Console that fared best vs PC's available at launch

I don't understand how PS3 can have games like GT5P, Uncharted, Heavenly Sword, MLB 08 The Show, Ratchet and Clank, and countless unreleased titles like killzone2, haze, little big planet, MGS4, RFOM2 and be seen as anything other than a tremendous graphical success.

 

Gamecube it would be IMO. PS3 is shaping up with several intresting games although many games are not really up to what they have been hyped up to, in some cases massively hyped up. Although offtopic.

 

Just to be clear, I was only referring to graphical muscle, not the actual games on the system in which case PS3 seems to be doing pretty well.

Haze though.....??

 

Ah, yes. Let's compare graphical accomplishments on different platforms by picking the worst examples we can think of, shall we? PC's graphics are so awesomely powerful, every game released on Windows makes the best the current consoles manage look like sick dogs.

It's also worth considering that in graphical muscle, if 100% of the Cell was dedicated to graphics, it's achievements would be a lot better than what you see in games. And if you don't just limit yourself to comparing a $400 to a multi-SLI'd grand's worth of GPU gear, in some aspects the consoles trump the PCs nicely. It's fair to say GPU wise PS3 didn't launch in a strong position to the best PC had to offer, but that's a pretty blinkered view of the whole system.

 

If your referring to my mention of Haze then you should know that I mentioned it in response to ShaidarHairan referencing it as an example of excellent PS3 graphics.

I see from your reference to "the worst examples we can think of" that you agree with me that its not a good example to use.

By graphical muscle, I mean the whole game package. Graphics, physics, AI, world size, etc.... not just pure eye candy, i.e. which system is more capable of deliviering technically superior games. Obviously dedicating Cell to pure eye candy would sacrafice in other areas. On the other hand it does stand PS3 in good stead for the CPU driven tasks.

Even when we limit ourselves to single GPU systems, PS3 launched into X1950XTX territory . Certainly RSX is no match at all for that. As you say, Cell can certainly help out to an undeterminned amount but given that it would also be competing with Core2 Duo's at the time of launch I doubt it could afford to dedicate too much power to help out RSX and still keep up in the CPU depertment.

 

There may have been a bit of a difference, but I don't ever recall having to wait anything remotely like a year to get stuff like the TNT and GeForces in Europe. Give or take a few weeks, this stuff seems to appear around the world fairly close together - certainly compared to consoles!

These are the kind of numbers that Nvidia and ATI would be very happy to sell through on their high end products on day one. That's another plus for the PS1 and Dreamcast - they not only launched with technology and features that outstripped the very best the PC had to offer, they launched in (relatively) large numbers and sold at relatively low prices. A long way from the kind of paper launches the PC GPU world has seen form time to time, anyway.

Comparisons between the PS2 and PC aren't quite as straight forward IMO due to a difference in the way they seemed to direct their power (fillrate and polygons vs texturing features), but there's no doubting the PS2 had a huge amount of power and lots of silicone. I'd forgotten just what an amazingly strong start it hard - Nvidia and ATI would *kill* for those numbers on launch day of their new super high end stuff. Didn't the 8800 GT ship something like 100,000 for launch, and the 3850/3870 (a pretty small chip actually) ship something like 200,000 - 300,000 for it's first couple of months?

 

I'm looking forward to the Minesweeper DX 10.1 patch!

There's only one console that really stands out as doing really badly compared to the PC (I thought the GC was okay for mid 2001 tbh), and that's the Wii. The Wii, which is dominating everything in it's it's path, and threatening to drown Nintendo in money.

 

Ah, sorry. Missed his reference to that game. Haze looks 'okay', though there's some seriously primitive textures in that game. It's certainly not a graphical showcase to prove how capable a system is.

Well, we still don't really know. As with prior consoles, you can only know what the hardware is capable of at the end of it's life when it's being maxxed out. Then you get to compare those titles to what was possible on PC at launch. That's if you're intending to compare attained performance versus paper specs or factoring in easy-of-development into your analysis.

 

This is easy--the Atari VCS (later renamed 2600). Launching in 1977, the only personal computers it really competed with were the Apple II and Commodore PET, which with their monochrome, character-only displays, were hardly much of a challenge to the 2600's colorful graphics. 2600 just blew everything else out of the water.

 

Ok, in my original post I was somehow under belief Voodoo1 appeared in 95, hence I underestimated PS1 by quite a bit. With that in mind, I would have to put it ahead of PS2 (mainly due to PS2 software lagging badly behind HW).

