $$$ burden on mom's and dad's, you will be fine. The first bwc system I ever had was the PS2 (other than super gameboy) and never really cared. But I was someone who didn't throw or trade in systems. I kept them. The worse thing is to cut the system off from people who aren't ready to upgrade!!! That's what SEGA did.
    TBH, it really depends on the context. Compatibility isn't as important as on PCs or such, but if it can be done without excessive R&D overhead and without significantly compromising price point or features, then it's certainly worthwhile overall.

    You have several areas where this is useful:
    1. catering to existing users that might want to retire or sell/trade that system . . . if not a monetary concern, than the convenience at least is significant. (not having to have "one more" system left hooked up to the TV -space/clutter/connectivity limits)
    2. people who still haven't bought the old system but are very interested in the software available while also being interested in getting a new one.

    Category 2 is usually a small but not totally insignificant chunk of the market, varying somewhat by just how long the "old" system has been on the market and how well it's still stilling (both software and hardware wise).
    Number 2 could very well mean the difference from a buyer opting to go for the old machine alone and wait until later to buy the new one, while backwards compatibility might have been the breaking point for this decision. (making them willing to spend the extra cash sooner to buy 2 systems in one)

    I started a thread on this issue a long time ago, though it's been discussed at-length a few other times too. http://www.sega-16.com/forum/showthr...ty-Good-or-Bad
    In fact: I'm going to quote this post there now, so we can move further discussion on backwards compatiblity there.


    There's also the separate issue of forwards compatibility addressed in this thread with the whole Jupiter idea: intentionally scaling back a system while retaining support for expansion to "full" spec of the original design. (and allowing software to work on both as well -and programming to be very similar)


    The 32x isn't so much a backwards compatible console, as it is an upgrade that must be compatible to make any sense at all. Had Sega actually designed the Saturn itself expressly around being MD/MCD compatible, then it should have been far more capable and efficiently configured than the 32x itself. Without going into actual specifics, it probably could have been about as capable (overall) and cost effective as the Saturn itself, though that's only in the context that the Saturn could have been streamlined a fair bit already. (and a streamlined Saturn-like system would have been more cost-effective than either -aside from the marketing advantages of compatibility)
    OTOH, going way back to the Sega CD itself, had it been designed as a compatible standalone console rather than an add-on, it could have been more cost effective and much more capable overall (and made it more feasible for the Saturn to be built around compatibility too).

    Hence part of my premise in this thread:
    http://www.sega-16.com/forum/showthr...-Reality-stuff
    Last edited by kool kitty89; 01-29-2013 at 07:52 PM.
    6 days older than SEGA Genesis
    -------------
    Quote Originally Posted by evilevoix View Post
    Dude it’s the bios that marries the 16 bit and the 8 bit that makes it 24 bit. If SNK released their double speed bios revision SNK would have had the world’s first 48 bit machine, IDK how you keep ignoring this.
    Quote Originally Posted by evilevoix View Post
    the PCE, that system has no extra silicone for music, how many resources are used to make music and it has less sprites than the MD on screen at once but a larger sprite area?

  8. #38
    I remain nonsequitur Shining Hero sheath's Avatar
    Join Date
    Jul 2010
    Location
    Texas
    Age
    42
    Posts
    13,313
    Rep Power
    130

    Default

    Quote Originally Posted by kool kitty89 View Post
    Remember, the Genesis alone is capable of higher quality PCM sound capabilities than seen in pretty much all 32x games in general . . . it's only a matter of programming and ROM space. The few games making heavy use of sample synth on 4+ channels for the main music would be the exceptions (most of which also are less SH2 intensive 2D games).
    Right, the Genesis (or SNES for that matter) wasn't going to keep up with the 32X in simultaneous mid-to-high quality samples. For more normal sample usage the Genesis could and apparently did sound better than what developers knew about on the 32X for single sample usage and drums.

    Quote Originally Posted by kool kitty89 View Post
    VR Deluxe actually uses simpler PCM sound than Shadow Squadron. It's just plain 1 channel PCM (or maybe DPCM) played back by the Z80 (it cuts out to play other samples), and not even at sample rates that uncommon for the system (not sure, but it doesn't sound far off from 11-16 KHz). The main issue is that very few games used lots of speech samples of that quality due to ROM space, especially if using uncompressed 8-bit PCM. (4-bit DPCM a la SMPS or uncompressed 4-bit linear PCM a la GEMS would sound considerably worse at the same sample rates . . . granted, taking half the space and at least better than 8-bit at 1/2 the rate -depending on the bitrate and sample sound in question)

    Also remember that any distortion caused by uneven timing (including that from the 68k accessing the Z80 bus -really nasty in SFII) is greatly diminished at high sample rates (the error is spread out, diffused), so as long as the driver itself is capable of those sample rates without choking the CPU, then you'll get progressively less distorted playback on top of the higher quality samples in general. (albeit, this also begs the question of why games didn't tend to just scale up lower rate samples too, or at least done so more often -there's several games with very obviously higher sample rate drums with very low rate -and distorted- speech)
    Wouldn't these timing issues be greatly diminished by most of the game code running on the 32X's SH2s anyway? If I recall the timing issue has to do with the Z80 halting the 68000 to play a sample or vice versa. No, that was blocking access to RAM and the cartridge Bus I think. At any rate, wouldn't the 68000 not being as busy as it would be with a Genesis game make the timing issue significantly easier for samples?

    Mortal Kombat II plays more, and better sounding, simultaneous samples than SNES MKII does and I think it even sounds clearer and cuts out less than Genesis MKIII. Dangit I still need to make that MKII comparison video.

    Quote Originally Posted by kool kitty89 View Post
    Anyway, with the resource of the SVP available, the only disadvantages of using the YM2612 DAC would be mono (or hard pan -all PCM channels panning hard L/R/center) and using up an FM channel. There's the potential for higher res and higher max sample rates for mixing on PWM, but without DMA being used there's very limited practical use of that, and 9 or 10-bit resolution on PWM still won't sound far off from 8-bit output (especially with the other quality trade-offs of using PWM output in general, especially below ~18 kHz where the squeal/aliasing becomes obvious).
    The MD DAC can capably handle up to 26 kHz (above that and you start missing writes), and a 22-26 kHz PCM playback driver ding a simple 8-bit stream via Z80 to complement the SVP's own sound mixing (mixing to SVP DRAM and having Z80 read from that) should have been fairly realistic and a decent workaround for not having built-in DMA sound on the SVP. Granted, games maxing out the SVP DSP for 3D computation or such would be limited to more typical genesis driven sound. (which, gain, VR Deluxe and several other 32x games already do)
    Oh yeah, I forgot to ask also, where are you getting this information for the "32X Standard J" driver that Virtua Racing Deluxe uses? I can't find any info on it, let alone its sample capabilities using Genesis DAC exclusively.

    Quote Originally Posted by kool kitty89 View Post
    Darxide does all its samples with PWM, but it certainly doesn't sound impressive . . . and no more so than Metal head for that matter. (some of the best samples on the 32x are played by the MD alone )
    Blackthorne is probably the only example that would have been impractical to do on the MD alone music-wise . . . but only then because it totally relied on a (still low quality) MOD player of sorts without any use of the onboard synth hardware. Techically, it could do that quality music using the 68k in a demo (mono), but not in-game. (like Atari ST demos, but using a real DAC rather than the YM2149 PSG) Only 1 MD game even attempted that (Toy Story) and did it quite poorly considering what the ST managed.

    -So if I didn't imply this well enough: SVP sound mixing played by the Z80 ~22 kHz could have sounded significantly BETTER than anything actually done on the 32x.
    Yeah, I don't think we were going to see 22Khz samples on cartridges at that time though. It would have been great if the only SVP game in existence sounded anywhere near as clear as Virtua Racing Deluxe too.

    Quote Originally Posted by kool kitty89 View Post
    Assuming Sega's $200 price projection for the Neptune was a genuine reflection for selling at cost, then the Jupiter really shouldn't have been much more than that anyway. The custom chips were bigger (SCU+VDP1+VDP2 vs MD VDP/ASIC+SuperVDP+32xASIC), but the really big difference would have been RAM, and motherboard complexity would have been fairly similar too.
    Given the pricing of the Saturn in Japan, I still think $250 should have been do-able for "Jupiter" (especially with the low-RAM removed -so 1MB SDRAM main, 512k sound, 512k+512k+512k video, 2 MB SDRAM and 512kB cheap DRAM total). Given Sega's manufacturing volume and industry negotiation/connection advantages as well as them selling at cost, compared to the Jaguar's $250 price point (low volume, poorer negotiation/connections, and above cost), then this should have been very realistic.
    $250 would have been doable for Jupiter in 1995, probably $200 by the time the Saturn and PS1 were launched nationally. 1994 is a wash though, and as we have discussed before Sega needed the 32X for the boost in revenue that could then be rolled over to the Saturn launch the next year. We have also already discussed that the Sega CD 2 and Genesis 2, Sonic 3 & Knuckles, the SVP and probably Sega's most expensive/impressive software lineup to date failed to keep that year on year decline in revenue from hitting Sega harder than Nintendo.

    Quote Originally Posted by kool kitty89 View Post
    This is on top of the many other advantages the Jupiter provided. (and monetary/resource advantages -especially software development logistics)

    However, if retail/consumer cost was the more important factor, than SVP was the way to go by far. The add-on would have been very cheap and potentially get a head-start as a direct replacement for standalone VR (bundled launch at same price), and an integrated (with MD) version not only would have been much cheaper than Neptune, but should have been cheap enough to integrate as a standard feature of all late model MD/Genesis. (again, once consumer/developer interest merited it -otherwise keeping the slightly cheaper to build, more profitable base model would have been better)
    Not only retail/consumer cost, it's the total package. SVP doesn't address the color issue or makes it worse, isn't powerful enough to even approach 3DO/Jag level 3D, and cannot likely address the audio problem any better than Genesis carts could without adding more hardware to the cart/expansion. Then there is the games. Apparently the 32X was powerful enough to have stuff like Space Harrier and After Burner, not to mention Doom, Star Wars Arcade and Virtua Racing Deluxe, hastily slapped together for the break neck launch. Do we think that all of these games would have turned out well with a similar amount of resources spent (ignoring the 15-color problem)? I think the SVP is far less versatile in what kinds of games could be enhanced by it.

    Quote Originally Posted by kool kitty89 View Post
    You are ONLY looking at the consumer pricing context though, and on that front is where the SVP wins by miles anyway.

    Compared to the Jupiter, it's cheaper, but it also sucks away software development resources, is way too weak to competently compete in the long run, and has far poorer consumer value in general than the Jupiter. 32x is just a final boost to an aging platform, while Jupiter is the introduction of a progressive (semi-modular) new platform, a real investment.

    Targeting the low-end isn't the point here, but targeting the lower-end of the "next gen" new market segment is. For the low-end, the 32x was too costly for consumers and Sega/3rd parties to invest in with too many compromises and conflicts to really be practical. SVP would have been a slight conflict with Saturn by comparison, but cheap enough all around to at least make sense for the low-end.
    In 1994 I didn't see the 32X as a stop gap, I saw it as an inexpensive way to get the Genesis and Sega CD into the range of 3DO and Jaguar 3D while enhancing everything else about the Genesis that the Super Nintendo could possibly win at. SVP basically would have given an answer to SNES SFX, which I already thought the Sega CD beat handily. The total value proposition of SVP versus 32X favors the 32X ESPECIALLY if Sega hadn't canceled it six months later but even with its existing library.

    Now if come mid 1995, with no real world 32X, Sega offered me a Jupiter for $250 or a Saturn for $400, and then dropped the prices to $200 and $300 that Fall we are just talking about a new system launch and nothing else. That doesn't do anything for 1994, the Genesis, or the Sega CD. More on that later in relation to your question about my not buying a Super Nintendo.

    Quote Originally Posted by kool kitty89 View Post
    This is also the reason I don't really favor SVP either: making preparations to facilitate the Saturn's mass-market success was more important than trying to extend the breadth of capabilities of the MD/CD. (keep supporting the "16-bit" market as the primary interest for the time being in either case, of course, but aim at focusing on the full next-gen hardware for the future, and push projects onto the new platform that would be overly compromised by the old -along with doing a few cross-platform releases where desirable)
    Jupiter introduces both consumers and developers to the Saturn architecture sooner, and provides a lower initial investment for consumers to get onboard (buy CD later, especially after prices drop) rather than introducing a totally separate platform needing its own specific programming and learning curve, and the same investment for consumers and developers to make when it came time for the Saturn in general.
    Yeah, I don't think focusing exclusively on the new hardware was the smartest move in 1994-95. The Genesis was going to remain Sega's most mass market product for at least a couple of years while the next generation limped into shape. In my ideal scenario the Genesis and Sega CD would have been supported through 1998 and the 32X makes that more likely to have been interesting, not less. Simultaneous 32X/Saturn releases shouldn't have cost that much more resources considering companies ported games to the Saturn using a single individual code jockey. The only thing that Jupiter adds to developer familiarity with the Saturn is the VDPs +DSP. I have mentioned that the 32X already having a 2D / 3D hardware VDPs in its setup is conceptually like the Saturn setup. The only thing the Genesis VDP couldn't easily be used for is scaling floors/ceilings, but they ended up being relatively limited to the peculiar Saturn setup anyway.

    Quote Originally Posted by kool kitty89 View Post
    [snip]
    OK, but how about this: did you refrain from buying an SNES because you spent too much on Sega stuff, or because the Sega stuff you bought made the SNES superfluous?

    TBH, I can't really see how it could have been the latter, since the main reason to get an SNES would have been for the unique/exclusive software available, and the games offered by MCD/32x hardly addressed that better than the MD alone . . . buying a system "for the hardware" just doesn't make sense --sure, some idiots who fall for marketing might do it (just as easily could fall for software-specific marketing), or the much less common but more legitimate real tech-heads who take the risk based on platform technical potential, but the latter is so small that there's really no marketing catering to that for consoles.
    I was and still am predominantly interested in Action-Arcade style games, not RPGs and not console exclusive franchises. The 32X was perfect for my tastes. Space Harrier and After Burner seeing their first "Arcade perfect" releases on the 32X plus Virtua Racing Deluxe alone took the SNES library out of my consideration. Say what you will about Cosmic Carnage but it gave me all kinds of hope for Mortal Kombat II and Super Street Fighter II, not to mention NEO GEO ports. By late 1994 the only SNES exclusive games I really was tempted by was Super Double Dragon, Star Fox and Super Mario Kart. Nintendo's other mainstay franchises were and are of no interest to me. Even today, when Super Nintendo games are relatively cheaper than they were at the time, I only own about ten more of them than 32X games and I use the 32X a lot more.
    "... If Sony reduced the price of the Playstation, Sega would have to follow suit in order to stay competitive, but Saturn's high manufacturing cost would then translate into huge losses for the company." p170 Revolutionaries at Sony.

    "We ... put Sega out of the hardware business ..." Peter Dille senior vice president of marketing at Sony Computer Entertainment

  9. #39
    Hero of Algol kool kitty89's Avatar
    Join Date
    Mar 2009
    Location
    San Jose, CA
    Age
    30
    Posts
    9,725
    Rep Power
    64

    Default

    Quote Originally Posted by sheath View Post
    Right, the Genesis (or SNES for that matter) wasn't going to keep up with the 32X in simultaneous mid-to-high quality samples. For more normal sample usage the Genesis could and apparently did sound better than what developers knew about on the 32X for single sample usage and drums.
    SNES could do better, for the most part (compared to existing 32x stuff), with the exception of having the forced heavy filtering on everything. (technically, the SPC unit could do better than anything the 32x historically put out . . . though an SH2 mixing to DMA sound could certainly do a lot of things the SPC unit can't . . . or at least normally can't)
    The is the major limit of the 64kB sample RAM and the nasty bottleneck for streaming added samples, so that's a realistic consideration too, and the fixed sample format (no option to use 16 or 8-bit PCM vs 4-bit vs 2-bit ADPCM, CVSD, or some custom)

    Wouldn't these timing issues be greatly diminished by most of the game code running on the 32X's SH2s anyway? If I recall the timing issue has to do with the Z80 halting the 68000 to play a sample or vice versa. No, that was blocking access to RAM and the cartridge Bus I think. At any rate, wouldn't the 68000 not being as busy as it would be with a Genesis game make the timing issue significantly easier for samples?
    It's all about being able to do cycle counting and precisely timing code to do the same thing at a consistent rate without deviating from that timing. (this involves knowing the execution times of the instructions you're using on a specific processor at the specific speed its running) This sort of programming is also the ONLY way to write any working software on the Atari 2600.

