Quantcast

Page 52 of 210 FirstFirst ... 24248495051525354555662102152 ... LastLast
Results 766 to 780 of 3143

Thread: PS2 vs Dreamcast Graphics

  1. #766
    Hero of Algol
    Join Date
    Aug 2010
    Posts
    7,790
    Rep Power
    177

    Default

    Quote Originally Posted by j_factor View Post
    Sorry but this doesn't add up. There is no way RS3 doubles RS2's poly counts. At least, it certainly doesn't do that by way of putting RS2 in split screen. RS2 ran at 640x480. RS3 still runs at 640x480. Even the exact same models and environments with not a single polygon removed will not give you the same number of polys/sec in a 640x240 window as it does in a 640x480 window. It was clearly truncated. Take a look again at this comparison shot from your link:



    The viewing area is way cropped in split screen. The alternatives are to squish the screen, which would look awful, or to zoom out the camera, which would make stuff harder to see. Comparing a frame of RS2 to a "frame" of half of RS3's split screen, the cropped out area pushes a lot of polygons out of frame. So even if they did say they made no cuts (neither of your links quotes anyone from Factor 5 as saying that), and even if they were not stretching the truth even a little, it's still not double the polygon count.
    Good points (except for the resolution part; the view size part makes all the sense though). Can't rep you, sorry.
    I still would like to see a proper video comparison or images with better quality to judge it with more details on hand.



    Quote Originally Posted by j_factor View Post
    No, it's clearly not "just good texture work on limited polygons". In fact it's the opposite -- simple texture work on lots of little polygons. Not that you need intricate textures on such small objects. You can see it more clearly when it's zoomed in (example) -- there's a base flat ground with a texture on it, and on top of that blades of grass are rendered as distinct objects. It looks like they created a couple of different models for a small group of blades of grass, and then just littered the ground with those models. As for the grass facing the same direction and rotating with the camera, a number of games did that with various things. Toejam & Earl III did it a lot, both in the unfinished Dreamcast version and in the Xbox version. That's one game where it's particularly noticeable, but a lot of games did it less obnoxiously. The grass disappearing is just a clipping issue as the angle of the ground changes. Notice also how the grass doesn't react to being stepped on; it looks like it has no collision detection at all.
    Yep, the movement is strange but it's not always facing you (I just think it's using a very simplified perspective calc) AFAIK, it depends on the camera angle.
    This segment shows that they seem to have a 3D view: http://youtu.be/YCVOQf-W1EA?t=18m54s

  2. #767
    Raging in the Streets A Black Falcon's Avatar
    Join Date
    Jun 2008
    Age
    37
    Posts
    3,238
    Rep Power
    40

    Default

    Quote Originally Posted by Barone View Post
    I like how you forgot to mention the floating point bar chart which is pretty much self-explanatory and quite relevant IMO.

    Context:
    I got a new computer back in early 2001, just a few months before I got a Gamecube. The GC's graphics were great in games like RS2, but overall I think it clearly wasn't as powerful as that P4 1500, even with only a GeForce2 GTS32MB for a graphics card. Of course that's a much faster CPU, however. At least the GC can also do good image quality, though, unlike the PS2 usually... though yeah, sure, of course the PS2 is also in that generation, power-wise.

    Monte Carlo simulations and FFT charts in that article are more about benchmark of classical algorithms than anything else IMO, FFT benchmarks are pretty common in FPGA articles.
    Monte Carlo-based algorithms can be used for lots of applications, including ray casting and AI (since we must find game-related applications to "validate" benchmarks like that, lol).
    Given that consoles exist to play games, I would think that game-related things would be the most valuable things for them...


    The EA's article which compares PS2/GC/Xbox/PC was pretty clear, unless you can read a table (well, yeah, maybe not...).
    But 3D rendering scenarios aren't in the context of this debate, right?
    That article may have been clear, but comparing its GC numbers to Factor 5's makes it also clear that they weren't as good at programming for the system as F5 was...

    Agreed. I'd still like to see a side-by-side comparison of RS2 and the RS3's split-screen version though.
    Thanks for the links.
    Doesn't one of the IGN articles kind of do that? More would be nice, sure, but they have some. It's better than Youtube videos or something for sure anyway, given the usual Youtube video quality/framerate issues...

