PS2 vs Game Cube graphics?

[Bastardcat]

I never knew much about the GC specs but always thought the graphics looked much cleaner than the PS2. It definitely handled textures better. My impression of the PS2 is that it pushed more polys but they weren't as good looking.

As someone who plays with lighting, textures and shaders on an almost daily basis, I can tell you that proper lighting and high res textures can make even lower poly count models look brilliant.

[gamevet]

I never knew much about the GC specs but always thought the graphics looked much cleaner than the PS2. It definitely handled textures better. My impression of the PS2 is that it pushed more polys but they weren't as good looking. Seems like comparable machines with different emphases (I aways felt like Sony perhaps rushed the PS2 because of the DC, and it was a bit slapdash in the same way the Saturn was). But to my eyes, the GC was definitely superior.

I don't think either was as powerful as the Xbox though. I can't imagine the DC or PS2 doing Halo justice. It seemed like the Xbox had the best of both worlds, great textures and high polys. And the "off the shelf" PC design probably made it easy to program (especially for experience PC programmers).

I believe the GC was more powerful than the Xbox; Just look at the most demanding games for the GC. I don't think the Xbox could pull off Rogue Leader, or make Metroid Prime look any better than it did on GC. As was stated above, Madden 2006 looked better on the Gamecube, over the Xbox version.

[sheath]

The simplest way to end that argument is to look at the RAM. The XBOX has 64MB of 6.4 GB/s unified memory. The Gamecube has 24MB (2.6 GB/s) plus 16MB (81Mhz 81MB/s) and Video RAM: 3MB (20 GB/s) with access to the 2.6 GB/sec Main RAM. The PS2 has 32MB of 3.2 GB/s (?) RAM with 4GB of 40GB/s Video "cache" that was frequently used for both the frame buffer and textures. The Dreamcast has 16 MBytes 800 MB/sec RAM with 8MB Video RAM that included up to 5:1 compression supposedly without a hit to performance.

This means, bar none, the Xbox had the most potential for that generation regardless of theoretical polygon counts of central processors in a vacuum. In game, I have found the PS2 and Dreamcast to be the weakest in polygon performance. The Dreamcast, Gamecube and Xbox seem to have different strengths in texture mapping and effects that the PS2 lacks in the average game.

It is also worth noting that last I checked the Xbox had a max sustainable polygon count of 6 million per second, but only 3 million max in game, and the PS2 version of Resident Evil 4 ran at 900k polys per second according to Capcom. It will no doubt take decades to figure out which of these systems had any sort of absolute edge, and I doubt any of them really had all of the marbles.

[gamevet]

The simplest way to end that argument is to look at the RAM. The XBOX has 64MB of 6.4 GB/s unified memory. The Gamecube has 24MB (2.6 GB/s) plus 16MB (81Mhz 81MB/s) and Video RAM: 3MB (20 GB/s) with access to the 2.6 GB/sec Main RAM. The PS2 has 32MB of 3.2 GB/s (?) RAM with 4GB of 40GB/s Video "cache" that was frequently used for both the frame buffer and textures. The Dreamcast has 16 MBytes 800 MB/sec RAM with 8MB Video RAM that included up to 5:1 compression supposedly without a hit to performance.

This means, bar none, the Xbox had the most potential for that generation regardless of theoretical polygon counts of central processors in a vacuum. In game, I have found the PS2 and Dreamcast to be the weakest in polygon performance. The Dreamcast, Gamecube and Xbox seem to have different strengths in texture mapping and effects that the PS2 lacks in the average game.

It is also worth noting that last I checked the Xbox had a max sustainable polygon count of 6 million per second, but only 3 million max in game, and the PS2 version of Resident Evil 4 ran at 900k polys per second according to Capcom. It will no doubt take decades to figure out which of these systems had any sort of absolute edge, and I doubt any of them really had all of the marbles.

Yet, the Xbox never (appeared to) pushed beyond that 6 million polygons per second count.