    An alternative to that is using hardware timed interrupts or polling a timer, but both of those (especially the latter) require more overhead than good, cycle timed code, more so on some CPUs than others. (but it's a lot easier to accomplish) The problem is that the Z80 in the MD isn't hooked up to such a timer controlling interrupts (the YM2612 supplies timers well suited to this though, just not hooked up), though the Z80 can poll the YM timers, but that eats up a good bit more CPU time than even interrupts. (GEMS uses timer polling to play PCM) -On the note of interrupts, the 6502 is extremely fast at handling interrupts (in large part due to having so few registers), so doing interrupt driven playback on the PC Engine can be done using only a small fraction of CPU time (using the onboard 7 kHz timer takes only ~5% CPU time to handle PCM playback) and hucards could potentially add more capable timers onboard as well (interrupt input on the cart slot). Also note the NES has no timer interuppts to facilitate this. (at least without added hardware on cart)

    Aside from managing good timed code, there's the issue of other influences screwing up timing: having games using long, sustained periods of DMA (loading VRAM data in long, uninterrupted blocks) can throw things off and make the Z80 miss DAC writes, leading to the characteristic distortion seen in much of the MD's sample playback. This is also seen in games using 68k sound drivers (dedicating Z80 to sample playback) where the 68k spends too much time working on the Z80 bus (updating YM2612/PSG registers) and causing missed writes, and some games have both problems. The solutions to these are: break up DMA into small enough chunks to avoid huge blocks in Z80 access, and avoid letting the 68k stay on the Z80 bus too long in general. (or, take the work to actually make a pure Z80 sound driver also capable of good PCM playback) The absolute worst case is with the likes of preSMPS Z80 drivers that halt music to play samples. (some SMS games actually halt gameplay for the same reason: need to switch to cycle timed code that can't cope with other tasks being processed)
    SFII SCE seems to mostly (or only) suffer from 68k access conflicts: it uses a 68k sound driver and the samples are just as distorted in the sound test as in-game (sound test should have no DMA issues). Plus, the samples sound even if you halt the 68k immediately after selecting a sample to play. (works on emulators and real hardware)

    Also, as noted before, using higher playback rates in general (assuming you've got the code to handle it) will exhibit less audible distortion than the same sound driver playing back at lower sample rates. (which, again, begs the question why not at least scale up the lower rate stuff to minimize that -and mind, you, simple double/triple/etc scaling would take virtually no added overhead, unlike useful scaling needed for pitch/note control . . . some games probably do do that though)

    Mortal Kombat II plays more, and better sounding, simultaneous samples than SNES MKII does and I think it even sounds clearer and cuts out less than Genesis MKIII. Dangit I still need to make that MKII comparison video.
    MK I and II both used Shaun Hollingworth's Krysalis sound engine (the one used in all games Matt Furniss did the music for). That engine isn't foolproof, obviously, given the distortion seen in the first game. So, either the engine itself was tweaked/improved somewhat, or it's more a matter of other programming issues being avoided (including DMA conflicts) and/or the use of higher sample rate playback making the missed writes less noticeable.
    That sound driver was consistently good at doing sampled drums though.

    MKIII wasn't done by probe, and uses the GEMS sound engine instead (hence the weaker arrangements and strict use of arcade compositions -rather than Matt Furniss's meddling with things and swapping in his own awesome original stuff ). GEMS only supports 1 PCM channel, hence the "cutting out" problem.

    Interestingly enough, EA sound engines tended to handle samples fairly well, though there's notable distortion in some things too (again, mostly in-game and with lower sample rate stuff). Madden '96 always sticks with the streaming intro music, speech, and great sounding sampled percussion in the synth (and pretty nice compositions from Tommy Tallarico). Actual use of the sound engines (music wise) was kind of inconsistent (often similar to GEMS stuff bland/grating) which I assume is partially due to the typical sound design techniques used in American studios (PC games and arcades had some of the same problems using FM synth with many US devs). That said, several of EA's sound drivers were programmed by some pretty notable sound programmers. In the case of Madden '96, it was Rob Hubbard, well known for his home computer music in Europe. And, to be fair, EA sound engines on average tend to at least have better sounding music than GEMS on average.

    Several of those EA engines were also capable of relatively decent sample rates and multiple sample channels. (Skitchin' supports 4 channels iirc and at least one scaled for pitch as a MOD player -albeit only using a pretty limited guitar sample that gets strained when pushed to far out of its native sample rate . . . it sounds like it's at least ~11 kHz, I'm not super good at guessing that in general though)
    https://www.youtube.com/watch?v=e9AYz6CIq2E (madden '96 intro plus menu music)

    Oh yeah, I forgot to ask also, where are you getting this information for the "32X Standard J" driver that Virtua Racing Deluxe uses? I can't find any info on it, let alone its sample capabilities using Genesis DAC exclusively.
    http://gdri.smspower.org/wiki/index....nd_Engine_List
    Also the source for pretty much all other mentions of sound engines. (though some were also pointed out by other members on here)


    Yeah, I don't think we were going to see 22Khz samples on cartridges at that time though. It would have been great if the only SVP game in existence sounded anywhere near as clear as Virtua Racing Deluxe too.
    I'm talking about 22 kHz (or 26 kHz) for mixing purposes mostly, which is a decent output rate for doing software mixed/scaled samples (MOD players and such). 26 is the hard limit for the DAC (not the Z80), you'll get missed writes beyond that.

    However, I'd also argue that 22 kHz samples would indeed be doable at the time . . . compressed samples especially (1 or 2-bit CVSD/ADPCM derivatives would mean 22 kHz samples use the same bitrates as 5.5/2.75 kHz 8-bit PCM). Several existing MD sound engines use 4-bit ADPCM or DPCM derivatives that would still mean cutting bitrate in half.
    And, even for uncompressed 8-bit PCM, you could still fit a decent amount in the 2 to 4 MB (16-32Mbit) carts common in '94/95. Even Tiido's full percussion sample set of uncompressed 22 kHz samples takes 512 kB, and putting a stricter limit on that you could still fit a decent amount of 22 kHz instrument samples onboard. (significant amounts of speech would be another matter, and also something much better used with compression -ADPCM/DPCM/CVSD algorithms are relatively well suited to speech compression) There's plenty of examples of demos on this too, including Chilly Willy's CVSD/T-ADPCM player and Stef's 4-bit ADPCM (actually DPCM). http://www.sega-16.com/forum/showthr...612#post398612

    Using the SVP would allow for more flexibility and resource for compression too (along with mixing, scaling, and potential added effects -echo, interpolation, etc), and in this contex we're back to using the Z80+DAC as a simple 1-channel 8-bit PCM player streaming the data mixed into RAM by the SVP. 26 kHz 8-bit is the best you get, and I'm sure you've seen Joe's video demoing that: http://www.sega-16.com/forum/showthr...on-the-Genesis

    http://www.sega-16.com/forum/showthr...l=1#post398502

    Not only retail/consumer cost, it's the total package. SVP doesn't address the color issue or makes it worse, isn't powerful enough to even approach 3DO/Jag level 3D, and cannot likely address the audio problem any better than Genesis carts could without adding more hardware to the cart/expansion. Then there is the games. Apparently the 32X was powerful enough to have stuff like Space Harrier and After Burner, not to mention Doom, Star Wars Arcade and Virtua Racing Deluxe, hastily slapped together for the break neck launch. Do we think that all of these games would have turned out well with a similar amount of resources spent (ignoring the 15-color problem)? I think the SVP is far less versatile in what kinds of games could be enhanced by it.
    Most of the games done on the 32x could probably have been done reasonably with SVP or (at very least) SVP-CD. In any case, it would beat Super FX level 3D, and offer some other potential advantages as a general-use enhancement chip (sprite/tile compression, sound, etc)

    Simplicity and cost are the main advantages over everything else . . . that and (like the Jupiter) it already existed at that point. The SVP and Saturn chipset were there, and the SVP was ready for mass production already. 32x was an all-new design. Being weaker on top of all that would also be an advantage in not directly competing with the Saturn as such.
    32x would mean more R&D, more tooling for manufacturing, another major platform for programmers to learn, and a major financial investment on top of that for something that couldn't possibly ever become a long-term competitor for the next generation. (and was also horribly cost ineffective -MD+32x+CD)

    Even in the context of thinking from their perspective at the time, I really don't see how a "neither here nor there" design like the 32x could be seen as preferable to such alternatives . . . (too expensive to be an easy to a "simple" add-on most users could just pick up, and too weak to really be an all-new platform in the long run, especially when the Saturn was already planned for that role)

    And, as to the color problem, look at Sega CD games. Most of the ASIC-rendered games used very large areas with only 15 colors (Batman and Robbin does the road, cityscape, and all buildings/cars/objects with 15 colors). Star Fox and Vortex both used 15 colors, as did the many Atari ST and Amiga polygon games. Silpheed is a good example too. Star Wars Arcade already uses a color set well suited to a limited palette (lots of gray). Shadow Squadron probably would have been best off shifting to a more subdued color scheme. (space games normally went with gray for ships -metallic and such)

    Plus, you've got the potential to mix textures and flat shaded polygons, especially with the CD ASIC in use. Due to RAM space limits (and eventually ASIC bandwidth too), using decal style textures sparingly might be more efficient. (similar to Super FX games, but used more frequently)

    A game like X-Wing should have been possible to do decently well on the MCD, at least with SVP. (model detail would probably take a big hit with MCD alone . . . the PC version could run playably on a 386SX-16 at min detail, but that's got really basic models that don't look like much) Getting it to run in the limited RAM wouldn't be that crazy either, given the 1993 PC game already worked in just 1 MB on the PC. (plus, some of the non-RAM limited stuff from the CD version could be added -like the enhanced cutscenes and the voiceovers, albeit streamed from CD rather than HDD)
    Controls might be a tough fit, but the 6 button pad would probably be OK. (maybe offload a few things to menus rather than hot keys)

    In 1994 I didn't see the 32X as a stop gap, I saw it as an inexpensive way to get the Genesis and Sega CD into the range of 3DO and Jaguar 3D while enhancing everything else about the Genesis that the Super Nintendo could possibly win at. SVP basically would have given an answer to SNES SFX, which I already thought the Sega CD beat handily. The total value proposition of SVP versus 32X favors the 32X ESPECIALLY if Sega hadn't canceled it six months later but even with its existing library.
    What Sato/Nakayma are depicted as addressing with SoA staff in January 1994 sounds more like an interim platform to be superseded by the Saturn when it became available.


    Now if come mid 1995, with no real world 32X, Sega offered me a Jupiter for $250 or a Saturn for $400, and then dropped the prices to $200 and $300 that Fall we are just talking about a new system launch and nothing else. That doesn't do anything for 1994, the Genesis, or the Sega CD. More on that later in relation to your question about my not buying a Super Nintendo.
    Why not release the Jupiter in '94 in line with the 32x? . . . OK, I have a bunch of possible "why nots" for that, but more to the point: if Sega wanted "something new" hardware wise in that context, why not press on with Jupiter for '94.
    As it was, the 32x absorbed manufacturing resources that could have gone to Saturn, so why not do the same with Jupiter. (which would, again, be much more helpful to the Saturn on the market)


    Yeah, I don't think focusing exclusively on the new hardware was the smartest move in 1994-95. The Genesis was going to remain Sega's most mass market product for at least a couple of years while the next generation limped into shape. In my ideal scenario the Genesis and Sega CD would have been supported through 1998 and the 32X makes that more likely to have been interesting, not less.
    IMO, no 32x could have been more interesting: if the MCD managed to eek out enough mass-market interest to keep it as a lower-budget platform, we might have seen some actual 3D stuff being pushed on it . . . maybe some more PC ports (that should have been there earlier) or other home computer platforms. (Geograf Seal would have been awesome, for example )
    That, or SVP with MCD too, since that would have been cheap enough for a potential high adoption rate in the low-end of the market, more so if it was embedded in the MD (and/or a duo console . . . an affordable one).

    I was and still am predominantly interested in Action-Arcade style games, not RPGs and not console exclusive franchises. The 32X was perfect for my tastes. Space Harrier and After Burner seeing their first "Arcade perfect" releases on the 32X plus Virtua Racing Deluxe alone took the SNES library out of my consideration. Say what you will about Cosmic Carnage but it gave me all kinds of hope for Mortal Kombat II and Super Street Fighter II, not to mention NEO GEO ports. By late 1994 the only SNES exclusive games I really was tempted by was Super Double Dragon, Star Fox and Super Mario Kart. Nintendo's other mainstay franchises were and are of no interest to me. Even today, when Super Nintendo games are relatively cheaper than they were at the time, I only own about ten more of them than 32X games and I use the 32X a lot more.
    I can understand that, but that also wasn't the big selling point (at least outside Japan) at the time. Arcade ports were not usually a selling point anymore . . . some games that happened to be arcade ports might be selling points for other reasons (inherently good games and/or licensed IPs like Star Wars)

    Hell, the relatively small/niche arcade status is why a ton of people from the 90s (experienced with games at the time) remember most contemporary arcade games as the console ports, if they remember them at all. (and may not have even known they were ever arcade titles) As a mid/late 90s kid myself, I knew of quite a few games I never realized were arcade ports until many years later.

    TBH, I really don't see the sense in putting out very old games like Space Harrier and After Burner as flagship console releases at the time. As lower priority arcade classic releases, sure (as on the Saturn and later), but not high priority showcases in general.
    On the MCD, those games would already have been old to show off the system . . . tech demos, but not mass-market compelling games. They did have some newer, more relevant scaler games that could have been pushed on the CD though. (in the end, it's kind of moot since SoJ didn't bother with either ports of old -let alone newer- arcade ports showcasing the MCD's scaling hardware)



    In these discussions, I try to stick to the perspective of the general/collective/average consumer perspective, ie stuff that's popular to large segments of the viable consumer base. If that happens to overlap with my personal likes, then fine, but I'm not going to force a fantasy around what I would have liked specifically. -Sure, I can comment on a few specific examples that might fall in my personal catgories, but I tend to either qualify those as not worthwhile (too niche) or as an actual viable mass market product. (I think X-Wing would fall into the latter due to being a well designed game and a massively popular IP . . . even if not typical "console gamer" fare as a simulator . . . though the MD did have plenty of sim combat games already, including some actual flight sims rather than the scifi/fantasy stuff)
    6 days older than SEGA Genesis
    -------------
    Quote Originally Posted by evilevoix View Post
    Dude it’s the bios that marries the 16 bit and the 8 bit that makes it 24 bit. If SNK released their double speed bios revision SNK would have had the world’s first 48 bit machine, IDK how you keep ignoring this.
    Quote Originally Posted by evilevoix View Post
    the PCE, that system has no extra silicone for music, how many resources are used to make music and it has less sprites than the MD on screen at once but a larger sprite area?

  10. #40
    I remain nonsequitur Shining Hero sheath's Avatar
    Join Date
    Jul 2010
    Location
    Texas
    Age
    42
    Posts
    13,313
    Rep Power
    130

    Default

    Quote Originally Posted by kool kitty89 View Post
    SNES could do better, for the most part (compared to existing 32x stuff), with the exception of having the forced heavy filtering on everything. (technically, the SPC unit could do better than anything the 32x historically put out . . . though an SH2 mixing to DMA sound could certainly do a lot of things the SPC unit can't . . . or at least normally can't)
    The is the major limit of the 64kB sample RAM and the nasty bottleneck for streaming added samples, so that's a realistic consideration too, and the fixed sample format (no option to use 16 or 8-bit PCM vs 4-bit vs 2-bit ADPCM, CVSD, or some custom)
    Two games I can directly compare between the 32X and SNES, Blackthorne and MKII, favor the 32X for audio and graphics. That is primarily where I was coming from. Even if the SNES samples were higher bitrate they sound flatter and more muffled.