Those are at the very bottom of PS2 software selection in terms of system utilization. They were an order of magnitude behind average PS2 title in terms of polygon complexity, and the amount of work done per pixel (and vertex) isn't even close to good PS2 titles either.

I dislike bringing resolution as some benchmark though because hw was build for different market targets. Even in cases where video-out existed(PS2, XBox1), there was essentially no market for games running in hi-def.
It's obvious the hw could run low-tech games like Quake3 in higher resolution just fine though.

But yea, it took a long time for PS2 SW to start catching-up to HW, which puts it in less favorable light in early years.

I don't think it really happened until V1 though. Ports like Quake2 easily looked and ran better then SW rendered on PC (I played Q2 in SW on P133, and later K6233, for long time - accelerated look threw off my game online).
Forsaken looked like PC accelerated version minus bilinear filtering (and still 60hz).

Yea the resolution aspect is always there of course, but that was true of DC too.

 

Sure, but here's my point. Say that we use your metric of RV630+25% in COD4 for Xenos. That would be much higher than the 7900GTX, whereas RSX would be much lower. However, both would be usually 60 fps at 1024x600 on a TV.

Just because you see similarity with cross-platform games doesn't mean they're performing equally. We just don't have the data to make the sweeping assertion of equality that you're suggesting. Maybe if the hackers can keep making progress on RSX, we can.

RV670 has a 50% perf boost over R580. That doesn't happen with older PC games, nor does G80 nearly triple G71. There really does seem to be some advantage for the unified architecture.

Well, like I said, you're rarely going to see onscreen evidence even if one was faster. Hopefully we can get more real evidence with homebrew code. We do have all those comments from devs, though.

You keep saying this, but really there is no evidence from cross-platform games suggesting this. From what I can gather, you're using similar framerate on PS3 and 360 implies RSX = Xenos implies R580 is faster. Not only is the first link hard to measure, but it doesn't correlate to the second.

You're probably talking about that remark about the Toy Store demo, right? That's solely a pixel shader demo, and I already agreed on that point.

Note that you're replying to my comment about vertex load. I wasn't trying to say a 360 would beat an X1950XTX in COD4 (though I think it's possible at 1024x600 when looking at resolution scaling on the PC), but that a unified architecture will show advantages on games that are primarily targeted for consoles.

The interface doesn't matter because it's never a bottleneck. That's all I was trying to say. Framebuffer BW never limits Xenos. It does limit R580 and RV630 in various situations, though, regardless of compression.

If math ability mattered much, RV630 would clobber G84 instead of falling behind. Memory BW doesn't help R600, as we saw with RV630 (EDIT: sorry, meant RV670). So the only reason for RV630 to perform well below 50% (look here, for example, noting that R600 performs similarly to the 8800GTS) is ROPS. If ROPs are the limiting factor for any portion of the workload, my guess is that Xenos will usually smoke RV630.

Okay, this sounds better, though it should be noted that sometimes RV670 is notably faster than R600 (like in those COD4 benchmarks) so it really does seem like R600 has some bugs/oddities. There's no reason to assume Xenos had them too, as the internal arrangement is quite different.

Considering NV30 had twice the pipelines and a massively higher clock, it was slower. Even per GB/s it was slower, from what I remember.

COD4 doesn't exactly put the 360 to shame. As I reasoned above, there's a decent chance that the 360 matches a 2005 PC there. What other examples do you have?

As for other differences, you have things like rapid renderstate change on the console. Xenos can quickly switch between 7 different contexts, AFAIK. I don't know if R600 has that or if its driver can even get the related commands from DX10. Some optimizations can be made too, like using Xenos' 64 sample/clock Z-only rate for shadow maps, but I guess you can argue that we don't know how often these are used.

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

In any case, software was a big part of my argument as to why 360 should be considered to be faring well with PCs at release. Naturally, console games can have a much bigger budget for art and content. Even though this has always been true, in the past PCs had one claim to fame that even inferior software could take advantage of: resolution.

That's why it's hard for me to say Dreamcast was leagues above the PCs of the time. A Voodoo2 could produce a picture at 800x600 that no console could, even though a Dreamcast could do much better stuff at TV resolutions. In the HD era, though, that loophole is gone. Sure, you still have ultra-high resolutions on the PC, but the advantage of that over HD isn't nearly as important as the leap PCs had over SD for the last decade.

 

Ps1 all the way... But only because back then, the leap in 3d perfomance wasn't on the Pc market, but rather in arcades.