Rogue Squadron II pushed the GC beyond the specs Nintendo listed for the console (6-12 million polygons per second).

http://www.segatech.com/gamecube/overview/

First off, lets point out that Nintendo is being very conservative in it's 6 to 12 million polygons/seconds rating, as the developer Factor 5 who is developing Star Wars: Rogue Squadron is already doing 12 million polygons/second, and the developer claims they are only using 50 percent of Gamecube's power.

Factor 5 indicated they could get 20 million polygons/second per second with all effects. Effects stands for texture layers, and not polygonal lighting.

[Tho]

The simplest way to end that argument is to look at the RAM. The XBOX has 64MB of 6.4 GB/s unified memory. The Gamecube has 24MB (2.6 GB/s) plus 16MB (81Mhz 81MB/s) and Video RAM: 3MB (20 GB/s) with access to the 2.6 GB/sec Main RAM. The PS2 has 32MB of 3.2 GB/s (?) RAM with 4GB of 40GB/s Video "cache" that was frequently used for both the frame buffer and textures. The Dreamcast has 16 MBytes 800 MB/sec RAM with 8MB Video RAM that included up to 5:1 compression supposedly without a hit to performance.

This means, bar none, the Xbox had the most potential for that generation regardless of theoretical polygon counts of central processors in a vacuum. In game, I have found the PS2 and Dreamcast to be the weakest in polygon performance. The Dreamcast, Gamecube and Xbox seem to have different strengths in texture mapping and effects that the PS2 lacks in the average game.

It is also worth noting that last I checked the Xbox had a max sustainable polygon count of 6 million per second, but only 3 million max in game, and the PS2 version of Resident Evil 4 ran at 900k polys per second according to Capcom. It will no doubt take decades to figure out which of these systems had any sort of absolute edge, and I doubt any of them really had all of the marbles.

It's not about ram. Every piece of the machine counts.



This video explains how only around 50% of the ps2 was being used by 2003.
Here's a link to SCEE research : http://www.technology.scee.net/files...rHaveWeGot.pdf

They used Silent hill 3 and Jak 2 as examples among many other ps2 games. SH3 and Jak 2(a game with 10 million polygons at 60 fps) were using less than or equal to 56% of the ps2 . Gran turismo 3 was using only 26% of the ps2 cpu.

Gamecube is also explained in the video. The highest performing GC game is rogue squadron due to its huge levels full of bump mapping and yet the developers state they were only using 50% of the gamecube.

The gamecube is slightly stronger than the ps2 but both consoles were extremely underused due to their different architectures.

Also it is absurd to compare ports. Ports are never an indication of one console's superiority over another. Programmers usually specialise in one machine architecture and if not the developers outsource programming to developers who may or may not be up to the task of porting perfect visuals. RE4 was a rushed port done in one year by a studio in capcom who were not very good at ps2 programming. I forgot what studio number it was, but it wasn't the studio responsible for the resident evil outbreak series on the ps2 who shouldve done the RE4 port. It is so strange that capcom delegated programming to a studio other than the RE outbreak guys. RE outbreak ran at 60 fps and had the dynamic lighting engine needed for RE4. All they had to do was cut the frame rate down to 30 fps to accommodate RE4 areas.

[Bastardcat]

Outbreak was more an example of what NOT to do.

[sketch]

As someone who plays with lighting, textures and shaders on an almost daily basis, I can tell you that proper lighting and high res textures can make even lower poly count models look brilliant.

I believe it. When I first saw the GC running, I was amazed by how crisp and clear everything looked. When I compared it to the PS2, the PS2 left me underwhelmed.

I believe the GC was more powerful than the Xbox; Just look at the most demanding games for the GC. I don't think the Xbox could pull off Rogue Leader, or make Metroid Prime look any better than it did on GC. As was stated above, Madden 2006 looked better on the Gamecube, over the Xbox version.

I think the Xbox could have pulled off Metroid Prime easily. I do not believe the Gamecube could have pulled off Halo or Halo 2. I owned both systems simultaneously and never saw anything on the GC that I thought couldn't be done on the Xbox. So we'll just have to agree to disagree; I'm no spec monkey, but to my eyes, Xbox ruled that generation for graphics.