    Quote Originally Posted by kool kitty89 View Post
    MKIII wasn't done by probe, and uses the GEMS sound engine instead (hence the weaker arrangements and strict use of arcade compositions -rather than Matt Furniss's meddling with things and swapping in his own awesome original stuff ). GEMS only supports 1 PCM channel, hence the "cutting out" problem.
    I'm pretty sure MKIII mixes more than one sample without cutoffs and with much higher quality than typical Genesis DAC. The sample clarity isn't quite on MKII 32X's level, but I'd put it up against the SNES games. Also there are far more samples played than MK1-2 on Genesis.

    Quote Originally Posted by kool kitty89 View Post
    Interestingly enough, EA sound engines tended to handle samples fairly well, though there's notable distortion in some things too (again, mostly in-game and with lower sample rate stuff). Madden '96 always sticks with the streaming intro music, speech, and great sounding sampled percussion in the synth (and pretty nice compositions from Tommy Tallarico). Actual use of the sound engines (music wise) was kind of inconsistent (often similar to GEMS stuff bland/grating) which I assume is partially due to the typical sound design techniques used in American studios (PC games and arcades had some of the same problems using FM synth with many US devs). That said, several of EA's sound drivers were programmed by some pretty notable sound programmers. In the case of Madden '96, it was Rob Hubbard, well known for his home computer music in Europe. And, to be fair, EA sound engines on average tend to at least have better sounding music than GEMS on average.
    Some EA games might have handled samples well, but they all tend to have horrible FM instrumentation that represents the worst "twang" Genesis sound I can think of.

    Quote Originally Posted by kool kitty89 View Post
    Several of those EA engines were also capable of relatively decent sample rates and multiple sample channels. (Skitchin' supports 4 channels iirc and at least one scaled for pitch as a MOD player -albeit only using a pretty limited guitar sample that gets strained when pushed to far out of its native sample rate . . . it sounds like it's at least ~11 kHz, I'm not super good at guessing that in general though)
    https://www.youtube.com/watch?v=e9AYz6CIq2E (madden '96 intro plus menu music)
    I ended up picking up Skitchin' because of our previous discussion on it. It might be mixing a lot of channels but it is not pleasant to listen to.

    Quote Originally Posted by kool kitty89 View Post
    http://gdri.smspower.org/wiki/index....nd_Engine_List
    Also the source for pretty much all other mentions of sound engines. (though some were also pointed out by other members on here)
    I saw that earlier, what I don't see is a description of how 32X Standard J or U works. Looking straight at this list I can't tell that these engines aren't using 32X sound.

    Quote Originally Posted by kool kitty89 View Post
    I'm talking about 22 kHz (or 26 kHz) for mixing purposes mostly, which is a decent output rate for doing software mixed/scaled samples (MOD players and such). 26 is the hard limit for the DAC (not the Z80), you'll get missed writes beyond that.

    However, I'd also argue that 22 kHz samples would indeed be doable at the time . . . compressed samples especially (1 or 2-bit CVSD/ADPCM derivatives would mean 22 kHz samples use the same bitrates as 5.5/2.75 kHz 8-bit PCM). Several existing MD sound engines use 4-bit ADPCM or DPCM derivatives that would still mean cutting bitrate in half.
    And, even for uncompressed 8-bit PCM, you could still fit a decent amount in the 2 to 4 MB (16-32Mbit) carts common in '94/95. Even Tiido's full percussion sample set of uncompressed 22 kHz samples takes 512 kB, and putting a stricter limit on that you could still fit a decent amount of 22 kHz instrument samples onboard. (significant amounts of speech would be another matter, and also something much better used with compression -ADPCM/DPCM/CVSD algorithms are relatively well suited to speech compression) There's plenty of examples of demos on this too, including Chilly Willy's CVSD/T-ADPCM player and Stef's 4-bit ADPCM (actually DPCM). http://www.sega-16.com/forum/showthr...612#post398612

    Using the SVP would allow for more flexibility and resource for compression too (along with mixing, scaling, and potential added effects -echo, interpolation, etc), and in this contex we're back to using the Z80+DAC as a simple 1-channel 8-bit PCM player streaming the data mixed into RAM by the SVP. 26 kHz 8-bit is the best you get, and I'm sure you've seen Joe's video demoing that: http://www.sega-16.com/forum/showthr...on-the-Genesis

    http://www.sega-16.com/forum/showthr...l=1#post398502
    If it is true that 22 Khz samples could be compressed and used the same as ~2-5Khz samples then I wonder why they didn't do that back then, was the compression method not common, or was there too much overhead? All of the examples of streaming music I have seen have huge ROM sizes, though you are right that in a 32-40Mbit game the developer might have chosen to dedicate 512-1028KB of that to samples. The Sega chime in Sonic 1 is (supposedly) 64KB of the 512KB ROM after all.

    Quote Originally Posted by kool kitty89 View Post
    Most of the games done on the 32x could probably have been done reasonably with SVP or (at very least) SVP-CD. In any case, it would beat Super FX level 3D, and offer some other potential advantages as a general-use enhancement chip (sprite/tile compression, sound, etc)

    Simplicity and cost are the main advantages over everything else . . . that and (like the Jupiter) it already existed at that point. The SVP and Saturn chipset were there, and the SVP was ready for mass production already. 32x was an all-new design. Being weaker on top of all that would also be an advantage in not directly competing with the Saturn as such.
    32x would mean more R&D, more tooling for manufacturing, another major platform for programmers to learn, and a major financial investment on top of that for something that couldn't possibly ever become a long-term competitor for the next generation. (and was also horribly cost ineffective -MD+32x+CD)
    I don't see the SVP being fast enough for even Cosmic Carnage, though that one's colors might not have suffered too much. Mainly what I was getting at is that the 32X was fast enough for developers to hastily slap together its games and have them run fairly well. SVP versions of reasonable quality would have taken much more careful development. And forget about the possibility of scaler type games looking like a source port, add in color optimization and redrawn art to that experience at a minimum.

    Quote Originally Posted by kool kitty89 View Post
    Even in the context of thinking from their perspective at the time, I really don't see how a "neither here nor there" design like the 32x could be seen as preferable to such alternatives . . . (too expensive to be an easy to a "simple" add-on most users could just pick up, and too weak to really be an all-new platform in the long run, especially when the Saturn was already planned for that role)
    Again, when these publishers could and did dedicate only one single programmer to the process of porting between PC, PS1 or Saturn and got the game out in time and usually in good form, why does it matter that the 32X doesn't have the Saturn VDPs from a programming perspective? Most of the complaints I see about Saturn development were in regard to optimizing SH2 Assembly, the 32X would have the same issue. The rest of the complaints I see about Saturn programming are in regard to managing the busses or dealing with Quads, or trying to get the system to operate at a Model 3 level (freaking insane). Besides, we already know from the Bayless interviews that the original 32X design was beefier and probably closer in cost to your Jupiter design. Sega thought it was too expensive to market and stripped it down to "Peripheral cost". In the real world if Sega Japan or America started with a Jupiter instead of "Two SH2s and dual frame buffers" it would have lost most of the system before it made it to market anyway. More on that later though.

    Quote Originally Posted by kool kitty89 View Post
    And, as to the color problem, look at Sega CD games. Most of the ASIC-rendered games used very large areas with only 15 colors (Batman and Robbin does the road, cityscape, and all buildings/cars/objects with 15 colors). Star Fox and Vortex both used 15 colors, as did the many Atari ST and Amiga polygon games. Silpheed is a good example too. Star Wars Arcade already uses a color set well suited to a limited palette (lots of gray). Shadow Squadron probably would have been best off shifting to a more subdued color scheme. (space games normally went with gray for ships -metallic and such)
    Star Fox's backgrounds on the planet levels would have to be dithered, Shadow Squadron's light sourcing/shading would too if it would even be possible. If the colors for these games could be optimized to Silpheed's level I think it would have worked fine.

    So this:


    Would end up more like this (please no lectures on tile formats and not using JPGs, I know):



    Quote Originally Posted by kool kitty89 View Post
    Plus, you've got the potential to mix textures and flat shaded polygons, especially with the CD ASIC in use. Due to RAM space limits (and eventually ASIC bandwidth too), using decal style textures sparingly might be more efficient. (similar to Super FX games, but used more frequently)

    A game like X-Wing should have been possible to do decently well on the MCD, at least with SVP. (model detail would probably take a big hit with MCD alone . . . the PC version could run playably on a 386SX-16 at min detail, but that's got really basic models that don't look like much) Getting it to run in the limited RAM wouldn't be that crazy either, given the 1993 PC game already worked in just 1 MB on the PC. (plus, some of the non-RAM limited stuff from the CD version could be added -like the enhanced cutscenes and the voiceovers, albeit streamed from CD rather than HDD)
    Controls might be a tough fit, but the 6 button pad would probably be OK. (maybe offload a few things to menus rather than hot keys)
    I agree, a game like X-Wing should have been adaptable to the Sega CD especially if SVP was involved.

    Quote Originally Posted by kool kitty89 View Post
    What Sato/Nakayma are depicted as addressing with SoA staff in January 1994 sounds more like an interim platform to be superseded by the Saturn when it became available.
    Yes, they clearly wanted the Saturn to be the flagship product and so they made the 32X not only a stop-gap but a late term abortion. What Sega of America suggested is what I bought in 1994 and expected to see supported at least into 1996 if not beyond. It is the simple difference between a "stop gap" and a lower end budget system. My interest in the 32X (or any like kind of late Genesis add-on) is certainly not mass market, but I think the 32X actually was of mass market interest. Certainly enough to boost Sega's revenues in certain regions while the Saturn and PS1 slowly took over market share from the 16-bit consoles.

    For myself, heck, I just turned a Pentium III 500Mhz machine into a 90s game box just to see the Matrox Mystique at its best. Back in the day I couldn't get half the games I'm playing now to run just because I only had a Cyrix 686 133Mhz coupled with that Mystique, now everything runs great and who cares about texture filtering?

    Quote Originally Posted by kool kitty89 View Post
    Why not release the Jupiter in '94 in line with the 32x? . . . OK, I have a bunch of possible "why nots" for that, but more to the point: if Sega wanted "something new" hardware wise in that context, why not press on with Jupiter for '94.
    As it was, the 32x absorbed manufacturing resources that could have gone to Saturn, so why not do the same with Jupiter. (which would, again, be much more helpful to the Saturn on the market)
    Why not release the full blown Saturn in the West in 1994 then? Cost factors for one, imagine the internet rumors today if the Saturn launched at $500. Also, they apparently couldn't make enough of the things for the Japanese market alone and didn't have enough SH2s to replace a faulty one in an early 32X development kit. Summer of 1995 Sega apparently couldn't supply all retailers with Saturn systems and so chose the limited "pre-launch" approach instead. No, the lack of a CD-ROM subsystem wouldn't have changed these facts for Jupiter. At best it would have launched in Japan in 1994 and shared sales with Saturn there. Whether it be the 32X's memory or VDP or something else, for whatever reason the Saturn was not going to be viable in the West as early as 1994 and the 32X was. I'm not sure if it was the faster SH2s or the type of memory or the VDPs, or DSP or SH1 that held back the Saturn's manufacturing early on but there is no reason to think that it wouldn't have affected Jupiter as well.

    Quote Originally Posted by kool kitty89 View Post
    I can understand that, but that also wasn't the big selling point (at least outside Japan) at the time. Arcade ports were not usually a selling point anymore . . . some games that happened to be arcade ports might be selling points for other reasons (inherently good games and/or licensed IPs like Star Wars)

    Hell, the relatively small/niche arcade status is why a ton of people from the 90s (experienced with games at the time) remember most contemporary arcade games as the console ports, if they remember them at all. (and may not have even known they were ever arcade titles) As a mid/late 90s kid myself, I knew of quite a few games I never realized were arcade ports until many years later.

    TBH, I really don't see the sense in putting out very old games like Space Harrier and After Burner as flagship console releases at the time. As lower priority arcade classic releases, sure (as on the Saturn and later), but not high priority showcases in general.
    On the MCD, those games would already have been old to show off the system . . . tech demos, but not mass-market compelling games. They did have some newer, more relevant scaler games that could have been pushed on the CD though. (in the end, it's kind of moot since SoJ didn't bother with either ports of old -let alone newer- arcade ports showcasing the MCD's scaling hardware)

    In these discussions, I try to stick to the perspective of the general/collective/average consumer perspective, ie stuff that's popular to large segments of the viable consumer base. If that happens to overlap with my personal likes, then fine, but I'm not going to force a fantasy around what I would have liked specifically. -Sure, I can comment on a few specific examples that might fall in my personal catgories, but I tend to either qualify those as not worthwhile (too niche) or as an actual viable mass market product. (I think X-Wing would fall into the latter due to being a well designed game and a massively popular IP . . . even if not typical "console gamer" fare as a simulator . . . though the MD did have plenty of sim combat games already, including some actual flight sims rather than the scifi/fantasy stuff)
    Right, so that leaves MKII, Knuckles Chaotix, Star Wars Arcade, Doom and maybe Virtua Racing Deluxe as relevant mass market products. Space Harrier and Afterburner, and T-Mek were apparently just easy enough to port/adapt that they were worth it to do so. However Carmack got Doom 32X running so smooth in a mattter of weeks is pretty amazing too. I don't think the SVP would have been as quick/easy for adapting these games. Doom SNES SFX-2 coming out in 1995 pretty much points to that, but at least it got more of the original levels in.
    Last edited by sheath; 01-31-2013 at 11:50 AM.
    "... If Sony reduced the price of the Playstation, Sega would have to follow suit in order to stay competitive, but Saturn's high manufacturing cost would then translate into huge losses for the company." p170 Revolutionaries at Sony.

    "We ... put Sega out of the hardware business ..." Peter Dille senior vice president of marketing at Sony Computer Entertainment

  11. #41
    Hero of Algol kool kitty89's Avatar
    Join Date
    Mar 2009
    Location
    San Jose, CA
    Age
    30
    Posts
    9,725
    Rep Power
    64

    Default

    Quote Originally Posted by sheath View Post
    Two games I can directly compare between the 32X and SNES, Blackthorne and MKII, favor the 32X for audio and graphics. That is primarily where I was coming from. Even if the SNES samples were higher bitrate they sound flatter and more muffled.
    I disagree on MK II . . . the higher color sprites are noticeable (directly comparing to the Genesis version), but IMO the difference is so minor overall that it's pretty inconsequential. The BGs look better on the SNES (at least color wise) in a far more noticeable way. Both Sega versions have a resolution advantage over the SNES.

    Blackthorne is a separate issue entirely. It's the best looking version of that game, and a remake of sorts rather than a port of the older versions. The 256 color DOS game really should have been able to look close to that . . . maybe even trade in the dither mesh for look-up table based color blending and shading (like Doom, among others)
    An SNES port of that probably could have looked much closer to that too, had it been based on the same version of the game . . . the aspect ratio difference would be noticeable (and more of a problem in Europe -typical for SNES, like SMS, NES, and low res MD games), but the color/animation and overall graphics should have been able to be much closer to the 32x game than what's on the SNES right now. (still weaker on the SNES, granted, but not a hugely obvious difference to the point of it immediately jumping out to the average gamer without seeing the 2 side by side)

    TBH, the SNES/DOS version of Blackthorne probably could have been done on the MD alone without looking too bad, but the 32x game would have been a much bigger jump to be sure. (all that smooth shading, added color, etc)
    After all, Toy Story didn't look that much worse color/shading wise on the Genesis. It's noticeable, but still quite reasonable on the MD version. (and with tasteful use of dithering)


    I'm pretty sure MKIII mixes more than one sample without cutoffs and with much higher quality than typical Genesis DAC. The sample clarity isn't quite on MKII 32X's level, but I'd put it up against the SNES games. Also there are far more samples played than MK1-2 on Genesis.
    Hmm, it's really hard to notice from the SFX alone, especially since the PCM drums do indeed seem to cut out, but listening to the "finish him" speech, there does seem to be SFX playing at the same time. (the low quality of the samples makes it hard to notice at first) Interesting, I hadn't realized GEMS could do that.