    Yet 18 million polygons per seconds are in the realm of the GPC's performance if MH is also not lying. Remember that such game is heavy in both physics and AI.
    It'd be interesting to see if that that game does indeed push that many polys, since I haven't heard anyone mention any other PS2 games that manage to do so well polycount-wise... but who knows, it could have. They did do probably the highest-poly DC game. But is there any analysis of the game anywhere? I remember there have been some about Test Drive Le Mans, looking at how many polys it's probably actually pushing.

    20 million is a bit more than 10% of advantage,
    10% (or something) is a definite edge, but I don't know what the actual number would be... but while the polygon-count thing is an advantage, the bigger one, as I've said all along, is the image quality gap problem the PS2 has.

    so, do you consider 10% such a huge gap that no other games could have matched or surpassed it on the PS2 later on?
    Should we consider the PS2 insanely underpowered due to a 10% disadvantage?
    If there were other games with poly counts in that level, I'd think they'd have been mentioned somewhere sometime...

    That level of performance can only be achieve in split-screen mode on Rebel Strike?
    No, I just haven't seen articles specifically saying how many polygons F5 used in the new campaign in Rebel Strike. It uses more polys than Rogue Leader did, for sure, but how many more exactly? Unless there's more out there somewhere, with a number attached to them for the main campaign. How many more polys, exactly, is the single player campaign in RS3 pushing, compared to RS2? Yeah, can't say for certain. But even just video comparison shows that RS3 is doing more than RS2, there's more stuff going on in Rebel Strike. And if they could do ~18 million polys, I'd think they'd use them.

  3. #768
    Raging in the Streets A Black Falcon's Avatar
    Join Date
    Jun 2008
    Age
    37
    Posts
    3,238
    Rep Power
    40

    Default

    I should say one bad thing about RS3, graphically. I like the game and think it's much better than it's generally thought to be -- the mediocre on-foot sections are under 20% of the game and I found them just bland, not really awful -- but it was pretty disappointing that the versus mode is two player only. F5 came up with several pretty interesting versus modes, but it's hamstrung by being two player only. The game would be so much better with the four player splitscreen it should have had! Sure, they probably couldn't get 3/4 player modes working at a good enough framerate or something, but they really needed to come up with something. the game could have been the GC's answer to Crimson Skies on the Xbox (by the way, that game is great), but because unlike Crimson Skies RS3 is two player only, it isn't that. Too bad. The two player co-op campaign is great because co-op splitscreen campaigns almost never support more than 2 players anyway, but the versus modes should have supported more.

    Quote Originally Posted by j_factor View Post
    Sorry but this doesn't add up. There is no way RS3 doubles RS2's poly counts. At least, it certainly doesn't do that by way of putting RS2 in split screen. RS2 ran at 640x480. RS3 still runs at 640x480. Even the exact same models and environments with not a single polygon removed will not give you the same number of polys/sec in a 640x240 window as it does in a 640x480 window. It was clearly truncated. Take a look again at this comparison shot from your link:



    The viewing area is way cropped in split screen. The alternatives are to squish the screen, which would look awful, or to zoom out the camera, which would make stuff harder to see. Comparing a frame of RS2 to a "frame" of half of RS3's split screen, the cropped out area pushes a lot of polygons out of frame.
    That's true, but with two player windows to render, unless you cut stuff you'd still need to be pushing a lot more stuff than a game would in single player. That's why in the 5th gen particularly so many games made graphical cuts in splitscreen mode, smaller viewing areas or no the games couldn't handle it at full detail. Even with some of the top and bottom cut off, you'd still need to be able to render almost double the polygons because often the highest-detail stuff is going to be in the center of the screen, not at the edges...

    So even if they did say they made no cuts (neither of your links quotes anyone from Factor 5 as saying that), and even if they were not stretching the truth even a little, it's still not double the polygon count.
    Uh, what? From the article: "As it turned out, though, when the company figured out how to double its polygon counts, a discovery made through practically rewriting the original engine completely, it also realized it could handle the new lighting model." "When they figured out how to double its polygon counts" means exactly what it sounds like, Factor 5 says they doubled RS2's polygon count in RS3. And when RS2 released they said that game was pushing 9 or 10 million polygons.