[Televator]

Bah! I can't stand robotic voiced over vids. Fragmented and strangely timed sentence structure is a pain to follow.

[sketch]

This thread should have been set up as a poll; it's always nice to get numbers on a simple A/B question.

[kool kitty89]

One thing that pissed me off about Resident Evil 4 on the Gamecube was that it was letterboxed in a shitty 4:3 image area. The PS2 version was true 16:9 and as a result that means the game has a higher vertical resolution than the GC version. There was no excuse for the GC version to be like this.

Weird, the GC only runs letterboxed, there's no fullscreen 4:3 mode even?

The polygons on the PS2 version are rendered in 16:9 so yes, it does have a higher resolution than the GC version. That doesn't mean the textures or lighting are better, though.

And yes, it is 4:3 no matter what video connection or mode you run the game in. The GC version is gimped for no reason, unless the GC couldn't handle those graphics at 16:9 just like Nintendo claims that it couldn't handle Wii Zelda in 16:9 (though I don't think Nintendo ever actually made any 16:9 games for the GC so they could just be full of shit).

Many games on the GC support anamorphic 4:3 (like many support 480p), but I'm not sure how many Nintendo published games support it. The first game that comes to mind is Star Fox Adventures.

One is anamorphic and uses the full 720x480 resolution. One is 720x404 inside a 720x480 frame. 480 is more than 404. If you play Gamecube games on a 4:3 TV (which you shouldn't), then it wouldn't make a difference.

404 lines isn't letterboxed to 16:9, you'd need 360 lines for that. Given the usual 448 visible lines on most SDTVs, 404 lines isn't even that much letterboxing (less than SFII on the SNES and Genesis, and even less compared to most SMS games, Silpheed, Virtua Racing, etc -at 192 lines, equivalent to 384i). Hell, that's a LOT less letterboxing than most PAL users had to routinely deal with for SD gaming.

Still, I wonder why they decided to do that at all. . .

I consider Resident Evil 4 to be fairly representative of the differences in PS2 and Gamecube graphics. Typically the PS2 has the worst texture maps of that generation. Dreamcast and Gamecube are roughly second place to the Xbox in texture mapping quality and effects.

That's the problem with comparing the PS2, the architecture was significantly different from the competition, so ports don't represent themselves well. (let alone the programming difficulties on the PS2 -especially related to documentation/tools for using the VUs, though that improved late-gen and more ambitious developers took it upon themselves to build/improve their own tools -the quality/performance curve for the PS2 was obviously far, far steeper than any other console that generation though)

There are very few games (let alone late-gen/mature ones) that were anywhere close to equally represented on Xbox/PS2/GC. There are almost no games where equal attention was given for development of all versions with equally mature tools and equally experienced developers. You usually saw games that were made for the Xbox or GC first and ported to the others, usually with good results of GC/Xbox (still weaker than the native system in most cases), but significant degradation on the PS2 versions. Likewise, you have some cases of really well optimized PS2 games that make poor showings on the competition. (and many more games that were exclusive)

Aside from the difficulties in programming in some cases, there's also the fundamental differences in architecture (hardware effects, flexibility -PS2's VUs emphasize that heavily, memory bandwidth and capacity, and compromising bottlenecks). Each if better at some things than the others, and all can have games tailor made for the strengths of said platform. (more so if the hardware is supported with equally powerful tools -in the PS2's case, the VUs made for a very flexible system, but also meant lack of direct hardware support of effects handled by "conventional" GPUs in the competition -the VUs could replicate many of those effects, but needed to be programmed to do so . . . they also were flexible for offloading some other tasks or doing some things the fixed-function GPUs of the competition couldn't do at all, but you'd need software tailored to support that -the fact that PCs of the time were far more like the DC/GC/Xbox architecture obviously played a role in game design tendencies of the time; granted, the PS2's massive popularity ended up inducing strong software support in spite of situational disadvantages, and some games really did push the hardware, though perhaps never truly showing the full potential of the VUs' flexibility -sort of the opposite case of the PSX, where it was the system that was mot fixed-function, PC-accelerator like, and straightforward to program with great SDKs from day 1, except both the PS1 and PS2 shared the massive hype and consumer interest driving software support)