    And as to being "good" compared to "average" MD stuff . . . that's highly debatable. It's pretty average for speech in GEMS games (which pretty much never have distortion -GEMS polls the YM timers for accuracy . . . which has the massive disadavantage of using tons of Z80 time -worse than timer interrupts would be if it could use them; that's also why GEMS can only handle up to ~10 kHz).
    It's better than BAD examples of speech on the MD . . . ie either games using really crappy playback systems, or just ones that end up poorly suited to the specific samples in question. (like ones capable of high sample rates, but still have some distortion from missed writes -timing issues- that become obvious at low sample rates -again, not sure why scaling up samples to play at higher rates wasn't used more often -Mortal Kombat 1 is a good example of this)

    I'm going to make an educated guess here and say that Williams published MK games on the Genesis along with a few other GEMS titles (maybe Earthworm Jim) are opting to use 4-bit PCM to save space rather than using more ROM for sound or using 8-bit PCM at half the sample rate (since GEMS only does 4 and 8-bit linear PCM, no DPCM or ADPCM, etc). I say that due to the type of aliasing audible in that sound, which is not really like 8-bit PCM at very low bitrates (usually muffled along with weird harmonics), but does sound a lot like examples of 4-bit PCM I've heard. (which is more like radio static or white noise)
    Actually, the aliasing is kind of similar to what you get with playback through the PSG too . . . only After Burner II does that on the MD (and the low sample rates make it muffled too). Actually, ABII's speech still manages to sound better at times than Capcom did with SFII.
    Hell, if you know a game is only going to be using 4-bit PCM anyway (and better quality isn't practical for whatever reason), then using PSG playback might have been an interesting alternative. (unless having those 3 PSG channels was more important than a 6th FM channel . . . you don't have to use all 3, but that tends to sound the best and it's how ABII does it)

    I'll also modify my previous comment on music quality in MK3 (and UMK3). It's actually pretty decent sounding music/instrument wise, at very least for GEMS based stuff. (not as good at Furniss's Krysalis stuff though . . . unless you're hung up over the arcade compositions )

    Some EA games might have handled samples well, but they all tend to have horrible FM instrumentation that represents the worst "twang" Genesis sound I can think of.
    I think it's more the arrangements used than the instruments themselves . . . "twangy" can sound fine if used properly. (that said, the common instruments used seem rather limited . . . like with GEMS stuff -I assume a MIDI sequencing related practice from composers not experienced enough to customize things . . . same problem with many PC sound engines of the time -that, and the actual sequencer software often being limited or awkward to customize)

    Madden '96 is probably one of the best examples of music done on the EA driver. (or drivers, rather, since there's more than one used) That game has some good examples of PCM too. (in music and speech -though the play by play speech is lower quality)



    I ended up picking up Skitchin' because of our previous discussion on it. It might be mixing a lot of channels but it is not pleasant to listen to.
    I like a few of the tracks, and on average it's better than the Road Rash games at least.


    I saw that earlier, what I don't see is a description of how 32X Standard J or U works. Looking straight at this list I can't tell that these engines aren't using 32X sound.
    I used an emulator (Gens GS) to manually shut off sound channels and halt CPUs. The sound driver is directly driven by the 68k and Z80 and uses only the PSG and FM/DAC in-game with Virtua Racing. Turing off the DAC gets cuts out all samples, as does halting the Z80. (music keeps playing in both cases)
    Resetting the SH2s will halt/crash the game itself, but music continues playing, while resetting the 68000 causes music/sound to crash.
    So, overall implying a fairly typical MD 68k driver set-up with the 68k actually controlling all the music/sound playback, but the Z80 acting as a PCM (or compressed sample) playback engine. (preSMPS 68k, SMPS 68k, Capcom's drivers, Data East -Atomic Runner and Vapor Trail, Krysalis, Mega Turrican, among others)

    If it is true that 22 Khz samples could be compressed and used the same as ~2-5Khz samples then I wonder why they didn't do that back then, was the compression method not common, or was there too much overhead? All of the examples of streaming music I have seen have huge ROM sizes, though you are right that in a 32-40Mbit game the developer might have chosen to dedicate 512-1028KB of that to samples. The Sega chime in Sonic 1 is (supposedly) 64KB of the 512KB ROM after all.
    CVSD was nothing new, but it also wasn't common to implement in software . . . but technically, it's not really more complex than doing a software ADPCM decoder. 2-bit ADPCM derivatives weren't that odd either, and Covox promoted that (along with 4-bit and 3-bit formats) for use with their 8-bit DACs for PCs. (they also provided sample code for relatively computationally simple methods of decoding intended for being practical on low end 80s x86 PCs -similar techniques would be applicable to the Z80 as far as decoding goes; software playback through the DAC would still be a separate hurdle)
    4-bit ADPCM formats were more commonly used in general (though what the SNES uses is not normal 4-bit ADPCM and has a slightly lower compression ratio too, closer to 4.5-bits per sample), but it certainly wouldn't be unthinkable that programmers couldn't come up with their own custom hybrid formats too. (actually programming an encoder and decoder is another matter, of course)

    DPCM should also be possible, but anything below 4-bit DPCM probably won't be worth it, at least for speech and complex sampled sounds (1-bit DPCM is what the NES decodes, and you it sounds pretty bad for some things, even with the sample rate really high it's not great . . . good for some things for that bitrate though -33 kHz is 33 kbps or 1/8th that of 8-bit PCM). Still, 1-bit DPCM isn't good for much in this situation (especially given sounds FM can do), but a specialized 2-bit DPCM derivative might be useful.
    4-bit DPCM is definitely useful and Stef uses his own that as well as Sega with SMPS.

    The difference between DPCM and ADPCM (or CVSD) is that each word of a sample (every 4 bit chunk, for 4-bit DPCM) defines a specific delta value in a table in RAM (a single set of 16 deltas -16 values in 4 bits), while ADPCM uses a much broader set of delta values (either a large set of tables or an algorithm that modifies a single base set/table of deltas) which requires the decoder to do additional computation to interpret/predict how the waveform should be changing (and what delta sets to change to for the next sample word). DPCM is a lot simpler to do as such, hence part of why Stef managed a 2 channel 22 kHz player and 4 channel 16 kHz vs a single 22 kHz channel for Chilly Willy's CVSD/TADPCM decoder.
    And if you don't know how delta encoding works, it's basically adding/subtracting a value to the last sample value played, rather than PCM which just stores the sample values directly.

    Additionally, higher sample rates aren't always better . . . compression reduces quality, and depending on the sound type in question, different formats would be favored for a given bitrate. (and some things will sound really bad in any format at a given bitrate) In certain situations, plain 4-bit linear PCM can even be better suited than 4-bit DPCM or ADPCM, and DPCM might be better suited than ADPCM . . . logarithmic (stepped/nonlinear) 4-bit PCM may also be better suited in some cases. (all those would be the same sample rates for a given bitrate too, since they're all 4-bits per sample) You've got cases where 4-bit PCM could sound better than 8-bit PCM at 1/2 the rate (a lot of 4 kHz 8-bit PCM on the MD probably applies to this), or similar for 1-bit vs 2-bit vs 4-bit CVSD/ADPCM . . . and 8-bit and 16-bit, and 8-bit nonlinear (like uLAW and ALAW) vs linear. Nonlinear 8-bit PCM would be pointless on the MD though, since you've got to decompress to 8-bit linear anyway. (if you're going to use any 8-bit per sample format, linear PCM is the only one that makes sense)

    I don't see the SVP being fast enough for even Cosmic Carnage, though that one's colors might not have suffered too much. Mainly what I was getting at is that the 32X was fast enough for developers to hastily slap together its games and have them run fairly well. SVP versions of reasonable quality would have taken much more careful development. And forget about the possibility of scaler type games looking like a source port, add in color optimization and redrawn art to that experience at a minimum.
    It depends, but like with the MCD, VDP DMA bandwidth will also be a bottlneck still. (letterboxing will always help though, since it gives more vblank time . . . plus it automatically means a lower screen res too -PAL is also better off than NTSC)


    I'm short on time right now, so I'll try to address the rest of your post later.
    6 days older than SEGA Genesis
    -------------
    Quote Originally Posted by evilevoix View Post
    Dude it’s the bios that marries the 16 bit and the 8 bit that makes it 24 bit. If SNK released their double speed bios revision SNK would have had the world’s first 48 bit machine, IDK how you keep ignoring this.
    Quote Originally Posted by evilevoix View Post
    the PCE, that system has no extra silicone for music, how many resources are used to make music and it has less sprites than the MD on screen at once but a larger sprite area?

  12. #42
    Hero of Algol kool kitty89's Avatar
    Join Date
    Mar 2009
    Location
    San Jose, CA
    Age
    30
    Posts
    9,725
    Rep Power
    64

    Default

    Quote Originally Posted by sheath View Post
    Again, when these publishers could and did dedicate only one single programmer to the process of porting between PC, PS1 or Saturn and got the game out in time and usually in good form, why does it matter that the 32X doesn't have the Saturn VDPs from a programming perspective? Most of the complaints I see about Saturn development were in regard to optimizing SH2 Assembly, the 32X would have the same issue.
    This a a huge oversimplification.

    For one thing, developing for ANY other platform is always going to consume more time and resources. The more different that platform is in raw capabilities and programming/rendering techniques requires as well as raw programming complexity, the greater the burden is to develop for that platform. Equally, any resources going to work on said platform can and will detract from that of other platforms.

    For now, let's stick to 32x vs Saturn vs Jupiter programming specifically. With 32x, you've got the MD hardware to work with and the SH2s doing software rendering and (maybe) software mixed sound . . . at the time (even in 1995), documentation issues screwed up the sound capabilities. (this is a significant point to include, as documentation is a big issue for the Saturn -and Sega CD too)
    Games on the 32x may use the MD graphics for a lot, a little, or none at all (usually the far backdrop at least), and any graphics done on the 32x end must be software rendered (aside from simple screen/linescroll -single plane- and line fill -for clearing framebuffer and solid filling polygons). One or both of the SH2s will basically be taking the place of a GPU as well as handling the 3D math and game logic/AI/etc. (some games might make more use of the 68k) You've got relatively limited work RAM that barely made it very difficult to design some types of games for, ROM could make up for some of this (often wasteful though -more uncompressed data), but it's still a big problem for early 5th gen multiplat games. (hell, PC games already had to be optimized/reworked to work in the RAM available on the 3DO/Jaguar/PSX/Saturn -most PC games of that time needed 4 MB minimum) The Sega CD potentially helps in some areas, but it makes some areas worse (unless you use cart+CD) and in all cases the complexity of programming for all 3 (to efficiently work around the bottlenecks) would increase the development resource draw significantly more. (again, in the context of actually trying to push mainstream mid-90s 3D games -or higher-end 2D ones)

    For a reasonably good looking multiplat game with the 32x as a target, you've got to optimize the art for both the memory(ROM and RAM)/bandwidth limitations and the color limitations (depending on color depth on the framebuffer and how MD graphics are used), plus the sound, and overall processing/rendering. It wouldn't be too tough to design a game specifically optimized around the 32x and then port that over directly (with minimal changes) to more powerful systems, but then you've got a game's potential being heavily limited by the weakest platform. (lowest common denominator effect)

    Programming experience for the 32x would not be directly applicable to Saturn programming. Sure, learing SH2 assembly would be useful (but it's a well designed, well-supported, bug-free CPU with a straightforward ISA that any assembly language programmer should learn quickly -even the Jaguar RISC language could be learned quickly). More important would be experience with multiprocessing with the SH2s in that configuration. However, that experience has a very limited overlap with how the SH2s are most useful in the Saturn. 32x programmers would be emphasizing software rendering techniques and totally different uses for multiprocessing (like 1 dedicated to graphics, one to 3D math+game logic), or even optimizing for games making significant use of the 68000 (offloading some game logic overhead -or all of it in simpler -but graphically demanding- games). 2D games might barely use the SH2s at all on the Saturn while being hugely intensive on them in the 32x with very different programming techniques being used in each.

    With the Jupiter, by comparison, programmers would learn to use the hardware in basically the same manner as the Saturn itself. They could learn to optimize around the VDP capabilities, sound capabilities, and use the CPUs in basically the same capacities, thus focusing on the strengths, weaknesses and general quirks of the system. Even with documentation and tools just as sketchy as they were early on for the Saturn, developers would at least have a head start at working around those problems.
    More likely IMO, is that pushing the Jupiter as such in 1994 would also give a general boost in resources focusing on compiling better documentation in general (and better translations of existing documentation) as well as building more comprehensive tools for use in the SDK "out of the box." The Saturn wasn't built around API programming like the 3DO, PSX, or 3D accelerator cards (or most later consoles), but it could still support graphics libraries to a reasonable extent. (it would just take more effort to code efficient libraries and more work to optimize games in general -using even "good" libraries would mean a bigger hit than low-level vs API programming on the PSX, but doing a mix of API stuff with low-level optimized tweaks to that could be a good compromise for multiplat releases or perhaps some lower budget Saturn specific games too)

    Additionally, since you'd basically be coding a Saturn game outright, it would be pretty straightforward to expand upon that for a CD-ROM version of the game and/or to release expanded content on CD. Some games heavily exploiting ROM-specific advantages would tend to only favor the latter.

    There's also the manufacturing/distribution/marketing logistics to consider too:
    The RAM content of the Jupiter would allow games to be heavily compressed and load data into RAM as needed, only using uncompressed (or lightly compressed) data when it was really favorable (ie the performance boost would be worth the space/cost). That would mean load times, but they should still have been faster than CD load times. (assuming efficient decoding algorithms were used) With the 32x, you have little space to decompress into, so it's mostly down to uncompressed data and continually streaming and decompressing data. (much more CPU overhead and/or weaker compression)
    So, like N64 games, you'd get a lot more bang for your buck at a given ROM size. (and, unlike the N64, the cost overhead for CD-ROM drives was still significant enough to be a major burden when the system was released -for 1996, that was just silly though)

    Further, even for games specifically made as cart-only for the Jupiter and not expanded/re-released with Saturn CD installments, you'd still be directly expanding the Saturn's software library since Jupiter cart games would be directly compatible. -As a very vague scenario, let's assume that Jupiter only got similar support in '94/95 to what the 32x already had (quantity/quality wise), but that all software at least reflected the system's general capabilities (roughly as well as 32x games did for that platform). So you'd have some pretty decent early-gen Saturn-quality looking games (some specifics due to ROM formats) with nice sample/synth sound, so genuine "next gen" class software. And then apply that to the Saturn's launch lineup as well . . . perhaps also with the addition of some otherwise cancelled (would-be) 32x projects and some degree of increased quality in the early (Western) Saturn software due to the head start in learning the architecture.
    Even if (for some inconceivable reason) Sega still went with the stupid, poorly planned, May 1995 "launch" of the Saturn in the US, at least it would have a much better established infrastructure to work with. (Jupiter games, Jupter CD -presumably released at the same time- allowing an affordable upgrade path for existing -and future- Jupiter owners; and the Jupiter itself being a complement to the Saturn's market position, rather than a segmented mess with the 32x --both fit together rather well, both play the same cart games, one can be upgraded to allow the rest of the CD games too, one is already at a mainstream price point while the other will stick to the higher-end niche for a while)


    TBH, the real arguments aren't whether Jupiter would have made more sense than 32x, but whether it made more sense than SVP or nothing at all until Saturn.
    I'm really not sure what I'd decide on this. With the Jupiter, there's a genuinely good argument (IMO) for releasing an intermediate (NOT interim) platform on the market to smooth transition to the next generation. Costs of CD-ROM drives was a big part of that, and the main reason the new hardware couldn't reach mass-market home console friendly prices right away at the time. (which had not been the case for the previous 2 generations, albeit most pre-crash consoles had been very expensive at launch -taking inflation into account-)
    -Perhaps another reasonable option from the early 1994 perspective would have been to implement the SVP as a bundled module with VR (and launch at the same date VR was planned to), and then push plans for Jupiter along as well. Use the short-term public response (and 3rd party interest) in the SVP that Spring to guide further plans for SVP support and the Jupiter. If SVP is well accepted all-around, they might consider dropping Jupiter before officially announcing/unveiling it (probably at the Summer CES), or decide to shift away from SVP in favor of emphasizing Jupiter. (status of the MD/CD would also contribute to that) Finally, if the Saturn chipset hit delays, then Jupiter would be off the table for the immediate time, so SVP would remain the de-faco choice in the interim anyway.