 

You ever play Luigi's Mansion? It was rather good looking for a launch game. Much, MUCH better than any PS2 launch game.

Note: I checked and confirmed it launched with the GC in Japan. The system and game were released on September 14th, 2001.

 

I think remember getting a beta Rendition Verite v1000 board in summer 1996 to play the Rendition version of Quake1 before I ever got a Voodoo card.

Tommy McClain

 

PS2 came out during the era of Geforce 2 right? Quake 3, Unreal tournament etc have all been released, and played at much higher resolution than any PS2 game. I simply don't see PS2 outclassing the equivolent PC hardware at the time.

 

True, but CoD4 is a bit of a one off as in most games the 2600XT is more in line with a 7900GS. CoD4 does indeed show us that in some cases the G71 architecture fares very poorly vs R6xx (and perhaps by extension Xenos) but thankfully for the PS3, that doesn't seem to be the case the majority of times.

If CoD4 represented the performance norm for G71/R6xx then even by my more conservative estimate Xenos would be 50% + faster than RSX across the board! And i'm sure that kind of sweeping advantage would come across in games.

Older PC games are likely to be more texture limited than more modern games and texturing is an areas were RV670 isn't particularly more powerful than R580. Its possible the performance delta between R580 and R6xx in this case is merely down to the lack of VRAM on R580. They are only using a 256MB X1900XT so not exactly the fastest R580 available. In fact that point is furthr confirmed when looking at the firingsquad results which show the 512MB X1950XTX to be right behind the X2900XT.

Based on R580's comparitive performance to other US arch GPU's i'm still of the believe that its somethng else thats responsible for G71's poor performance here.

The evidence i'm talking of are cross platform games and their performance on an XTX powered system in comparison to how they perform on the 360. Are there any cases were its not possible to dial the resolution/AA/AF/texture detail etc... higher on the PC version and still maintain the same or better framerate?

Here's a good review with plenty of modern games on the XTX:

http://xbitlabs.com/articles/video/display/radeon-hd-2900-games_9.html#sect1

Just look at how it performs at 1280x1024 with 16xAF and often 4xAA to boot. I think we'd be hard pressed to say the 360 would manage the same feats (why the lack of AF in these games on xbox for example).

No it was just a general statement in some interview around the time of R580's launch. Unfortunatly I have no idea were to find it now so can't back this up at all but as I recall, when questioned which was more powerful, the ATI rep responded that R580 had more power but the onscreen results would be similar due to the closed box nature of a console. I assume that ignores things like higher resolutions though.

I could be wrong but i'm sure I have read here that there are things you can't do with the edram bandwidth because of that interface. i.e. It may not be a bottleneck if you use the edram in a restricted manner but its not the same as having a normal 256GB/s connected directly to the GPU because the link acts as a limitation in using it that way. At the very least it can't be used for textures or (as far as i'm aware) geometry.

Xenos has exactly 125% the ROP performance of RV630 on paper (with 2xAA or less). The only reason I can see for it performing beyond that in the real world would be bandwidth restrictions. But if RV630 is bandwidth restricted rather than ROP limited then it would indeed explain its poor performance in comparison to R600.

I think the original source of this line of discussion has got a little lost, one of the pit falls of forum discussion The context of the statement was around whether consoles and PC's had similar or different workloads to deal with when it comes to games. CoD4 was my example of how they have very similar workoads in that the end result (and the genral process of producing it) is very similar between consoles and PC's. Other examples in this context would be every cross platform game out there that runs on the same engine on both platforms.

In other words, pretty much every cross platform game you look at poses the same challenge to the hardware whether its on PC or console. Thus, in the context of gaming, the workloads are pretty much identical.

Its true that Xenos probably has a few features that are lacking in R6xx but the reverse is almost certainly true to a greater extent. Yes those features are likely to see more use in the console due to its closed box nature but in pure capability terms (which is what I have argued in this thread - rightly or wrongly ) they should be taken into account. In other words, there are unlikely to be unique features of Xenos which take it beyond RV630 +25% that can't be equally or moreso countered by unique features of RV630.

I agree, consoles do fare better than specs would suggest because they are optimised to a much greater extent than the high end PC hardware at the time of their launch. Art and content budgets may also play a factor in that. But as I said above, my arguments has revolved around hardware capability rather than what was done with that hardware as I think that when you start to bring graphics of specific games into the equation, things tend to get a little too subjective. Hell, I probably missed the whole point of the thread though!