Aside from the VUs, the memory bottlenecks on the PS2 are also notable. (32MB 16-bit RDRAM -with the inherent latency issues- on a shared bus with a smaller dedicated embedded RAM pool/buffer for the graphics hardware to access -but you'd either treat that as the designated video RAM, or you could use that as a buffer and implement a paging mechanism for updating that from main RAM -though the trick would be doing that while maintaining max bandwidth . . . the GC sort of did that too, but with 3 separate memory chunks -embedded CPU RAM, embedded video RAM, and shared general purpose RAM, the DC had more dedicated video RAM and no shared RAM buffer as such, and the Xbox was optimized totally around the idea of a unified high-bandwith bus with no dedicated buffers -aside from dynamic hardware caches, different from scratchpads/buffers- and the 360 did the same thing -so did N64, but with more latency/contention issues, and the Jaguar with far more contention than that obviously)

CPU performance is one of the more straightforward areas to compare though, and from raw power alone, it goes Xbox, GC, PS2, and DC. (it's not just about clock speed either, but it happens to work out in that order in this case -and you could get more detailed about just how much faster one is than the others, etc) Memory bandwidth is also a bottleneck for the CPU, but that's not really a mitigating factor for the Xbox either. (though you could argue the GC's dedicated CPU bus has advantages)

The simplest way to end that argument is to look at the RAM. The XBOX has 64MB of 6.4 GB/s unified memory. The Gamecube has 24MB (2.6 GB/s) plus 16MB (81Mhz 81MB/s) and Video RAM: 3MB (20 GB/s) with access to the 2.6 GB/sec Main RAM. The PS2 has 32MB of 3.2 GB/s (?) RAM with 4GB of 40GB/s Video "cache" that was frequently used for both the frame buffer and textures. The Dreamcast has 16 MBytes 800 MB/sec RAM with 8MB Video RAM that included up to 5:1 compression supposedly without a hit to performance.

Huh, I didn't realize the GC's shared DRAM buffer was only 81 MHz, that's slower than the Dreamcast by a good margin . . . and crap, it's only on an 8-bit bus, that's horrible. 81 MB/s is just pittiful, even the DC's 66 MHz 32-bit audio SDRAM was 266 MB/s. (let alone the 100 MHz 64-bit main/video buses at 800 MB/s)

The 2.6 GB/s and 20 GB/s for the CPU/GPU on the GC also seem a bit odd relative to the CPU/GPU clock speeds.
In the CPU's case, it's quoted as having a 162 MHz 64-bit front side bus (with 1.3 GB/s peak bandwidth), but the RAM is quoted at 2x that clock speed. (so 2.6 GB/s max, but the CPU can't use that if the FSP speed is accurate -and assuming DDR interfacing wasn't supported, which I don't believe any G3 PPC chips did -except maybe the Wii's derivative)

The video buffers also are quoted at far less than 20 MB/s: 10.4 GB/s peak for the 1MB texture buffer and 7.6 GB/s for the 2M framebuffer. (note that that's only 3 MB, less than the 4 MB scratchpad on the PS2 -obviously, both the PS2 and GC must update that scratchpad/buffer for new textures -3/4 MB is a pretty small space to hold ALL textures and framebuffers, or just 1 MB for textures in the GC) How efficiently the GC's buffer can be updated vs the PS2's is another matter though, and I'm not sure of the details on that. (GC has both the 16 MB buffer and 24 MB CPU work RAM to use for secondary graphics storage vs PS2 only having the 32 MB RDRAM bank, though the 16MB bank on the GC is going to be pretty limited with that piddly bandwidth)

The PS2's listed video bus speed is just crazy . . . if that's accurate, it must be embedded RDRAM (like main memory) on a much wider bus. (like 128 bits or 256 bits rather than 16-bits like main is)