    Oh, and technically, the SVP would share many of the programming/development disadvantages as the 32x list includes above (actually worse in some respects -limited RAM, limited CPU resource, no use of SH2s, etc). However, as an overall product, there's still many advantages: namely cost, but also convenience of installation (compare a lock-on cart to the 32x). In terms of cost, I don't just mean to the consumer, but to Sega in general. The risks for PR and monetary loss from the SVP would be much smaller than those of the 32x since it would be far cheaper to manufacture, cheaper/easier to distribute, and (arguably) easier to market.
    -Though, thinking on the SH2 programming issue . . . there's also the interesting possibility of using a configuration very similar to SVP, but using a Super H CPU instead. (SH1 should have been cheaper and still uses the same instruction set, plus it's usable as a general purpose CPU -unlike the SVP's DSP- but also has performance trade-offs compared to the SVP for raw computational power -faster for some things, slower for others- . . . the SVP also isn't just a DSP, but has other custom logic onboard -it's an ASIC with DSP onboard, and I'm not sure what that includes -could have hardware drawing assist for line fill and such, I don't know any more on the SVP specifically)

    Star Fox's backgrounds on the planet levels would have to be dithered, Shadow Squadron's light sourcing/shading would too if it would even be possible.
    Well, either dithered or posterized . . . in either case minimized by optimization. It's just like any MD vs SNES 2D comparison really. Actually, it's better for the MD than usual, since you've got 3 palettes to dedicate to the BG without contention with sprites (and the optional use of the "famebuffer" palette). You can also use sprites to enhance the BG detail in either system's case.

    If the colors for these games could be optimized to Silpheed's level I think it would have worked fine.
    Again, Star Fox already used a very similar sort of color optimization as silpheed (except Silpheed also shared that palette with the streaming BG . . . so arguably MORE carefully optimized there). Games like X-Wing and Star Wars Arcade already used color schemes well suited to optimizing a 15 color palette.
    Shadow Squadron is too colorful though . . . you'd either get ugly/gaudy color usage with lots of (fairly high-contrast) dithering or (preferably) optimize for a much more limited color set in general. (and omit the custom color options) It wouldn't have to be that gray either, but just optimized around more limited/specific colors . . . (Elite on the ST/Amiga had a bit more color, but gray was the stereotypical choice for polygonal space sims/shooters) Also, sticking to pastel/lower saturation colors would be smart, since those would tend to blend better and allow less obvious dithering and better shading. (desaturated colors wouldn't look weird when shaded with limited other colors -ie shading gradients that transition through different desaturated colors can look OK . . . sticking to true shades of specific colors is extremely limiting in 16 colors and still very limiting in 256 colors -hence why you see subtle "wrong" colors in the approximations used in many examples of 256 color look-up-table shading)


    http://www.gamepilgrimage.com/sites/...uadron_171.JPG

    Would end up more like this (please no lectures on tile formats and not using JPGs, I know):
    http://www.gamepilgrimage.com/sites/...1_16-color.jpg
    I was thinking more of a total recolor . . . since you're only going to be using checkerboard dithering, the combinations for blended pseudo colors is limited (even more limited in actually getting different-looking colors/shades too -some combinations might look the same)

    So, no I wouldn't really consider it looking like that specifically.

    I agree, a game like X-Wing should have been adaptable to the Sega CD especially if SVP was involved.
    TBH, most PC games up to 1993 probably would have been doable on the Sega CD. X-Wing and Doom would be kind of the far end of "practical" in that sense, more so for Doom due to the memory requirements. (X-Wing 1 MB vs Doom 4 MB) You'd need a lot of optimization and selective omission/quality reduction in textures/sprites/SFX for Doom. (if you did fixed per-sector lighting like the 32x -actually static lighting for sprites- it might be reasonably possible to optimize Doom color-wise too, especially dithered -since it would be 1/2 res anyway)

    For SVP games in general:
    Compared to Super FX you're better off overall by a good margin . . . DMA bandwidth is better (updating VRAM framebuffer), higher res is possible (also better for dithering), and SVP added more RAM than Super FX (and more processing resource than Super FX2 -at least taking the 68k into account).
    The DMA/color optimization stuff also generally applies to the Sega CD, of course.

    Compared to 32x though, for full-frame framebuffer rendered games (so not FX animation to specific sprites/tiles, decompression, sound mixing, etc), you'd generally be limited to smaller screen sizes and/or resolutions than the 32x allows.
    This is for several reasons: DMA bandwidth limits max framerate and/or screen size, but using smaller screens means less data to copy and higher max framerates. Additionally, clipping the screen (letterboxing) can be used to add vblank DMA time and greatly increase bandwidth. (h20 has proportionally higher bandwidth than h16, so a high res mode game of the same screen size could run just as fast, or one at the same resolution -smaller screen size- could run at higher framerates in H20)
    There's also the issue of leaving space for BG tile data (if using a separate 2D BG), and if you're double buffering the framebuffer layer, you can use up space pretty fast. (if not double buffering, you'd tend to have screen tearing since you can't update at 60 FPS unless you use small screens and/or a huge amount of letterboxing)
    PAL stuff is much better off bandwidth wise (much more vblank time per frame), but PAL-specific optimization would tend to be relatively limited.

    For Shadow Squadron, the 2D BG really isn't a problem, so you could potentially use a screen of a very similar resolution, but practically speaking you'd probably at least use a moderately lower vertical res for DMA reasons. (it's already 288 wide, so 288x208 probably would have worked well, and allow updates up to 20 FPS)
    Again, all this also applies to Sega CD games . . . or plain MD games using software rendering.


    Yes, they clearly wanted the Saturn to be the flagship product and so they made the 32X not only a stop-gap but a late term abortion.
    The plan with the Saturn was always going to be the flagship next-gen product . . . well "always" since the Saturn design was definitively laid down . . . and barring strange late-development delays, or (for whatever reason) scrapping of the project entirely. (none of which should have been the case by late 1993)

    The US Saturn launch was stupid all-around though. It was made stupider by the position of the 32x, but even ignoring that it was horrible in every possible way for Sega. (software developers, retailers, pricing -inciting Sony, distribution, consumer perception, available software, overall marketing cost/investment/hype invested . . . and lost potential build-up for a proper fall launch that would easily flow into the holiday sales season)

    This is a different topic than I intended this thread for, but:
    IMO, with the way things were going prior to that stupid launch of the Saturn, the best thing Sega could have done was the continue (vaguely) as planned with the 32x, but just tactfully scale things back as events developed so as to better redistribute resources between all of Sega's active platforms. At that point, focus should have been on damage control, maintaining still currently profitable older platforms, and carefully working towards the transition to the next generation.
    Part of this would include maintaining some degree of 32x software development and overall market presence (try to at least minimize monetary and PR loss, or possibly even eek out some profits in the long-run) while carefully managing things to also facilitate a reasonable introduction of the Saturn as well. (while 32x wasn't a specifically good platform to complement Saturn technically, Sega still could have taken steps to further facilitate ease of development for the platform and ease of transition to the Saturn -including multiplat programming- . . . plus, careful PR work in press releases, advertising, and keeping 3rd party developers in the loop for planned support)

    While I maintain that releasing the 32x itself was a bad idea compared to various alternatives Sega had, I certainly also maintain that Sega could have made both the 32x and Saturn far more successful than they were historically had the right management steps been taken.
    The problem was that Sega just made mistake after mistake after terrible mistake, and didn't have the luxury of all the competition screwing up too. (ie had Sony been as incompetent at managing/marketing a console as NEC had been -with similar monetary resources- then Sega's position wouldn't have been quite as bad by comparison )

    What Sega of America suggested is what I bought in 1994 and expected to see supported at least into 1996 if not beyond. It is the simple difference between a "stop gap" and a lower end budget system.
    The 32x didn't make sense as a lower-end budget system though; the hardware is just too limited for that, or too expensive to compete well in the budget market alongside the SNES and MD.

    Or, that is to say, it didn't make much sense as a budget system for SEGA. If some 3rd party wanted to get into the market with a new system, then something in line with the 32x's hardware would be understandable (albeit still pretty weak to be competitive). Or, if Sega really didn't have any good next-gen options at the time (say the Saturn either became unworkable hardware wise -and was scrapped- or they'd taken a different design that ran into delays a la N64) and thus a simple alternative with very short design cycle was thus an attractive option. (even then, they should have added at least a little more than what the 32x got -to make it more realistic)

    But the point is that Sega DID have better options, both for targeting the low end segments, and for the next gen in general. If they wanted a next-gen class platform that could be released almost immediately and fit into the mainstream or "budget" segment as-is, then Jupiter was the most practical choice given the context of established plans for the Saturn design in general. (the rest of the low-end being covered by MD/CD) Or, if they really wanted to add some extra power to the MD itself (and specifically a lower end of the market than the Jupiter itself would cater to -including building on existing MD users), then SVP also makes far more sense than 32x. (more limited, sure, but also much cheaper, much less risky to aim at the "budget" end of the market . . . and less risky in general -much less monetary and PR loss if it failed)


    My interest in the 32X (or any like kind of late Genesis add-on) is certainly not mass market, but I think the 32X actually was of mass market interest. Certainly enough to boost Sega's revenues in certain regions while the Saturn and PS1 slowly took over market share from the 16-bit consoles.
    It was obviously intended as mass-market . . . and it probably could have been a modest success (or at least much less of a detractor) if things had been handled differently after the fact. (as above)

    However, my argument remains that Sega had much better alternatives at their disposal than the 32x itself.

    For myself, heck, I just turned a Pentium III 500Mhz machine into a 90s game box just to see the Matrox Mystique at its best. Back in the day I couldn't get half the games I'm playing now to run just because I only had a Cyrix 686 133Mhz coupled with that Mystique, now everything runs great and who cares about texture filtering?

    Why not release the full blown Saturn in the West in 1994 then? Cost factors for one, imagine the internet rumors today if the Saturn launched at $500. Also, they apparently couldn't make enough of the things for the Japanese market alone and didn't have enough SH2s to replace a faulty one in an early 32X development kit. Summer of 1995 Sega apparently couldn't supply all retailers with Saturn systems and so chose the limited "pre-launch" approach instead. No, the lack of a CD-ROM subsystem wouldn't have changed these facts for Jupiter.
    Given the shared/competing engineering and manufacturing resources of the 32x and Saturn, then yes, at very least some of those factors should have changed. Specifically planning for manufacturing logistics as such would be significant . . . getting sufficient volumes of the core chipset would be the main difference. (SCU, VDPs, SCSP)

    That said, this could also be an area where the SVP would come back into play. As addressed above, they could have used the SVP module as an interim/fallback solution as well as a potentially significant product in its own right. (maintained in a distinct market segment than Jupiter or Saturn in the long run)
    I'm not sure a 1995 Jupiter would have been totally worthwhile though . . . the price point possibilities can't be ignored, but there's a lot of trade-offs compared to going straight to Saturn instead. Then again, it made a hell of a lot more sense than Nintendo launching a cart based system in 1996, and one with no CD module option (let alone standalone CD counterpart) . . . or even plans for a CD module later on. (which would have made far more sense than the DD -a proprietary CD-based format should have meant much cheaper media and drive hardware than the DD along with much higher capacity . . . releasing it in 1999 would still have been ridiculous though, but '97 might have been understandable -totally different topic though )

    Right, so that leaves MKII, Knuckles Chaotix, Star Wars Arcade, Doom and maybe Virtua Racing Deluxe as relevant mass market products. Space Harrier and Afterburner, and T-Mek were apparently just easy enough to port/adapt that they were worth it to do so. However Carmack got Doom 32X running so smooth in a mattter of weeks is pretty amazing too. I don't think the SVP would have been as quick/easy for adapting these games. Doom SNES SFX-2 coming out in 1995 pretty much points to that, but at least it got more of the original levels in.
    TBH, several of these games would make more sense targeting the Sega CD than the SVP . . . unless they HAD to target the cart market. (aside from going SVP+CD) Sega actually putting out documentation to facilitate use of cart+CD simultaneously would have been significant too. (regardless of SVP)

    MKII was already done very well on the MD alone, so that's not really an issue IMO. (a Sega CD version of equal quality optimization would have been cool though -unlike the CD version of the first game)

    Space Harrier and Afterburner on the CD (while less relevant than some alternatives -including hypothetical original titles), still could have made decent additions and good tech demos, especially if released early enough. But, again, that's kind of another topic too. (I still think it's weird that SoJ managed to put together a bunch of games within a few months of the 32x's completion, but didn't do anything close to that for the Sega CD -which should have been a far more substantial platform from their perspective)
    6 days older than SEGA Genesis
    -------------
    Quote Originally Posted by evilevoix View Post
    Dude it’s the bios that marries the 16 bit and the 8 bit that makes it 24 bit. If SNK released their double speed bios revision SNK would have had the world’s first 48 bit machine, IDK how you keep ignoring this.
    Quote Originally Posted by evilevoix View Post
    the PCE, that system has no extra silicone for music, how many resources are used to make music and it has less sprites than the MD on screen at once but a larger sprite area?

  13. #43
    I remain nonsequitur Shining Hero sheath's Avatar
    Join Date
    Jul 2010
    Location
    Texas
    Age
    42
    Posts
    13,313
    Rep Power
    130

    Default

    Quote Originally Posted by kool kitty89 View Post
    This a a huge oversimplification.

    For one thing, developing for ANY other platform is always going to consume more time and resources. The more different that platform is in raw capabilities and programming/rendering techniques requires as well as raw programming complexity, the greater the burden is to develop for that platform. Equally, any resources going to work on said platform can and will detract from that of other platforms.
    I think I might have let my point get lost in technical discussion again. The question I am addressing here is whether the SH2s, and overlay graphics style, in the 32X could help game developers get more familiar with the Saturn's architecture.

    Quote Originally Posted by kool kitty89 View Post
    For now, let's stick to 32x vs Saturn vs Jupiter programming specifically. With 32x, you've got the MD hardware to work with and the SH2s doing software rendering and (maybe) software mixed sound . . . at the time (even in 1995), documentation issues screwed up the sound capabilities. (this is a significant point to include, as documentation is a big issue for the Saturn -and Sega CD too)

    1) Games on the 32x may use the MD graphics for a lot, a little, or none at all (usually the far backdrop at least) [sheath: same goes for the VDP2]

    2) any graphics done on the 32x end must be software rendered. [sheath: big difference I agree]
    3) One or both of the SH2s will basically be taking the place of a GPU as well as handling the 3D math and game logic/AI/etc. (some games might make more use of the 68k)
    [sheath: replace GPU with GTE/Math Co Processor and I fail to see how this is any different than the Saturn.]
    4) You've got relatively limited work RAM that barely made it very difficult to design some types of games for,
    [sheath: This phrase isn't clear, but it might be argued that the work RAM plus ROM bandwidth was significantly different than the Saturn needing to load and process data from CD. Also, with the SH2s in the Saturn needing to manage high and low RAM the concepts don't sound too far off to me, though obviously different in overall purpose and potential.
    I have attempted to piecemeal reply in the above quote.

    Quote Originally Posted by kool kitty89 View Post
    For a reasonably good looking multiplat game with the 32x as a target, you've got to optimize the art for both the memory(ROM and RAM)/bandwidth limitations and the color limitations (depending on color depth on the framebuffer and how MD graphics are used), plus the sound, and overall processing/rendering. It wouldn't be too tough to design a game specifically optimized around the 32x and then port that over directly (with minimal changes) to more powerful systems, but then you've got a game's potential being heavily limited by the weakest platform. (lowest common denominator effect)
    I think we were talking about forward compatibility coupled with familiarizing developers with the Saturn's needs during the slow 5th gen start. How "5th gen" any 32X games look is irrelevant to the conversation in my view. We saw the lowest common denominator effect in the 6th gen with ports from PS2 to Xbox and Gamecube, and ports from PS1/PC to Dreamcast as well, but I don't see how that is on topic here.

    Yes, the Neptune would have been a mostly different hardware design than the Jupiter as we have discussed. What I see in the Neptune is a similar main setup to that of the Saturn though, just a lot cheaper (and less powerful) and not requiring the consumer to purchase all of it. Ignoring the sound subsystem as well, as we can assume that would have changed in the Jupiter from the Saturn as well, the graphics are set up conceptually very similarly as opposed to a fully 3D system like the PS1 or 3D acceleration cards. The idea of using the 2D aspects of the system to optimize the performance of the 3D aspects is so similar I can't see a difference, regardless of whether the 3D is software or hardware rendered.