Regadring resolution, that should also theoretically act as a disadvantage for the current gen of consoles since they now have a comparitively much higher workload than their predecessors compared to PC's of the day. i.e. if Xenos only had to render at 640x480 then you could argue that it would take an 8800GTX to match it at PC levels of image quality. However because Xenos is expected to render at 720p with AA, the GTX has bags of spare power to poor into other aspects of the graphics. Unfortunatly, because of the ever decreasing PC exclusives, that doesn't seem to be happening too much.

 

minimum framerate.

30fps average != 30fps locked.

 

Forced vertical sync makes me puke.

 

Well how many examples do you have of games primarily targetting the console? Here's Bioshock (UE3 based, AFAIK), again showing G71 falling behind (scaling 7900GS, RSX would at best match the 2900XT if BW is ignored).

How? Assume that there is a 50% advantage. If it runs at 60fps on PS3, then you'll only see 60fps on the TV for the 360. No difference. If a frame takes 30ms to render most frames on PS3, you'll see 30fps. If it takes 20ms to render most frames on 360, again you'll only see 30fps on the TV unless triple buffering is enabled or v-sync is completely disabled.

The gamespot results also show the R580 fairly close to R600. It's RV670 that's farther ahead.

Memory isn't a limitation here. The 7900GTX 512MB is 36% ahead of the 7900GS 256MB, which is less than the clock/pipeline advantage. The 8800GTS 320MB is at no disadvantage either. This is 1280x1024 with no AA.

R580 is only close to R600. It's advantage over RV630 is less than normal, and it's further behind RV670 than normal. R600 is the odd one out, probably with some bug.

How many of those games were primarily targetted for the console? And how are you able to compare performance between the PC and 360, anyway?

That's not a particularly compelling anecdote. They're trying to pimp up R580 as well, and like I said I agree that PS power is higher. That's all the justification needed by the rep to make that claim.

The things you can't do are related to not being able to texture from there, i.e. read into the pixel shader. In terms of the ROPs, they are always running as fast as they can. That is not true for R580 or RV630.

Sorry, I meant to say "as we saw with RV670". R600 doesn't get any advantage from the 512-bit bus. Comparing RV630 and RV670, BW is not a good explanation for the former performing well below 50% of the latter most of the time. BW per pipe per clock is nearly identical.

BW restrictions are very real for RV630 in fillrate limiting scenarios, because usually alpha blending is enabled then. Look up 3DMarks's single texturing fillrate test, and then note that the texture BW is very low and there's no Z-test either. In games Xenos will destroy RV630 when fillrate matters.

COD4 is an example of a console-like workload and has been quite useful to me in proving my points (high vertex load, beneficial for unified shaders, bad for G71). However, it is not a typical example of workloads for PC games. For other cross-platform games, please, give examples. Find me games that primarily target the consoles, have PC versions, and have benchmarks out there.

I'm not challenging the fact that cross-platform games have similar workloads regardless of platform. I'm saying that your notion that R580 >> RV630+25% ~= Xenos and R580 ~= G71 is not based on modern cross-platform games, but PC games and often old ones.

Well you'd be wrong. RV630 couldn't get close to 32 Gsamples/sec Z-only fillrate. It couldn't get close to 4GPix/sec textured, z-tested alpha fillrate. There's nothing that will take it's performance way beyond Xenos in a similar way except more registers (I think) and quasi-scalar shaders, which are more useful for GPGPU than console games.

Well, suggesting that a 8800GTX must render at high res to match Xenos at 640x480 is pretty disingenuous. The thing is that regardless of how crappy a typical PC game looked in 1998 (GLQuake and QII didn't look bad, IMO), a Voodoo II at 800x600 or TNT at 1024x768 was unmatchable by a Dreamcast simply because those resolutions were such vast improvements over SD. Even though a Dreamcast was far more capable at SD, I can't give it a clear victory over the PC. In 2005, I can't say the ultra-high resolutions of R580 in older games had any value over 720p games on a 360. However, there isn't really a hardware advantage so it's also not clearly above a PC, but there was definately software that looked better on 360 than any PC game on any PC, and resolution was no longer an excuse to overwhelmingly prefer the latter.

That's why, in the era of 3D accelerators, I think Dreamcast came closest to a PC at launch with 360 a close second. Before 3D accelerators, though, all 3D consoles blew away PCs. It just wasn't a fair fight .

 

Thats fine, but my point is that just because a game is running at 30fps locked on a console doesn't mean the hardware isn't capable of doing 50 or more average. It really just means it is probably not fast enough to do 60. Showing average performance numbers of PC hardware doesn't really give a level comparison to consoles.