Also note that the PS2's CPU should max out at 2.35 GB/s, and that's if the external bus is full speed. (though the VUs and video DMA might be able to max out the RDRAM in certain situations -obviously, average bandwidth is going to be far less than peak in all cases on all systems, though RDRAM has the added hurdle of latency, etc -similar for the N64)

This means, bar none, the Xbox had the most potential for that generation regardless of theoretical polygon counts of central processors in a vacuum. In game, I have found the PS2 and Dreamcast to be the weakest in polygon performance. The Dreamcast, Gamecube and Xbox seem to have different strengths in texture mapping and effects that the PS2 lacks in the average game.

No, not really. RAM is a factor for sure, but it's certainly not the end all be all of anything. (it allows certain things that aren't possible with less RAM -or makes some things easier to do, but it's hardly the definitive factor) Even having RAM+bandwidth advantages isn't the end. (having the Xbox with the same CPU and same RAM, but a basic highcolor/truecolor framebuffer set-up in place of the GPU wouldn't be remotely close to performance of the contemporaries -even the DC . . . well, it would still probably be better at doing a port of Outcast than any of the others )

It is also worth noting that last I checked the Xbox had a max sustainable polygon count of 6 million per second, but only 3 million max in game, and the PS2 version of Resident Evil 4 ran at 900k polys per second according to Capcom. It will no doubt take decades to figure out which of these systems had any sort of absolute edge, and I doubt any of them really had all of the marbles.

Raw (or even tempered/average) polygon counts are also not all that useful since they don't take into account: resolution, color depth, game logic/AI complexity, physics, texture detail, various graphical effects (lighting, AA, reflection, shading, bump mapping, etc), etc, etc.

Plus, there's different ways you could measure polygon output: you could measure raw polygon rendering done for each frame (including those drawn off-screen/overwritten), or just those visible in the final screen as displayed (and obviously, more efficient back-culling/clipping/sorting/etc will result in more efficient rendering with less overdraw -the DC's tiled rendering mechanism is one way of optimizing a rendering pipeline as such, among others).

Yet, the Xbox never (appeared to) pushed beyond that 6 million polygons per second count.

Rogue Squadron II pushed the GC beyond the specs Nintendo listed for the console (6-12 million polygons per second).

http://www.segatech.com/gamecube/overview/

The currently quoted peak performance possible on the GC's GPU is 20M polys/s.

Not that that really matters without added context. (what matters is how the game comes together as a whole, not theoretical raw performance -albeit some machines never get a chance to show what their real-world capabilities are, due to lack of software support -not the case for any of those in question though)

It's not about ram. Every piece of the machine counts.



This video explains how only around 50% of the ps2 was being used by 2003.
Here's a link to SCEE research : http://www.technology.scee.net/files...rHaveWeGot.pdf

They used Silent hill 3 and Jak 2 as examples among many other ps2 games. SH3 and Jak 2(a game with 10 million polygons at 60 fps) were using less than or equal to 56% of the ps2 . Gran turismo 3 was using only 26% of the ps2 cpu.

Gamecube is also explained in the video. The highest performing GC game is rogue squadron due to its huge levels full of bump mapping and yet the developers state they were only using 50% of the gamecube.

"50%" is extremely vague as it might not actually mean using 1/2 the system overall. It might mean that you're getting 1/2 the potential performance in some specific areas (perhaps graphics) due to other bottlenecks (like CPU resource). There's a number of games on the PSX that could have run at close to 2x the framerate had it not been due to the CPU. (though others might be be simple enough CPU-wise that they are limited graphics wise -like tech demos tend to be)

It's important to look at a system as a whole. If the vast majority of games are using parts of the system well below peak performance, it may be due to poor programming/tools, or it may just be that the sort of games that are popular demand more resource than is possible in some other areas in the system. (again, having not enough CPU grunt to use the graphics at 100% bandwidth in a given game is a simple example)

The gamecube is slightly stronger than the ps2 but both consoles were extremely underused due to their different architectures.