    Quote Originally Posted by kool kitty89 View Post
    Programming experience for the 32x would not be directly applicable to Saturn programming.
    Perhaps not directly, but conceptually at the game design level I see a very solid connection that is validated by the 32X's relative low cost. Imagine if Akklaim, EA, Sega and a couple of other studios became much better over, say, a couple of years at making 3D games with extensive exploitation of 2D backgrounds and they then set themselves to the task of creating a game for the ground up for the Saturn. Consider even the fact that everything was software rendered on the 32X, and then consider custom VDP2 transparency and floor/ceiling effects on the Saturn. No, they would not be as familiar with the design as if they had been programming for a cart based Saturn the whole time, but I think that is a big if besides.

    We are assuming that the price of the Jupiter in 1994 would have been the same "shot in the arm" financial boon that the 32X was the same year. At $250 I doubt the Jupiter would have sold much better than the PS1 or Saturn did in 1995, and that wasn't as well as people like to think in comparison to the real world 32X. I'll have to address the rest of this post later.

    Quote Originally Posted by kool kitty89 View Post
    Sure, learing SH2 assembly would be useful (but it's a well designed, well-supported, bug-free CPU with a straightforward ISA that any assembly language programmer should learn quickly -even the Jaguar RISC language could be learned quickly). More important would be experience with multiprocessing with the SH2s in that configuration. However, that experience has a very limited overlap with how the SH2s are most useful in the Saturn. 32x programmers would be emphasizing software rendering techniques and totally different uses for multiprocessing (like 1 dedicated to graphics, one to 3D math+game logic), or even optimizing for games making significant use of the 68000 (offloading some game logic overhead -or all of it in simpler -but graphically demanding- games). 2D games might barely use the SH2s at all on the Saturn while being hugely intensive on them in the 32x with very different programming techniques being used in each.
    Not just SH2 assembly, but multiprocessing with SH2s and on a 16-bit bus at that. Introduce that and then give the same developer a Saturn, with its 32-bit High RAM, and do you think they would do better or worse at Saturn programming?

    Quote Originally Posted by kool kitty89 View Post
    With the Jupiter, by comparison, programmers would learn to use the hardware in basically the same manner as the Saturn itself. They could learn to optimize around the VDP capabilities, sound capabilities, and use the CPUs in basically the same capacities, thus focusing on the strengths, weaknesses and general quirks of the system. Even with documentation and tools just as sketchy as they were early on for the Saturn, developers would at least have a head start at working around those problems.

    More likely IMO, is that pushing the Jupiter as such in 1994 would also give a general boost in resources focusing on compiling better documentation in general (and better translations of existing documentation) as well as building more comprehensive tools for use in the SDK "out of the box." ...

    Additionally, since you'd basically be coding a Saturn game outright, it would be pretty straightforward to expand upon that for a CD-ROM version of the game and/or to release expanded content on CD. Some games heavily exploiting ROM-specific advantages would tend to only favor the latter.
    Yes, obviously Jupiter development would be more directly comparable to Saturn development. Whether the Jupiter would have resulted in better Saturn tools is debatable. There is still a big question mark as to whether $250 or more in 1994 would have resulted in a mass market product any more than a $500 Saturn would have.

    Quote Originally Posted by kool kitty89 View Post
    There's also the manufacturing/distribution/marketing logistics to consider too:
    The RAM content of the Jupiter would allow games to be heavily compressed and load data into RAM as needed, only using uncompressed (or lightly compressed) data when it was really favorable (ie the performance boost would be worth the space/cost). That would mean load times, but they should still have been faster than CD load times. (assuming efficient decoding algorithms were used) With the 32x, you have little space to decompress into, so it's mostly down to uncompressed data and continually streaming and decompressing data. (much more CPU overhead and/or weaker compression)
    So, like N64 games, you'd get a lot more bang for your buck at a given ROM size. (and, unlike the N64, the cost overhead for CD-ROM drives was still significant enough to be a major burden when the system was released -for 1996, that was just silly though)
    I've been glancing over the cost of CD-ROM previously, but how much do you think a CD-ROM sub system cost in 1994-95? I'm thinking no more than $150, and $200 for an add-on.

    Quote Originally Posted by kool kitty89 View Post
    Further, even for games specifically made as cart-only for the Jupiter and not expanded/re-released with Saturn CD installments, you'd still be directly expanding the Saturn's software library since Jupiter cart games would be directly compatible. -As a very vague scenario, let's assume that Jupiter only got similar support in '94/95 to what the 32x already had (quantity/quality wise), but that all software at least reflected the system's general capabilities (roughly as well as 32x games did for that platform). So you'd have some pretty decent early-gen Saturn-quality looking games (some specifics due to ROM formats) with nice sample/synth sound, so genuine "next gen" class software. And then apply that to the Saturn's launch lineup as well . . . perhaps also with the addition of some otherwise cancelled (would-be) 32x projects and some degree of increased quality in the early (Western) Saturn software due to the head start in learning the architecture.
    32X games should have indirectly expanded the Saturn library as well by means of more up ports like Sonic 3D Blast and Stellar Assault. Also, I'm detecting an assumption that the Saturn needed more games in 1995, which probably isn't the case. The Saturn might have needed more texture mapped games and better optimized games than VF and Daytona to sell it, but it didn't need more games. Least of all more games that looked or sounded inferior to PS1 launch games. Similarly, full blown Saturn development kits were available in the same quality as third party PS1 kits by summer of 1995, which is before the Saturn should have launched in the West anyway.

    Quote Originally Posted by kool kitty89 View Post
    Even if (for some inconceivable reason) Sega still went with the stupid, poorly planned, May 1995 "launch" of the Saturn in the US, at least it would have a much better established infrastructure to work with. (Jupiter games, Jupter CD -presumably released at the same time- allowing an affordable upgrade path for existing -and future- Jupiter owners; and the Jupiter itself being a complement to the Saturn's market position, rather than a segmented mess with the 32x --both fit together rather well, both play the same cart games, one can be upgraded to allow the rest of the CD games too, one is already at a mainstream price point while the other will stick to the higher-end niche for a while)
    The US "pre-launch" has one definitive thing going for it, for some silly reason the first console to launch of a new generation generally does quite well. I think the limited pre-launch has been journalistically blown way out of proportion. I also find it ironic that the NES' 1985 test market is considered the national launch by most journalistic history while the Saturn seeing a limited early release by mere months is somehow the biggest blunder ever. Pettus be praised for propagating such an idea.

    In regard to the Jupiter and Saturn scenario, I am honestly not sure if I would have received the Jupiter in May of 1995 any better than I did the Saturn (or PS1 for that matter). The games just didn't appeal to me at that point, but I don't count. Why I was willing to jump on the 32X in November 1994 for $160 had as much to do with enhancing my Genesis as it did with Virtua Racing Deluxe, After Burner and Space Harrier. An all new console at that point probably wouldn't have appealed to me as much. I found the Jaguar's $250 price point daunting as well considering the games I saw.

    Quote Originally Posted by kool kitty89 View Post
    TBH, the real arguments aren't whether Jupiter would have made more sense than 32x, but whether it made more sense than SVP or nothing at all until Saturn.
    I'm really not sure what I'd decide on this. With the Jupiter, there's a genuinely good argument (IMO) for releasing an intermediate (NOT interim) platform on the market to smooth transition to the next generation. Costs of CD-ROM drives was a big part of that, and the main reason the new hardware couldn't reach mass-market home console friendly prices right away at the time. (which had not been the case for the previous 2 generations, albeit most pre-crash consoles had been very expensive at launch -taking inflation into account-)
    -Perhaps another reasonable option from the early 1994 perspective would have been to implement the SVP as a bundled module with VR (and launch at the same date VR was planned to), and then push plans for Jupiter along as well. Use the short-term public response (and 3rd party interest) in the SVP that Spring to guide further plans for SVP support and the Jupiter. If SVP is well accepted all-around, they might consider dropping Jupiter before officially announcing/unveiling it (probably at the Summer CES), or decide to shift away from SVP in favor of emphasizing Jupiter. (status of the MD/CD would also contribute to that) Finally, if the Saturn chipset hit delays, then Jupiter would be off the table for the immediate time, so SVP would remain the de-faco choice in the interim anyway.
    I think you are overlooking some of the argument for 32X to say that there is no real contest. Jupiter would be a full blown next generation console in 1994 dollars and the SVP gave us this:



    and it cost half to two thirds of what gave us this:


    Quote Originally Posted by kool kitty89 View Post
    Oh, and technically, the SVP would share many of the programming/development disadvantages as the 32x list includes above (actually worse in some respects -limited RAM, limited CPU resource, no use of SH2s, etc). However, as an overall product, there's still many advantages: namely cost, but also convenience of installation (compare a lock-on cart to the 32x). In terms of cost, I don't just mean to the consumer, but to Sega in general. The risks for PR and monetary loss from the SVP would be much smaller than those of the 32x since it would be far cheaper to manufacture, cheaper/easier to distribute, and (arguably) easier to market.
    I would say that the SVP has none of the programming similarities with Saturn and cost is its only advantage.

    Quote Originally Posted by kool kitty89 View Post
    Well, either dithered or posterized . . . in either case minimized by optimization. It's just like any MD vs SNES 2D comparison really. Actually, it's better for the MD than usual, since you've got 3 palettes to dedicate to the BG without contention with sprites (and the optional use of the "famebuffer" palette). You can also use sprites to enhance the BG detail in either system's case.

    Again, Star Fox already used a very similar sort of color optimization as silpheed (except Silpheed also shared that palette with the streaming BG . . . so arguably MORE carefully optimized there). Games like X-Wing and Star Wars Arcade already used color schemes well suited to optimizing a 15 color palette.

    Shadow Squadron is too colorful though . . . you'd either get ugly/gaudy color usage with lots of (fairly high-contrast) dithering or (preferably) optimize for a much more limited color set in general. (and omit the custom color options) It wouldn't have to be that gray either, but just optimized around more limited/specific colors . . . (Elite on the ST/Amiga had a bit more color, but gray was the stereotypical choice for polygonal space sims/shooters) Also, sticking to pastel/lower saturation colors would be smart, since those would tend to blend better and allow less obvious dithering and better shading. (desaturated colors wouldn't look weird when shaded with limited other colors -ie shading gradients that transition through different desaturated colors can look OK . . . sticking to true shades of specific colors is extremely limiting in 16 colors and still very limiting in 256 colors -hence why you see subtle "wrong" colors in the approximations used in many examples of 256 color look-up-table shading)

    I was thinking more of a total recolor . . . since you're only going to be using checkerboard dithering, the combinations for blended pseudo colors is limited (even more limited in actually getting different-looking colors/shades too -some combinations might look the same)

    So, no I wouldn't really consider it looking like that specifically.
    I thought I was being generous, as SVP Virtua Racing's lower resolution makes dithered pixels even larger and visible even from Composite. I think posterized colors would be the best way to go unless they could get the resoluton to 320 wide, which I doubt.

    Quote Originally Posted by kool kitty89 View Post
    TBH, most PC games up to 1993 probably would have been doable on the Sega CD. X-Wing and Doom would be kind of the far end of "practical" in that sense, more so for Doom due to the memory requirements. (X-Wing 1 MB vs Doom 4 MB) You'd need a lot of optimization and selective omission/quality reduction in textures/sprites/SFX for Doom. (if you did fixed per-sector lighting like the 32x -actually static lighting for sprites- it might be reasonably possible to optimize Doom color-wise too, especially dithered -since it would be 1/2 res anyway)
    Exactly, as I mentioned above, we're talking limited framerate, lower resolution 15-color games with minimal sound enhancements. I am barring a fluke sound engine like Virtua Racing Deluxe's being used when such a thing didn't happen on Genesis titles for some reason. I don't think even Virtua Fighter 2 on Genesis, by the same team, sounds as good as Virtua Racing Deluxe sample wise. It's a real mystery.

    Quote Originally Posted by kool kitty89 View Post
    For SVP games in general:
    Compared to Super FX you're better off overall by a good margin . . . DMA bandwidth is better (updating VRAM framebuffer), higher res is possible (also better for dithering), and SVP added more RAM than Super FX (and more processing resource than Super FX2 -at least taking the 68k into account).
    The DMA/color optimization stuff also generally applies to the Sega CD, of course.

    Compared to 32x though, for full-frame framebuffer rendered games (so not FX animation to specific sprites/tiles, decompression, sound mixing, etc), you'd generally be limited to smaller screen sizes and/or resolutions than the 32x allows.
    This is for several reasons:...-smaller screen size- could run at higher framerates in H20)
    There's also the issue of leaving space for BG tile data (if using a separate 2D BG), and if you're double buffering the framebuffer layer, you can use up space pretty fast. (if not double buffering, you'd tend to have screen tearing since you can't update at 60 FPS unless you use small screens and/or a huge amount of letterboxing)
    PAL stuff is much better off bandwidth wise (much more vblank time per frame), but PAL-specific optimization would tend to be relatively limited.
    Right, I don't see any comparison between what the SVP/SFX-2 could do for their respective platforms versus 32X. When it comes to which ones could support the most "5th gen" looking games while competing with Jaguar and 3DO titles the 32X wins hands down while still being "peripheral" priced.

    Quote Originally Posted by kool kitty89 View Post
    The plan with the Saturn was always going to be the flagship next-gen product . . . well "always" since the Saturn design was definitively laid down . . . and barring strange late-development delays, or (for whatever reason) scrapping of the project entirely. (none of which should have been the case by late 1993)

    The US Saturn launch was stupid all-around though. It was made stupider by the position of the 32x, but even ignoring that it was horrible in every possible way for Sega. (software developers, retailers, pricing -inciting Sony, distribution, consumer perception, available software, overall marketing cost/investment/hype invested . . . and lost potential build-up for a proper fall launch that would easily flow into the holiday sales season)
    I don't think the pre-launch was as absolutely horrible as people like to think. To consumers (the 50 million + that buy mainstream devices) the Saturn pre-launch was as obscure as the 32X and Sega CD were already. The one kid I knew who got one felt like he was let in to an early sneak peak, not the full console launch. Usenet posts from Summer of 1995 express this as well, and EGM and Gamepro issues from the time were very positive of the Saturn. The narrative of the horrible mistep launch came much later and originated from Edge/Next Generation not consumer impressions.

    Quote Originally Posted by kool kitty89 View Post
    ...
    The 32x didn't make sense as a lower-end budget system though; the hardware is just too limited for that, or too expensive to compete well in the budget market alongside the SNES and MD.
    You keep saying this and I keep wanting to think you mean "to you" but I'm starting to doubt that. The 32X made sense from a variety of perspectives I have already described. The biggest thing that makes it questionable is its cancelation after six months and resultant small library of early games. This is arguable, but I would say that the SNES and Genesis were beyond budget at that point, they were aging/old tech that only super late adopters were still buying and in a totally different market region than "budget console". Back in 1993 they might have filled the role of being the low end market, but after Fall of 1994 they weren't even talked about as the mainstay consoles anymore. We're talking about the difference of buying a Retrogen today instead of a $99 360 or Wii (for whatever).

    Quote Originally Posted by kool kitty89 View Post
    Or, that is to say, it didn't make much sense as a budget system for SEGA. If some 3rd party wanted to get into the market with a new system, then something in line with the 32x's hardware would be understandable (albeit still pretty weak to be competitive). Or, if Sega really didn't have any good next-gen options at the time (say the Saturn either became unworkable hardware wise -and was scrapped- or they'd taken a different design that ran into delays a la N64) and thus a simple alternative with very short design cycle was thus an attractive option. (even then, they should have added at least a little more than what the 32x got -to make it more realistic)

    But the point is that Sega DID have better options, both for targeting the low end segments, and for the next gen in general. If they wanted a next-gen class platform that could be released almost immediately and fit into the mainstream or "budget" segment as-is, then Jupiter was the most practical choice given the context of established plans for the Saturn design in general. (the rest of the low-end being covered by MD/CD) Or, if they really wanted to add some extra power to the MD itself (and specifically a lower end of the market than the Jupiter itself would cater to -including building on existing MD users), then SVP also makes far more sense than 32x. (more limited, sure, but also much cheaper, much less risky to aim at the "budget" end of the market . . . and less risky in general -much less monetary and PR loss if it failed)
    Your first paragraph outlines exactly why the 32X was created. We also don't know what exactly was stripped out of the 32X, it might have been something very comparable to VDP1, more RAM or both, making it much more "Jupiter like" than the peripheral priced 32X add-on is.