    No, it's clearly not "just good texture work on limited polygons". In fact it's the opposite -- simple texture work on lots of little polygons. Not that you need intricate textures on such small objects. You can see it more clearly when it's zoomed in (example) -- there's a base flat ground with a texture on it, and on top of that blades of grass are rendered as distinct objects. It looks like they created a couple of different models for a small group of blades of grass, and then just littered the ground with those models. As for the grass facing the same direction and rotating with the camera, a number of games did that with various things. Toejam & Earl III did it a lot, both in the unfinished Dreamcast version and in the Xbox version. That's one game where it's particularly noticeable, but a lot of games did it less obnoxiously. The grass disappearing is just a clipping issue as the angle of the ground changes. Notice also how the grass doesn't react to being stepped on; it looks like it has no collision detection at all.
    That screenshot, it's so pixelated and ugly... I hope the game looks better on actual hardware than it does there...

  4. #769
    I DON'T LIKE POKEMON Hero of Algol j_factor's Avatar
    Join Date
    Jun 2005
    Location
    NorCal
    Posts
    9,330
    Rep Power
    129

    Default

    Quote Originally Posted by A Black Falcon View Post
    That's true, but with two player windows to render, unless you cut stuff you'd still need to be pushing a lot more stuff than a game would in single player. That's why in the 5th gen particularly so many games made graphical cuts in splitscreen mode, smaller viewing areas or no the games couldn't handle it at full detail. Even with some of the top and bottom cut off, you'd still need to be able to render almost double the polygons because often the highest-detail stuff is going to be in the center of the screen, not at the edges...
    It's certainly more, but it's not double, and I doubt it's even close to double. Hard to actually put a number on it, though.

    Uh, what? From the article: "As it turned out, though, when the company figured out how to double its polygon counts, a discovery made through practically rewriting the original engine completely, it also realized it could handle the new lighting model." "When they figured out how to double its polygon counts" means exactly what it sounds like, Factor 5 says they doubled RS2's polygon count in RS3. And when RS2 released they said that game was pushing 9 or 10 million polygons.
    Factor 5 didn't say that. Those are the words of the IGN writer, not Factor 5. They didn't quote anyone from Factor 5.

    That screenshot, it's so pixelated and ugly... I hope the game looks better on actual hardware than it does there...
    That was just a quick screengrab I did from the youtube video in fullscreen mode on my PC. A blown-up youtube video is always going to look worse than the real thing, though, it probably doesn't look that much better on real hardware, in terms of what you're talking about. But, keep in mind that that image is from the player zooming way in, not typical gameplay, and in any game that's going to look worse than the overall game.


    You just can't handle my jawusumness responces.

  5. #770
    ESWAT Veteran Team Andromeda's Avatar
    Join Date
    Jul 2010
    Location
    Wales, UK
    Posts
    5,684
    Rep Power
    61

    Default

    Well here's what some people who actually made games or development tools for the all the systems have to say

    Froniter developments ....

    While the Gamecube doen't have the raw fillrate of the PS2 and is slower at clipping polygons on screen, it does support multitexturning which gives it a signifacnt advantage
    PS2 has more polygon pushing power (than Cube) but Cube has better catching
    Cube is signifcantly faster than the PS2 but the smaller main memory cassues real hassles

    SN Systems ...

    If you're doing some crazy procedural rendering special effects with lots of crazy maths, then there's little to touch the PS2. For straightforward 3D graphics though, then the Cube can certainly match and sometimes exceed the capabilities of other consoles



    --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------







    Anyway, the 9/10 million polygons in RS2 is known and proven. As is that Factor 5 said that they doubled performance in RS3
    That would be at odd with NCL own technical facts and how they make out the Cube is able to handle 15 lit and texture polygons

    Those polygon counts are for the in-game models. And never mind 9 million, the GC did twice that. Resident Evil 4 is another game which pushes a lot of polygons, too
    Jak & Daxter is said to be pushing 10 million polygons isn't it ?. All I will say is for all this talk nothing beats the visuals the likes of Orta, Oddworld, Ninja Gaiden, DOA Ulimate, O.TO.GI offered on the XBox . That said I never see anything on the XBox or the Cube handle the amount of on screen action to these games or the secne in one of rhe hitman games where's there's hundreds and hundreds of characters on screen



    Panzer Dragoon Zwei is
    one of the best 3D shooting games available
    Presented for your pleasure

  6. #771
    It's that time again! Road Rasher HalfBit's Avatar
    Join Date
    Jan 2014
    Location
    Canada, Eh
    Posts
    316
    Rep Power
    8

    Default

    It'd be interesting to see if that that game does indeed push that many polys, since I haven't heard anyone mention any other PS2 games that manage to do so well polycount-wise... but who knows, it could have. They did do probably the highest-poly DC game. But is there any analysis of the game anywhere? I remember there have been some about Test Drive Le Mans, looking at how many polys it's probably actually pushing.