Different architectures???
I can see the PS2 (using the VUs at the time was even stranger than the Cell in 2006/07, and the tools were worse initially too), but the GC's architecture was relatively straightforward/similar to DC/PC/Xbox (in terms of hardware graphics functionality and such). The only major disadvantage of the GC was its poorer popularity.

The PS2 OTOH had the opposite problem: its popularity was so great that developers were willing to pour tons of effort into it regardless of being difficult to port to/from (or develop for in general with the early tools offered by Sony), and it most definitely had a number of high-quality late-gen games that push close to its max potential, at least for those genres of games. (there may be certain genres/game styles that the system could have shined with that simply weren't popular at all at the time -or possibly taking advantage of the flexible VUs to do non-standard effects or even accelerate non-polygonal rendering styles -ie things that would only otherwise be possible with brute force CPU rendering with perhaps modest assistance from GPUs -which were optimized for polygon rendering, but should help with more general purpose drawing tasks if programmed properly -though that would involve custom APIs for the most part since "standard" APIs tended to focus primarily on 3D polygon performance . . . OTOH most APIs had to be custom written to the PS2 anyway -though some developers opted to do a lot of stuff in assembly/coding to the hardware, especially early-on before anyone had invested in producing comprehensive APIs for the system)

Also it is absurd to compare ports. Ports are never an indication of one console's superiority over another. Programmers usually specialise in one machine architecture and if not the developers outsource programming to developers who may or may not be up to the task of porting perfect visuals. RE4 was a rushed port done in one year by a studio in capcom who were not very good at ps2 programming. I forgot what studio number it was, but it wasn't the studio responsible for the resident evil outbreak series on the ps2 who shouldve done the RE4 port. It is so strange that capcom delegated programming to a studio other than the RE outbreak guys. RE outbreak ran at 60 fps and had the dynamic lighting engine needed for RE4. All they had to do was cut the frame rate down to 30 fps to accommodate RE4 areas.

Comparing ports can be useful, but they are inherently limited. (as I already addressed to sheath above)

[Joe Redifer]

The highest performing GC game is rogue squadron due to its huge levels full of bump mapping and yet the developers state they were only using 50% of the gamecube.

This doesn't make much sense to me. Rogue Squadron was a launch game. If it was supposedly only using 50% of the GC's power, why didn't better looking games come out as developers got more familiar with the system? You mean to tell me that a friggin' launch game is the best the system will ever do? That's pathetic! And Rogue Squadron looks worse than a lot of Xbox games. I think the developers and publishers of Rogue Squadron were blowing smoke up people's asses, trying to impress people so they'd buy the game. It's only in 4:3, too. I don't even remember if it supports 480p but I am thinking it probably does.

[A Black Falcon]

Rogue Squadron II wasn't the best looking Gamecube game. It was a very, very good looking game, but the best? Certainly not. For one thing, its sequel, Rogue Squadron III: Rebel Strike is substantially better looking, to the degree that they actually managed to get the entire original game working in splitscreen mode in that one, with no cuts...

Both games do have extremely impressive graphics, though. Factor 5 did amazing work on both titles, visually. And the gameplay's pretty good too...

[TrekkiesUnite118]

I remember years ago there was a site that compared a lot of these specs for all 4 consoles of that Generation. The only interesting detail I remember from it is a lot of the times Dreamcast theoretically could out perform PS2 in some aspects it talked about. I also remember it mentioning that both Gamecube and Xbox also have support for texture compression similar to the Dreamcast, but they use newer more advanced algorithms for it.

[djshok]

Wow, turned off by the 32-bit era? Well, tastes are different I guess.
For me that era is the pinnacle of gaming. Never before or after did I enjoy games so much.

It wasn't so much a matter of taste as much as early 3D graphics giving me massive migraines. The blocky polygons with shifty pixel textures did something terrible to my eyes

[Clessy]

Bah! I can't stand robotic voiced over vids. Fragmented and strangely timed sentence structure is a pain to follow.

Yeah I cant watch that video and actually pay attention. The voice is awful.