    Quote Originally Posted by kool kitty89 View Post
    ...
    TBH, several of these games would make more sense targeting the Sega CD than the SVP . . . unless they HAD to target the cart market. (aside from going SVP+CD) Sega actually putting out documentation to facilitate use of cart+CD simultaneously would have been significant too. (regardless of SVP)

    MKII was already done very well on the MD alone, so that's not really an issue IMO. (a Sega CD version of equal quality optimization would have been cool though -unlike the CD version of the first game)

    Space Harrier and Afterburner on the CD (while less relevant than some alternatives -including hypothetical original titles), still could have made decent additions and good tech demos, especially if released early enough. But, again, that's kind of another topic too. (I still think it's weird that SoJ managed to put together a bunch of games within a few months of the 32x's completion, but didn't do anything close to that for the Sega CD -which should have been a far more substantial platform from their perspective)
    I think Sega saw CD-ROM as a completely different market than cartridge games, one that was poorly suited for quick pick up and play Action titles or straight Arcade ports. Furthermore, we are leaving out the likes of Spiderman Web of Fire, Blackthorne and even Knuckles Chaotix (especially the bosses and bonus rounds) from the mix. Besides that, I think without the pressing need for a launch lineup most of these games wouldn't exist or would have been folded into other projects.
    Last edited by sheath; 02-05-2013 at 12:11 PM.
    "... If Sony reduced the price of the Playstation, Sega would have to follow suit in order to stay competitive, but Saturn's high manufacturing cost would then translate into huge losses for the company." p170 Revolutionaries at Sony.

    "We ... put Sega out of the hardware business ..." Peter Dille senior vice president of marketing at Sony Computer Entertainment

  14. #44
    Hero of Algol kool kitty89's Avatar
    Join Date
    Mar 2009
    Location
    San Jose, CA
    Age
    30
    Posts
    9,725
    Rep Power
    64

    Default

    A note on the SVP:
    in the context of the CRAM DMA direct 9-bit color "mode," something like an SH1+RAM would have made a lot more sense than SVP+RAM (SH1 being an MPU rather than a DSP). Manufacturing may not have been cheaper, but R&D should have been less (no need to design the custom parts of the SVP ASIC) and it would certainly offer a lot more flexibility than the SVP. (and offer the modest advantage of adding Super H assembly programming experience)

    There's 2 problems with this relative to my original premise for this thread:
    1. Sega already had the SVP ready for mass production by the time in question (January 1994)
    2. it assumes Sega actually recognized the potential for that direct color hack . . . something that no-one used in any MD or MCD games or demos up to that time. (it doesn't seem like that mystifying of a technique though -and similar to tricks used on several home computers, and it's not very useful on plain MD games either, so only Sega CD programmers would even seriously consider it)

    For all we know, it was considered and simply never used for lack of interest. (it's worse for most types of FMV than the better optimized formats used on the MCD, and it can't really be used for CD ASIC rendering either -not very efficiently at least- . . . so that mainly leaves software rendering -interesting to be sure, but for whatever reason, there was very little interest in software rendered ray-casters or polygonal games on the CD)





    Quote Originally Posted by sheath View Post
    I think I might have let my point get lost in technical discussion again. The question I am addressing here is whether the SH2s, and overlay graphics style, in the 32X could help game developers get more familiar with the Saturn's architecture.
    As I said, it would help transition programmers modestly in some areas (in terms of going from MD to Saturn vs 32x to Saturn), but nothing that would really, consistently, give programmers a significant head start when formally introduced to Saturn programming.

    The video overlay situation is nothing like the Saturn's from a practical game design standpoint. VDP1 and VDP2 are SO much more powerful than software rendering+MD VDP that there's just no comparison. VDP1 alone should outperform any pure 2D stuff the MD+32x could pump out. The genlock video of the SuperVDP+MD VDP is much more limited and only allows the 32x and MD video to be treated as 2 solid planes when mixing. VDP1+VDP1 OTOH can be treated much more like a conventional sprite+BG set-up (be it arcade or console), since VDP2 handles the mixing digitally and can use per-layer and per-tile (and per-pixel -for some things) priority on any of its layers and the VDP1 framebuffer. As crazyace mentioned at one point, it's actually rather like an outgrowth of the System 32's graphics set-up, in some respects.

    The dual SH2 set-up could familiarize things somewhat, but (again) the SH2s are used for very different things in the 32x than the Saturn (unless you're doing a software renderer), and the use of ROM carts coupled with very limited RAM also leads to different sorts of compromises and game design and programming techniques otherwise useful for the Saturn. (even if doing a software renderer)
    And TBH, any developers dabbling in PC software would have a much better leg-up into 3D and multimedia game design anyway. (hell, you could even argue Sega themselves might have been good to get into Sega PC stuff sooner)

    The big problems with developers starting work on the Saturn specifically was learning the system as a whole. Learning the architectures of the custom processors (graphics and sound), the bottlenecks and quirks, and how to best make all that work together . . . plus working around weak documentation and building custom development tools for the lack of a comprehensive SDK. Kind of like what all older consoles went through, except more so due to the complexity (increasing with every tech generation), like with the Jaguar except not as buggy (though more complex in some respects) and with much more investment/interest from developers than the Jag ever saw.

    The Jupiter would do all of those things (aside from the few CD-specific game design areas), and not only would an earlier introduction of that hardware give devs a head start into building their own programming resources (both personnel experience and building tools), but it would very likely lead to Sega themselves having more complete documentation and development tools (maybe even graphics libraries) significantly sooner.

    For that matter, even if manufacturing the Jupiter in quantity for a late 1994 launch was impractical (logistics wise), planning on transitioning to Saturn (or Jupiter -still makes sense in '95) development sooner would make tons of sense for the same reasons as above. (release 1st gen Saturn/Jupiter dev units in mid 1994) Having real Saturn hardware and documentation to work with is the main factor here, as far as developers were concerned. (granted, having an actual platform on the consumer market will make for more immediate incentive for development -which isn't necessarily all good either, given it could distract from support for the MD/CD or lead to more rushed development of Saturn/Jupiter stuff)


    --Of course, this is all an argument for the Jupiter/Saturn specifically, not for the SVP. (which has less to do with the Saturn and more to do with maintaining Sega's mainstream market in 1994)



    I think we were talking about forward compatibility coupled with familiarizing developers with the Saturn's needs during the slow 5th gen start. How "5th gen" any 32X games look is irrelevant to the conversation in my view. We saw the lowest common denominator effect in the 6th gen with ports from PS2 to Xbox and Gamecube, and ports from PS1/PC to Dreamcast as well, but I don't see how that is on topic here.

    Yes, the Neptune would have been a mostly different hardware design than the Jupiter as we have discussed. What I see in the Neptune is a similar main setup to that of the Saturn though, just a lot cheaper (and less powerful) and not requiring the consumer to purchase all of it. Ignoring the sound subsystem as well, as we can assume that would have changed in the Jupiter from the Saturn as well, the graphics are set up conceptually very similarly as opposed to a fully 3D system like the PS1 or 3D acceleration cards. The idea of using the 2D aspects of the system to optimize the performance of the 3D aspects is so similar I can't see a difference, regardless of whether the 3D is software or hardware rendered.

    Perhaps not directly, but conceptually at the game design level I see a very solid connection that is validated by the 32X's relative low cost.
    Again, any major 3rd parties could be dabbling in 3D (and more advanced multimedia) on the PC at this point (the most viable 3D platform on the market in general at that point). Sega themselves had the arcade teams working on their own stuff (and Japanese Saturn devs already) as far as pure game design went. So that leaves Sega's in-house non-arcade developers . . . some game design aspects could already have been pushed on the MCD (and were to some extent, but not really by in-house teams), and otherwise could be pushed forward to start early development of Saturn stuff . . . or consider branching out to PC development too. (a 1994 Jupiter release would make that more immediately emphasized, but a lower intensity effort for early Saturn/Jupiter development for a planned '95 launch could have worked nicely too -it was a matter of getting documentation and development systems distributed earlier, and management/planning)

    The 32x is limited enough to not be that much better in game design potential than the Sega CD alone (or SVP). Its limitations put it behind the potential of what 3rd developers could push in higher-end PC games in 1994, and limit things well beyond what could be pushed on the Saturn too.
    If they'd wanted to get into pushing 3D/pseudo-3D game design in general, they should have pushed that on the Sega CD itself much earlier (several 3rd parties took that upon themselves too). By 1994, though, all the alternative options I mentioned above (as far as game design experience) would make far more sense. (particularly pushing western Saturn dev kits sooner in general -regardless of launch date)


    We are assuming that the price of the Jupiter in 1994 would have been the same "shot in the arm" financial boon that the 32X was the same year. At $250 I doubt the Jupiter would have sold much better than the PS1 or Saturn did in 1995, and that wasn't as well as people like to think in comparison to the real world 32X. I'll have to address the rest of this post later.
    You mean revenue boon? (surely not profits)

    As far as sheer revenue boosting, the 32x was a plain bad idea. Focusing on MD/CD software and competent marketing is what they needed on that front, and they'd already proven that (most of their mass market success was driven by good marketing). If they wanted to artificially inflate revenue by inciting hardware sales (bad for profits for any pretty much console company), then there should have still been room to push the Sega CD harder in '94 (lots of room for marketing that, especially with the price drop late that year). SVP would have some room here too . . .

    But, looking at Jupiter specifically . . . I can see plenty of scenarios where it would have driven up revenue better than the 32x at the time. For one, it was more expensive, so even if it got only 2/3 the actual consumer sales rate (including software) of the 32x or Jupiter, it should have been similar. On top of that, you've got to look at the people who might be buying it:
    -long-time Genesis owners who have enough money saved up (or on hand) to look into a new platform,
    -someone with more money to spare in general who just bought a genesis but is willing to upgrade immediately,
    -or a non-genesis owner who's interested in buying a new system in general (could be moving over from a competitor or could have skipped the generation entirely -plenty of people/families who hadn't yet gone beyond owning an NES or a gaming computer in 1994 and might be interested in skipping a generation).

    The only one of those cases where the 32x a real edge over the Jupiter is the long-time genesis owner one, assuming they're on a tighter budget. Even then, the power of the Jupiter should have been compelling enough to sway many people beyond what the 32x offered. (screenshots and game footage for one, but much more so for kiosk displays in-stores or for rentals, or seeing it in action at a friend's house) Sensationalistic ads will do a lot by themselves, but having concrete power to back it up can make marketing magic happen. Getting consumers to see games in motion (especially in person) is a HUGE part of marketing for a technically impressive platform. (software+hardware)

    The other (bureaucratic/PR driven) reason behind the 32x would be market prestige and hype of having new hardware out in general, and this is something the Jupiter should have been able to pull off much better than 32x. (on top of all the real advantages)

    Not just SH2 assembly, but multiprocessing with SH2s and on a 16-bit bus at that. Introduce that and then give the same developer a Saturn, with its 32-bit High RAM, and do you think they would do better or worse at Saturn programming?
    Again, the problem there is that the 32x uses the SH2s extremely different than good Saturn games should.

    On most 32x games, the brunt of work the SH2s handle is software rendering and (sometimes) sound mixing. A much smaller chunk of CPU resource will be available for actual game logic and such, and the limited RAM in the system will force different programming and game design methods than would be possible for Saturn games. (you can cut down a game to work in very limited RAM, but that's not teaching you to make use of the advantages of what the Saturn can push)

    A lot of Saturn games might use 1 CPU to compute the 3D stuff and the other to handle the game engine itself . . . optimizing beyond that to max out both CPUs as best as possible would be the trick there. (or making use of the SCU DSP when applicable)
    The 32x doesn't have a good counter example for this.
    The Saturn had enough CPU resource for 2D game logic to be relatively trivial without needing multiprocessing . . . 32x needs to slave one CPU for 2D rendering in general. (and might use the other to run the game, or pass that to the 68k, or share it with the rendering CPU, depending on the specific example)

    Honestly, if the 32x had only had 1 SH2 along with the Saturn's VDP1, even that probably would have given better programming experience to apply to the Saturn. (and still a ton of differences in that case too)

    Yes, obviously Jupiter development would be more directly comparable to Saturn development. Whether the Jupiter would have resulted in better Saturn tools is debatable.
    I don't think there's any question about it. Launching the jupiter in 1994 would mean having dev kits out for the Saturn chipset much earlier. And, assuming the documentation was about the same as it was for the Saturn's initial dev kits (ie roughly translated documentation and some sample code to work with), then developers would have that much more time to work around those limitations even if further improvement to the SDKs didn't appear any sooner than Sega managed historically. (I find the latter hard to believe too though, since SoA dev teams should have been approaching the programming difficulties as well or better than 3rd parties)

    Again, much of that could have been done without the 32x or Jupiter being released in 1994 at all. The issue here is more with making Saturn documentation and development systems available sooner. (in that sense, Jupiter is just an example of a catalyst to force that in the given situation -ie, with Sega being very interested in launching an intermediate platform in 1994)

    There is still a big question mark as to whether $250 or more in 1994 would have resulted in a mass market product any more than a $500 Saturn would have.
    It easily would have been much more mainstream than trying to launch the Saturn in 1994 . . . in terms of raw short-term marketability, I'd argue no worse than the 32x itself. (potentially better depending on how they managed it, but obviously much better for the long run)

    Similarly, it would have fit more into the mainstream than the Sega CD did at its launch. (or a lot longer than that even, since the $230 price point lasted until late 1994)

    Hell, it would have been more a mass market product than the Genesis in 1989 too. (relatively similar price point -for the time- and much better marketing/market position)


    I've been glancing over the cost of CD-ROM previously, but how much do you think a CD-ROM sub system cost in 1994-95? I'm thinking no more than $150, and $200 for an add-on.
    The SH1 and 512 kB of SDRAM would have inflated it. With retail mark-up (so after the bare component+manufacturing+distribution costs on Sega's end), a 2x speed CD drive along with CD-ROM data decoder (and associated components), and then the SH1 and 512 kB SDRAM, then $200 would be a reasonable estimate. That's also taking Sega's component procurement and manufacturing status at the time. (a company like Atari or Commodore would be a good deal worse off given their positions at the time)

    That said, Sega could have cut costs in other areas of the Saturn had it been designed a bit differently, and gotten to the point where it might have made a real mainstream price by 1995. Honestly, that would make more sense than releasing a new cart based system (especially with some provisions for expansion on the CD system).
    However, my premise here is for what would make the most sense given their existing 1994 situation, not how Sega could have designed a "better" 5th gen console in general. (OK, there's some areas they could have cut back in the Saturn even that late in the design, but I really can't see Sega of Japan management being forthcoming with that)


    32X games should have indirectly expanded the Saturn library as well by means of more up ports like Sonic 3D Blast and Stellar Assault.
    The same argument could be made that similar games being developed as Saturn (or Jupiter) titles in the first place would have been better all-around.

    That, and the Genesis/CD could be conducive to a few transitional multi-platform games too. (which is exactly what Sonic 3D Blast did) TBH, remaking higher profile late-gen MD/CD games into upgraded Saturn/Jupiter counterparts wouldn't be much different from doing the same for 32x games. (ie both would be programmed rather differently . . . except with the MD/CD stuff you'd be talking mostly about 2D and some limited 3D/scaling stuff -Batman and Robin would have been awesome on the Saturn)

    Also, I'm detecting an assumption that the Saturn needed more games in 1995
    More and higher quality is always good. The crappy early launch ruined that for the US and ruined hype too. And yes, the existing library by that fall was perfectly adequate for a good launch, but the added quality (if not quantity) of games facilitated by western developers having access to the platform months earlier could have been huge.

    In fact, it's more important to point out that (aside from marketing/PR alone), there were genuine holes in the Saturn's lineup for all of 1995 . . . or beyond that even. Sonic and Sega Sports both come up here, and honestly, I think the latter is the bigger issue (even more so after reading Melf's recent article on it). Sega had an awesome position in the Sports game market in 1994, but it all started falling apart with weak support on the 32x and subsequent weak support for much of the Saturn's life. (especially for the all-important football games)
    With EA screwing up with a 32-bit Madden release in 1995, Sega could have cut in as the main competitor to Game Day with a Sega Sports NFL title on the Saturn.