    I'm pretty sure Le Mans was 3M. I remember hearing somewhere that shenmue 2 was 5M though.

  7. #772
    Road Rasher
    Join Date
    Feb 2013
    Posts
    423
    Rep Power
    16

    Default

    Quote Originally Posted by Barone View Post
    I like how you forgot to mention the floating point bar chart which is pretty much self-explanatory and quite relevant IMO.
    *glances at bar chart.

    Why are you using charts that show it takes longer for the EE and VUs to execute data compared to a P4 as evidence to the contrary?

    EDIT: ah, they were referring to the XBox CPU with the 733Mhz comment. But that begs the question: how good was that 733MHz CPU against a P4 itself?
    Last edited by spiffyone; 02-11-2014 at 02:09 PM.

  8. #773
    Raging in the Streets A Black Falcon's Avatar
    Join Date
    Jun 2008
    Age
    37
    Posts
    3,238
    Rep Power
    40

    Default

    Quote Originally Posted by Team Andromeda View Post
    That would be at odd with NCL own technical facts and how they make out the Cube is able to handle 15 lit and texture polygons
    You seem to have forgotten that I've already explained this. As I said earlier, Nintendo was very conservative in its GC polygon-count estimates, to the point where they actually gave an estimate that the system could surpass. It's basically the opposite of the PS2 or Xbox, where Sony and MS gave estimates so absurdly high that their systems could never even begin to hope to match the numbers the companies claimed. I don't know which approach, a too high or a too low estimate, is the better one, but those are the things they did.

    Quote Originally Posted by j_factor View Post
    It's certainly more, but it's not double, and I doubt it's even close to double. Hard to actually put a number on it, though.
    I would think it would be very close to double, at max. As Isaid, in a game often you'll have large objects in the center, not at the top and bottom.

    Factor 5 didn't say that. Those are the words of the IGN writer, not Factor 5. They didn't quote anyone from Factor 5.
    Oh come on, it's very clear that the IGN writer said that based on what Factor 5 told him. There's no question about that.

    That was just a quick screengrab I did from the youtube video in fullscreen mode on my PC. A blown-up youtube video is always going to look worse than the real thing, though, it probably doesn't look that much better on real hardware, in terms of what you're talking about. But, keep in mind that that image is from the player zooming way in, not typical gameplay, and in any game that's going to look worse than the overall game.
    True, but it did result in a typically PS2-ugly screenshot.

    Quote Originally Posted by Barone View Post
    That must be why I've being reading it for years and actually linked it in a reply to rusty many posts ago:


    ABF fails. Again.
    No, you fail, if you actually do know what the site is and still said something so incredibly stupid. Trolls like you should be ignored.

    Learning? From you? What? How to be a blind Nintendo fanboy? No, thanks.
    I'm not a blind Nintendo fanboy (how could I be a blind Nintendo fanboy when I love lots of non-Nintendo platforms too (just not Sony ones!), and think that the PC is better than any console? Oh come on!), but you clearly are a blind Sony fanboy...

  9. #774
    Hero of Algol
    Join Date
    Aug 2010
    Posts
    7,790
    Rep Power
    177

    Default

    Quote Originally Posted by A Black Falcon View Post
    No, you fail, if you actually do know what the site is and still said something so incredibly stupid.
    It is "incredibly stupid" to assume that all guys at beyond3d are some sort of authorities. There's a good share of developers and ex-developers there; but, as in all forums, there's also guys like you.

    However, it would be fun to see you dropping your absurd shit there, like Virtua Fighter being NOT a 3D game or how the N64 draws accurate polygons.


    Quote Originally Posted by A Black Falcon View Post
    Trolls like you should be ignored.
    Is anyone forcing you to read and reply to my posts?

  10. #775
    It's that time again! Road Rasher HalfBit's Avatar
    Join Date
    Jan 2014
    Location
    Canada, Eh
    Posts
    316
    Rep Power
    8

    Default

    Quote Originally Posted by Barone View Post
    It is "incredibly stupid" to assume that all guys at beyond3d are some sort of authorities. There's a good share of developers and ex-developers there; but, as in all forums, there's also guys like you.

    Reply: don't forget there's also guys like you. They just contradict everything and be negative.

    However, it would be fun to see you dropping your absurd shit there, like Virtua Fighter being NOT a 3D game or how the N64 draws accurate polygons.

    Reply: The N64 is a fine console, but compared to the dreamcast, it's graphics look dated and blocky.
    And virtua fighter is 3d. Duh. Only a complete idiot, which ABF is not, would think otherwise.



    Is anyone forcing you to read and reply to my posts?
    No, but he's making good points that should be heard and your contradicting him.
    So obviously he's defending himself.
    His gamecube poly claims seem reasonable and if you can give me one real reason to doubt them, I will be impressed. Obviously he's taken the time to find links and sources to back up his claims, so I'd say that they're probably legitimate, realistic claims. Now stop mudslinging.
    Last edited by HalfBit; 02-11-2014 at 07:45 PM.

  11. #776
    Hero of Algol TrekkiesUnite118's Avatar
    Join Date
    May 2010
    Age
    32
    Posts
    8,188
    Rep Power
    128

    Default

    Quote Originally Posted by Barone View Post
    I like how you forgot to mention the floating point bar chart which is pretty much self-explanatory and quite relevant IMO.
    You mean this one?:



    In this one the PS2 is still getting it's butt kicked by the Pentium 4. Look at the scale for time on the side. We're still in 100's of seconds for the scale and it doesn't go up linearly. So those small gaps between the PS2 and the Pentium 4 are actually very significant.

    For the Monecarlo simulation both P4's pull it off in under 100 seconds. The PS2 CPU takes over an hour to do it and the vector Units take about 5 minutes to do it. While that's a great improvement for the Vector Units, it's still 5 times slower than a Pentium 4.

    For the floating point the regular Pentium 4 takes about 1 Minute and 40 seconds to do it, the Pentium 4 with SSE takes about 4-5 seconds to do it. The PS2 CPU on the other hand takes about 5 minutes to do it, and the Vector Units take about 3 minutes to do it. So the CPU is at best 3x slower than a Pentium 4 and at worst is 6000x slower. The Vector Units at best are about 2x slower and at worst are about 4000x slower.

    The point is this doesn't really tell us a whole lot of anything about the PS2, only that it's weaker than a Pentium 4 by a rather large margin at it's worst. It doesn't tell us how the Gamecube or Xbox CPU compare in this situation.

    Quote Originally Posted by Barone View Post
    Context:
    That's referring to a Pentium 3. Which again isn't on that chart. I'm not saying a Pentium 3 is going to be better than a Pentium 4 (though it could be in some situations, the P3 core was more effiecient than the P4 core, hence why it was used as the base for the Core 2 Duo line if I remember correctly), but with how wide the scale is for performance on those charts, the Pentium 3 could very well end up clocking in between the PS2 and the Pentium 4's. And again, there's no Gamecube or Xbox CPU on these charts to give us any frame of reference for how those compare.



    Quote Originally Posted by Barone View Post
    Says the guy who wants to teach others how to code but can't even read a chart properly.
    How is the chart being misread then? Please enlighten us so we can understand it with you.



    Quote Originally Posted by Barone View Post
    Monte Carlo simulations and FFT charts in that article are more about benchmark of classical algorithms than anything else IMO, FFT benchmarks are pretty common in FPGA articles.
    Monte Carlo-based algorithms can be used for lots of applications, including ray casting and AI (since we must find game-related applications to "validate" benchmarks like that, lol).

    FFT is used a lot for image processing but I doubt the edge detection algorithm that rusty was referring to was using this sort of classical implementation.
    However, the "FFT" part in that article is represented just by an implementation of a radix-2 butterfly; which usually is just one addition, one subtraction and one multiply (all modulo) taking two inputs and giving two outputs.
    I suppose that any 3D game use things like addition, subtraction and multiply; right?

    Last, but not least, weighted sums are surely very used in any 3D game, since the 3D location of the vertex is usually a weighted sum. No rag-doll physics without good performance when computing weighted sums
    Thanks for explaining that, that gives us some idea for how these numbers can be applied to games.

    Quote Originally Posted by Barone View Post
    The EA's article which compares PS2/GC/Xbox/PC was pretty clear, unless you can read a table (well, yeah, maybe not...).
    But 3D rendering scenarios aren't in the context of this debate, right?

    The experiments using dot products, which compared the PS2 vector units to a PIII 600 MHz were pretty conclusive and easy to understand.
    Was that the one that showed us the GFLOPs of the CPUs? Yeah the CPU in the GC is rated lower than the combined GFLOPs of the Emotion Engine and the Vector Units in the PS2. It comes in at 1.9 GFLOPs. However ATi Flipper is rated at 8 GFLOPs. So combined the Gamecubes peak GFLOPs performance is close to 10 GFLOPs. This goes a bit hand in hand with what's stated in a lot of the IBM interviews about Gekko. The PS2's architecture is a very different design, so some specs don't compare very well. On PS2 the CPU and Vector units have to do your Floating Point operations, as the Graphics Synthesizer doesn't give you any floating point performance from what I can tell looking at the specs. Gekko on GC isn't as beefy in that sense, but that's partly because it has a GPU that can handle a lot of that for it. Which you seem to be ignoring here.

    And for what it's worth, according to Dolphin the LucasArts logo animation for Rogue Squadron 2 is rendering between 80k-120k polygons per frame. That is until the camera comes over top and you don't see all the individual storm troopers anymore where it drops to 12-18k.

    Though the game was really buggy for me so I'd take that with a grain of salt. When I tried to start up a level it froze and I got a bunch of garbage on the screen. Is anyone here able to get the game to actually run in Dolphin to get some solid polygon numbers? If the numbers I got were correct though, that sequence alone is rendering 6 million polygons per second, which holy god that's a lot for just that.

  12. #777
    Raging in the Streets A Black Falcon's Avatar
    Join Date
    Jun 2008
    Age
    37
    Posts
    3,238
    Rep Power
    40

    Default

    Quote Originally Posted by Barone View Post
    It is "incredibly stupid" to assume that all guys at beyond3d are some sort of authorities. There's a good share of developers and ex-developers there; but, as in all forums, there's also guys like you.
    I don't know about you though, people who troll as much as you do get banned at some forums...

    However, it would be fun to see you dropping your absurd shit there, like Virtua Fighter being NOT a 3D game
    Can we please not argue about this AGAIN? Seriously...

    Virtua Fighter is a 3d game in that it's got polygonal graphics, so it's 3d in that sense of course. As for the gameplay I don't really want to argue about this AGAIN, but sorry, I cannot consider a game to have fully 3d gameplay when you can't move around into and out of the screen! Sure, some attacks can shift the plane, so VF isn't strictly 2.5d like Street Fighter IV is, but it's also a very different, and very restrictive, kind of thing compared to later, better 3d fighting games which do give you full movement control. Sometimes Virtua Fighter is a 2.5d fighting game -- when you're out of attack range from the opponent, you cannot move into or out of the screen, period, and can only move along the set plane. This is not good design for a supposedly 3d game -- the gameplay isn't 3d at a distance, which is pretty awful sometimes. Sheath and you talked over and over about how the close-range stuff makes it 3d, but you always skip over this part... in a real fight, unless you're in a fencing match (since fencing uses a long and very narrow space to fight in) people can try to circle around eachother when out of range! The argument that circling around doesn't happen in real fights, or something, doesn't hold up at all. In a real fight, if you're stuck in the corner, you can try to circle around to the side to get out of the corner. In Virtua Fighter, you usually would have to rely on some attack move to hopefully shift the plane out of the corner. Forward or back won't always reliably get you where you want to go. See the difference? Also of course videogames are not real life and usually should not reflect real life. Fighting, racing, flight, whatever, in my opinion often the more fun games are the less realistic ones.

    But yes, when you are in range VF does become a 3d game. I just really dislike that you can't move around with the d-pad but instead are expected to memorize which moves shift things. Sorry, the game is too boring to make me want to do that. Most 5th gen polygonal (3d, 2.5d, whatever) fighting games aren't all that great, though... there's a reason why my favorite 5th gen 3d fighting game is Evil Zone (yes, a PS1 game), and not a more traditional title. But yes, at close range this style of game has 3d gameplay. But I would still add the asterisk there that you can't move around in 3d with the d-pad, a limitation which I really dislike. And it's only really 3d within attack range. So yeah, regardless of what you call that style, I don't like it from a design standpoint.

    (Note: I don't dislike all of the VF-styled games; Fighting Vipers and Dead or Alive 1 are decent fun, despite their VF-style control. They're faster paced and more fun than VF is.)
    or how the N64 draws accurate polygons.
    That the N64 was the first console which could draw polygons that don't have perspective problems, aka accurate polygons, is a very obvious fact that any sane person knows. If you define "accurate" some other way that's your problem, not mine. You know that's what I mean, and it was a massively important advance which made 3d gaming work much better than it did on systems with perspective problems, like all pre-N64 consoles have.


    Oh, on the note of pixelized textures, much earlier in this thread you falsely claimed that I have no problem with blocky Mode 7 textures, or something (I think that was it? Or was it Super FX? Either way, same answer.), which criticizing PS1/Saturn blockyness. That's absurd! SNES Mode 7 and Super FX stuff is indeed horrendously blocky and often pretty ugly looking, no question.

    Is anyone forcing you to read and reply to my posts?
    You keep posting this stuff, so of course I have to respond.

    Quote Originally Posted by HalfBit View Post
    No, but he's making good points that should be heard and your contradicting him.
    So obviously he's defending himself.
    His gamecube poly claims seem reasonable and if you can give me one real reason to doubt them, I will be impressed. Obviously he's taken the time to find links and sources to back up his claims, so I'd say that they're probably legitimate, realistic claims. Now stop mudslinging.
    He's more interested in insulting me than in doing that, most of the time, unfortunately.
    Last edited by A Black Falcon; 02-12-2014 at 12:14 AM.

  13. #778

    Join Date
    Jan 2014
    Location
    Frankfurt, Germany
    Posts
    181
    Rep Power
    0

    Default

    Quote Originally Posted by TrekkiesUnite118 View Post
    Was that the one that showed us the GFLOPs of the CPUs? Yeah the CPU in the GC is rated lower than the combined GFLOPs of the Emotion Engine and the Vector Units in the PS2.
    Correction: the Gecko is rated lower than the EE FPU + a single vector unit (VU0) in terms for FLOPS. Not just the EE FPU + both vector units. The Gecko is 1.9GFLOPs and the EE FPU + VU0 comes in at 3.8 FLOPS. There's also one important thing to note; you could run VU0 in parallel with the main core, so that the cost of a lot of general purpose logic in code that used the results generated by VU0 could be "hidden". On the Gecko, you didn't have that luxury. It was strictly linear.

    Also, you could pass data from VU0 to VU1 directly. So you could off-load some rendering calculations onto the CPU without going over the main memory bus.

    Quote Originally Posted by TrekkiesUnite118 View Post
    It comes in at 1.9 GFLOPs. However ATi Flipper is rated at 8 GFLOPs. So combined the Gamecubes peak GFLOPs performance is close to 10 GFLOPs. This goes a bit hand in hand with what's stated in a lot of the IBM interviews about Gekko. The PS2's architecture is a very different design, so some specs don't compare very well. On PS2 the CPU and Vector units have to do your Floating Point operations, as the Graphics Synthesizer doesn't give you any floating point performance from what I can tell looking at the specs. Gekko on GC isn't as beefy in that sense, but that's partly because it has a GPU that can handle a lot of that for it. Which you seem to be ignoring here.
    Flipper does have good floating-point performance, which is what I'd expect from fixed hardware. I just checked the docs again for Gamecube, and it turns out that Flipper doesn't have a post transform cache (I thought it did...hazy memory). That means that it has to transform EVERY vertex multiple times as it is rendered. So take a simple cube as an example. It has 8 vertices, 12 triangles and each vertex is used by four different triangles. This means that Flipper has to transform the same vertex, four times. That's not so far off what happens in real game geometry. So take the number of GFLOPS for flipper and divide by around three (because that's closer to what happens in geometry in a game) to get a real idea of it's realistic "per vertex" GFLOP rating in a simple lighting situation.

    Looking at numbers is all well and good, but understanding how both machines work in the real world, is critical here. And this is why I have such a problem with ABF's constant use of Factor 5's comparisons of the two machines. Julian Eggebrecht's team may have had a great understanding of the GC, but they knew very little about how to use the PS2. And because of this, I tend to dismiss their comparisons.


    Quote Originally Posted by TrekkiesUnite118 View Post
    And for what it's worth, according to Dolphin the LucasArts logo animation for Rogue Squadron 2 is rendering between 80k-120k polygons per frame. That is until the camera comes over top and you don't see all the individual storm troopers anymore where it drops to 12-18k.
    That number is believable, especially when there's not much going on. You can just pre-build all of your display lists, so you can spend the entire frame rendering, waiting for the GPU to finish before spending a couple of milliseconds updating animations, matrices and a few other things before kicking off the rendering again.


    Though the game was really buggy for me so I'd take that with a grain of salt. When I tried to start up a level it froze and I got a bunch of garbage on the screen. Is anyone here able to get the game to actually run in Dolphin to get some solid polygon numbers? If the numbers I got were correct though, that sequence alone is rendering 6 million polygons per second, which holy god that's a lot for just that.
    There wasn't any sort of performance analyzer for the Gamecube kits. If I remember rightly, there wasn't even a CPU profiler. I can't find one at any rate. And also, the dev-kits they weren't able to run retail discs. I have a tonne of Dolphin and SN kits here sitting in a box, but installing the drivers might be tricky. I might see if I can install XP on my old Netbook at home and do some tests, if I have the time at work. Explaining why I'm doing it might be a hard sell to my boss, though

    Ironically...I'm supposed to be writing a foliage renderer for the 3DS. Go figure.

    ** edit **

    Like the Dreamcast dev-kits, Dolphin kits are SCSSI based. Just looked at SCSSI to USB adapters. Holy crap, those things are expensive!
    Last edited by rusty; 02-12-2014 at 04:50 AM.

  14. #779
    Road Rasher
    Join Date
    Feb 2013
    Posts
    423
    Rep Power
    16

    Default

    Some info on Gamecube, including but not limited to Flipper:

    http://www.anandtech.com/show/858

    And XBox:

    http://www.anandtech.com/show/853

    Just in case anyone wants to read it and they haven't been posted already.

  15. #780
    End of line.. Hero of Algol gamevet's Avatar
    Join Date
    Jan 2008
    Location
    Dallas, Texas
    Posts
    8,723
    Rep Power
    123

    32X

    Quote Originally Posted by spiffyone View Post
    Some info on Gamecube, including but not limited to Flipper:

    http://www.anandtech.com/show/858

    And XBox:

    http://www.anandtech.com/show/853

    Just in case anyone wants to read it and they haven't been posted already.
    That was posted earlier. One of the readers made some very good points about some GC facts that may have been slightly off.

    Quote Originally Posted by Rusty
    Looking at numbers is all well and good, but understanding how both machines work in the real world, is critical here. And this is why I have such a problem with ABF's constant use of Factor 5's comparisons of the two machines. Julian Eggebrecht's team may have had a great understanding of the GC, but they knew very little about how to use the PS2. And because of this, I tend to dismiss their comparisons.
    It's not like Julian's team had a lot of time with the GC at that time. He only had 8 months to complete Rogue Squadron 2 and considering that his work, along with Retro Studios, looks better than anything Sony's 1st and 2nd parties produced for PS2, he couldnt have been too far off. Really, if the PS2 was really that powerful, Sony's own studios should have bested the efforts on the GC. Just look at what Naughty Dog has done with the PS3, which pretty much leaves no doubt about which console had the better hardware of that gen.
    Last edited by gamevet; 02-12-2014 at 01:45 PM.
    A Black Falcon: no, computer games and video games are NOT the same thing. Video games are on consoles, computer games are on PC. The two kinds of games are different, and have significantly different design styles, distribution methods, and game genre selections. Computer gaming and console (video) gaming are NOT the same thing."



Thread Information

Users Browsing this Thread

There are currently 5 users browsing this thread. (2 members and 3 guests)

  1. Cloofoofoo,
  2. Yharnamresident

Posting Permissions

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