    As for Sonic . . . having a spinoff in 1995 would have been nice (Knuckles Chaotix on the Saturn could have been great), or even an upgraded Sonic 3 and Knuckles or Sonic CD. (the latter more due to fewer users having access to it, but the former more due to it being a truly awesome game - and I'm talking actual technical graphics/sound upgrades here, not compilation a la Sonic Jam . . . S3&K already used a mix of CG rendered art in there too, so that combined with nice hand-drawn pixel art in the Saturn's graphics could have been really cool).
    It's more than Sonic though too . . . Sega had a wealth of IPs they'd established (some more notable than other) and still a few more notable ones they were just introducing. So building on all of that would have made tons of sense. (selectively continuing the more notable/established franchises -arcade and home- along with the interesting new ones)

    And a side note to 3D Blast: the isometric pseudo 3D style was interesting (and I like it more than most), but honestly, it probably would have been much smarter to just aim at a sidescroller (maybe with a few exceptions in select levels/areas -and special stages, of course). A sidescrolling game with high quality rendered CGI "eye candy" would have been the main point. (more so on Jupiter/Saturn than Genesis, since the color/2D graphics capabilities could be shown off) Honestly, Sonic's existing 2D style fits rather well with the CGI look of the time anyway. (and Sonic 3 was already using some of that in moderation)
    Not sure whether a Sonic 3&K remake or an all-new (or spinoff) 2D sonic game would have made the most sense. (maybe both -a remake for '95 and the new one for '96 -possibly in parallel with an actual 3D sonic title)

    The US "pre-launch" has one definitive thing going for it, for some silly reason the first console to launch of a new generation generally does quite well. I think the limited pre-launch has been journalistically blown way out of proportion. I also find it ironic that the NES' 1985 test market is considered the national launch by most journalistic history while the Saturn seeing a limited early release by mere months is somehow the biggest blunder ever. Pettus be praised for propagating such an idea.
    No. The 1985 NES test market was seen as a failure by the market critics and competition of the time. It wasn't until the expanded test market in early 1986 (and especially the cheaper bundle and better software arriving -including Super Mario Bros.) that Nintendo actually made a splash on the market. (at the Winter CES of 1986 they were still nobodies, but by the June Summer CES they were already being considered as a major competitor alongside Atari and Sega) At the time all 3 companies (Sega, Atari, Nintendo) had their nationwide launches in mid/late 1986 they were considered to be on even footing, and not until after the 1986 Christmas season was Nintendo notable as the market leader (followed by Atari and then Sega). By mid 1987, Nintendo had a huge lead in 1st place followed by Atari, with Sega in a distant 3rd place. (off hand, I think it was around 60-70% for Nintendo, 20-30% for Atari, and ~10% for Sega -and that's market share based on current sales, not install base -at least, it was 70/18/10 in early 1988 and 77.4/13.3/4.1 by mid November)

    Nintendo's big launch was focused clearly at fall of 1986, leading into the holiday season, just as ever successful North American launch has gone.

    As to the first console to launch being the most successful . . . this is sometimes true, but (again) all those cases are for consoles with a properly timed and planned launch with good marketing on top of that. This also means that most such consoles will have a 1 year lead over any competition in the same generation/class. (unless we're talking a few weeks difference within the same year -like August vs November)

    Honestly, there's not a consistent trend for the first contender of each generation being the most successful. In fact, that's rarely the case.
    Now, OTOH, if you go be the first to launch a well-designed, well-planned, well-priced, well-supported, well-managed, and well-marketed next gen platform, then yes, that happens relatively consistently . . . and always in summer/fall.
    In fact, the only exception to that rule would be the Dreamcast, since it fits all of those categories (at least in the US market), but failed to maintain that in the long run for a large number of other reasons.

    Hell, and even with the examples that fit that "rule" there's tons of arguments against it too. Genesis didn't definitively lead over the SNES . . . and in that case, it needed the 2 year advantage just to break into the market and establish Sega definitively in North America. With the Xbox 360 vs PS3, you've got Sony making some massive errors with the PS3 design (especially cost-wise) along with odd marketing for a while. (otherwise, the 360's horrible failure rate combined with Sony's strong market position easily could have compromised that -and, as it is, they've managed to catch up pretty well too)
    And that's if you ignore the Wii as being in the same market sector . . . which it is and isn't depending on the context. (there's a lot of overlap, but the Wii really fills a different market niche than the PS3 and 360 -there could have been even more overlap had Nintendo not been quite so skimpy on hardware, but that's not the case -I mean, they could have had roughly 360/2005 gaming PC level graphics but just limited to 480p -so roughly 1/3 the system performance, too bad the Wii is more like 1/8th the performance)

    In regard to the Jupiter and Saturn scenario, I am honestly not sure if I would have received the Jupiter in May of 1995 any better than I did the Saturn (or PS1 for that matter). The games just didn't appeal to me at that point, but I don't count. Why I was willing to jump on the 32X in November 1994 for $160 had as much to do with enhancing my Genesis as it did with Virtua Racing Deluxe, After Burner and Space Harrier. An all new console at that point probably wouldn't have appealed to me as much.
    You're not representative of the average user at all though. After Burner and Space Harrier are not good mass market attractions (not bad as added spice to the library though) Virtua Racing Deluxe arguably would be . . . but Star Wars Arcade and Doom was probably the real killer apps due to the recognizable IPs. (and Star Wars Arcade alone in terms of exclusives)

    Chilly Willy would hardly have been the average Sega user of the day either, but as far as personal accounts go, he already mentioned liking the premise of the Jupiter compared to 32x in this context. (initially back when I posed it in your alternate realities thread)

    As to the typical Sega gamer of the day (talking existing mainstream 1994 users), we'd be talking about a mix of average kid/teen users (with a scattering of adults -and families as a whole) with a varying set of interests addressed by many of Sega's (and major 3rd parties') major releases. (supplemented by a scattering of more obscure titles)

    However, also remember that a massive chunk of the Genesis's mainstream userbase was Sports gamers . . . both hardcore sports gamers (playing little else) and more general users with a strong inclination towards Sports. The Genesis was THE sports platform to have at the time, be it Sega's own publications or EA's games (the best versions of which were almost always on the Genesis/MD), it was the obvious choice. (hardcore sports gamers also tend to comprise a good chunk of the high-end users eager to buy the latest games and new systems if it means getting their hands on the best, newest sports titles out there -even glorified roster updates )

    This made a pretty sparse showing on the Sega CD (probably a big missed opportunity, and certainly one of the reasons it didn't sell better than it did -relative to Genesis), but that became a much bigger issue with the 32x and Saturn with the Genesis falling out of the forefront of the market and the new systems not meeting the challenge. (granted, this is more a software argument than hardware, but in that sense, the Jupiter at least could have offered a more fluid and focused transition for sports game development, and could offer much more than the 32x in terms of enhancing gameplay over MD counterparts)

    I found the Jaguar's $250 price point daunting as well considering the games I saw.
    Well, that's a terrible comparison for obvious, software support related reasons.
    (and slightly less obvious reasons in that the hardware is cheaper/simpler to manufacture than the Jupiter too . . . its just that Atari wasn't in a position to negotiate for components or push high volume production like Sega could -Sega probably could have pushed the Jaguar below $200 for late 1994)

    I think you are overlooking some of the argument for 32X to say that there is no real contest. Jupiter would be a full blown next generation console in 1994 dollars and the SVP gave us this:
    The point of the SVP scenario is for Sega wanting an expansion for the MD rather than a full-blown new platform (which the 32x was, regardless of form factor). If they wanted said expansion, the SVP makes far more sense due to the cost/simplicity.
    If however, they DID want a full new platform as a transitional next-gen console to complement the Saturn, then the Jupiter still makes far more sense than the SVP or 32x.


    and it cost half to two thirds of what gave us this:
    In my premise, SVP (if released standalone) would be less than 1/3 the price of the 32x. Or, in the context of your Virtua Racing comparison, just about 1/2 the price ($100 vs $200), if not slightly more in the SVP's favor. (also remember VR Deluxe was a 3 MB game vs 2 MB for SVP)

    Star Wars arcade was the most significant exclusive for the 32x at the time, and that game should have fared much closer to the 32x on the SVP given the color scheme of that game.
    Doom would be tougher though . . . probably needing a custom engine and/or done on the Sega CD instead. (or SVP-CD)

    I would say that the SVP has none of the programming similarities with Saturn and cost is its only advantage.
    Cost is a huge advantage in this case, along with all the associated benefits. (R&D cost, monetary investment/risk, marketability, etc) There's also the simplicity of installation and form factor compared with 32x. Then there's the earlier release (assuming they kept the SVP launch) and the far less direct conflict with the Saturn than the 32x did. (as far as marketing, market position, and PR)
    If the SVP failed to take off, it wouldn't be nearly as big a loss than the 32x. (PR, money, etc) OTOH, if it did show serious promise on the mass market, it could have been feasible to integrate the SVP as a standard feature of late model MDs without a major hit to manufacturing costs. (no way the Neptune could manage that, but tacking on that ASIC and 128kB DRAM should have worked out quite well)

    And TBH, I don't think the 32x would help programmer transition to Saturn much more than SVP, for the drastically different programming techniques already addressed. (the remaining advantage being game design, and among other things, that would be addressed largely by the SVP as well -especially on top of what the Sega CD already offered)

    I thought I was being generous, as SVP Virtua Racing's lower resolution makes dithered pixels even larger and visible even from Composite. I think posterized colors would be the best way to go unless they could get the resoluton to 320 wide, which I doubt.
    Super FX games resorted to dithering for the same reason. (same for Silpheed and the many other Sega CD games running in H32 -including most FMV)
    Using 15 or 16 colors alone is just too limited and ugly.

    That said, running in H40 should have been a very viable option. You get a good DMA bandwidth boost in that mode, but the disadvantage is that you'd be dealing with smaller screen sizes at similar resolutions. For games not needing a (semi)detailed 2D tile BG, you could go a fair bit beyond what SVP Virtua Racing does though. (otherwise you'd run out of VRAM or you could resort to single buffering and screen tearing)



    Exactly, as I mentioned above, we're talking limited framerate, lower resolution 15-color games with minimal sound enhancements. I am barring a fluke sound engine like Virtua Racing Deluxe's being used when such a thing didn't happen on Genesis titles for some reason. I don't think even Virtua Fighter 2 on Genesis, by the same team, sounds as good as Virtua Racing Deluxe sample wise. It's a real mystery.
    Virtua Racing deluxe is not a fluke. Most of the 32x games could have sounded just about the same on the Genesis alone. (or, potentially better depending on the circumstances)

    Several of the better games don't use PWM at all, and most others wouldn't be worse off without it. (in fact, I'd say Chaotix would be the only exception, and extremely slim even then -having 1 added FM channel) Be it genesis sound alone, or SVP mixing (played back by the Z80), there's no example of sound on the commercial 32x games that couldn't have been done similarly on the Genesis (in many cases it could have been done better on the MD/SVP).
    Granted, games making extensive use of the SVP for audio wouldn't be making much use of it for other things. (most examples of sound 32x could have been done without the SVP anyway . . . and those that do go beyond normal MD/Z80 capabilities still fall far short of what the SVP could have done if pushed for sound)


    I don't think the pre-launch was as absolutely horrible as people like to think. To consumers (the 50 million + that buy mainstream devices) the Saturn pre-launch was as obscure as the 32X and Sega CD were already. The one kid I knew who got one felt like he was let in to an early sneak peak, not the full console launch. Usenet posts from Summer of 1995 express this as well, and EGM and Gamepro issues from the time were very positive of the Saturn. The narrative of the horrible mistep launch came much later and originated from Edge/Next Generation not consumer impressions.
    That sneak peak still confused things substantially and started a price war with Sony earlier than necessary. Poor planning of it also screwed up relationships with retailers and 3rd party developers.

    It was botched, either as a pre-release or an actual launch. (TBH, they should have just announced pre-orders at that point . . . maybe started sending not-for-resale kiosk demo units for in-store displays)

    More than that though, I'd gotten the impression that Sega didn't follow up with a heavily promoted/hyped "full" launch that fall with massive marketing for the new platform moving into the holiday season. (there's still the big gap in the software lineup though, the big double SS -Sports and Sonic)

    You keep saying this and I keep wanting to think you mean "to you" but I'm starting to doubt that. The 32X made sense from a variety of perspectives I have already described.
    And I addressed many, many more areas where alternatives made more sense while retaining many of the desired attributes of the 32x.

    The biggest thing that makes it questionable is its cancelation after six months and resultant small library of early games. This is arguable, but I would say that the SNES and Genesis were beyond budget at that point, they were aging/old tech that only super late adopters were still buying and in a totally different market region than "budget console". Back in 1993 they might have filled the role of being the low end market, but after Fall of 1994 they weren't even talked about as the mainstay consoles anymore.
    In 1993 and 1994 the Genesis and SNES filled the roles of flagship mainstream consoles in North America and Europe (both for Sega and Nintendo, and from the consumer Standpoint). Like the NES in 1990, the PS1 in 1999, or the PS2 in 2005.
    For that matter, they were still the 2 mainstream NA/EU platforms of 1995 (especially from a software perspective), though by that point they were definitely fading into the budget market. 1996 marks the real shift to the new gen systems (the first mainstream year for the 5th generation), and the SNES and Genesis definitively entering the low-end budget market. (1996-1998 were big years for those systems on the budget market, and '96 still had a lot of new software on both 16-bit consoles)

    Software sales are far more important than hardware sales. After all, the only point of releasing console hardware is to achieve an install base for software profits.


    We're talking about the difference of buying a Retrogen today instead of a $99 360 or Wii (for whatever).
    No, we're pretty much talking about the PS2 in 2006 or PS1 in 2000. Or the NES in 1991. Not mainstream, but still huge platforms in the low-end niche with modest hardware sales and far more software sales of re-releases and a handful of newer releases. (which continued on the Genesis and SNES in considerably quantity through 1996 -and modestly in 1997)

    The 7th gen consoles are really terrible to compare since they've been on the market much longer with no next-gen competition. (the Wii is not sort of . . . except it fills a weird niche to begin with, was already dated when it launched and it's successor is in the same boat -actually worse off given the WiiU is weaker than the 360 in critical areas)

    Your first paragraph outlines exactly why the 32X was created. We also don't know what exactly was stripped out of the 32X, it might have been something very comparable to VDP1, more RAM or both, making it much more "Jupiter like" than the peripheral priced 32X add-on is.



    I think Sega saw CD-ROM as a completely different market than cartridge games, one that was poorly suited for quick pick up and play Action titles or straight Arcade ports. Furthermore, we are leaving out the likes of Spiderman Web of Fire, Blackthorne and even Knuckles Chaotix (especially the bosses and bonus rounds) from the mix. Besides that, I think without the pressing need for a launch lineup most of these games wouldn't exist or would have been folded into other projects.[/QUOTE]
    Last edited by kool kitty89; 02-10-2013 at 06:55 AM.
    6 days older than SEGA Genesis
    -------------
    Quote Originally Posted by evilevoix View Post
    Dude it’s the bios that marries the 16 bit and the 8 bit that makes it 24 bit. If SNK released their double speed bios revision SNK would have had the world’s first 48 bit machine, IDK how you keep ignoring this.
    Quote Originally Posted by evilevoix View Post
    the PCE, that system has no extra silicone for music, how many resources are used to make music and it has less sprites than the MD on screen at once but a larger sprite area?

  15. #45
    I remain nonsequitur Shining Hero sheath's Avatar
    Join Date
    Jul 2010
    Location
    Texas
    Age
    42
    Posts
    13,313
    Rep Power
    130

    Default

    Watching this just now, I still have to doubt that the emulators are correct that the 32X isn't doing anything for the sound in Deluxe, absolutely everything sounds better.

    "... If Sony reduced the price of the Playstation, Sega would have to follow suit in order to stay competitive, but Saturn's high manufacturing cost would then translate into huge losses for the company." p170 Revolutionaries at Sony.

    "We ... put Sega out of the hardware business ..." Peter Dille senior vice president of marketing at Sony Computer Entertainment

Thread Information

Users Browsing this Thread

There are currently 1 users browsing this thread. (0 members and 1 guests)

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •