• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

PC GPU comparable to PS3

I remeber in the PS3 promo vids before it was releaed they had pics and specs showing 2x 78xx or 79xx (cant remember which) cards and how the PS3 was an improvement over this. So basically, top end graphics in the 8 series are better than the PS3, but the comparison stops there.
 
I love my pc for gaming, and my ps3 dosnt get used as much as i would like it to be, but i think the ps3 has got massive potential horsepower wise

the cell can do a lot of the graphics work, the rsx was just put in there to give the programmers a bit of familiarity and ease programming from what i can tell

I read somewhere that some researcher is using a ps3 for some experiment with linux on the ps3, and the ps3 is on par with a very very high spec pc computing power wise, and a lot cheaper

also uncharted apparently only uses about 30% of the potential power of the ps3, and that looks pretty amazing to me already, I think that once the programmers really get to grips with the hardware of the ps3, we are going to see some amazing stuff on it

interesting read http://www.shacknews.com/onearticle.x/51632
 
Kojima Productions released some nice videos of MGS4 just a few months before PS3 launched. Then a few months after PS3 launched, they released more videos showing lower resolution textures, worse shading effects and lower poly models. The PS3's gpu may been going through different prototypes even just a few months before launch, so it's hard to pinpoint what it's even based upon.
 
Rubbish, the 360 has had better graphics for most games because of crap ports.

programmers are actually starting to prefer writing the games first on the ps3, and then porting them over to the other systems, so we might not have to live with crappy ports for much longer
 
Rubbish, the 360 has had better graphics for most games because of crap ports.

I don't want to get drawn into any tit for tat but I disagree.

Back on topic if thats alrite, sorry I shouldn't have been so loaded in the things I said.

PS3\360 are both fab....move on.
 
the cell is great at things like folding at home, thats the thing scientists were going on about, one guy was twittering on $399 for that, they bought a whole bunch of them to run it on. THey are accelerated cores, the SPU's would be useless as a multipurpose desktop cpu on their own, but great at basic accelerate singular tasks. so they are much faster than a much more multipurpose cpu, but thats it. Its like saying a plane is faster than a car, but how useful is a plane for day to day use, it really has no bearing in the argument, its not a "better" cpu, its simply different.

AS for SPU's being used for shaders, the whole graphics engine isn't just shaders, everything else would need to be done on the gpu. AGain i'll point out the difficulty both graphics companies have making IDENTICAL cores using the SAME drivers have working together and the lack of scaling due to the latency penalties when it needs to offload work or contact the other core. Can you imagine how much that latency penalty is when the cell can't do all the shader work, offloads it to the gpu, but translating the commands to a completely different format for a different driver then going to the gpu before it finally gets carried out.

AS with most things, the messing around with gfx on the SPU will remain in place for tech demo's and maybe even make a few basic 3d arcadey style games run completely on the spu, but the likelyhood of them working together to make a seamless up to date graphically advanced game is incredibly low. Its almost certain not to happen.

At the end of the day several other factors come in. THe RSX would still have to physically push this info to the screen and probably do final stages after calculations on the SPU, the RSX will already be balance, adding more stuff to go into the final stages wouldn't work well.

Games are going to be designed to mostly be multiplatform. So its 99% in their best interest to make a game that runs purely on the gpu thats of similar power to the 360 as in general that means graphically it will be compatible. If they could somehow get SPU's to give the console 3x the graphical power, porting games will be almost impossible, and pointless which kills lots of revenue, both ports to and from the PS3 platform. Which means only exclusive titles would maybe be used for the "extra" power. provoking massive backlash from fans who get 3-4 great titles and everything else would suck badly in comparison, again very unlikely.

In the end, they simply DO NOT NEED to do it, its complicated, means learning a whole new code set, brings about dozens if not hundreds of more issues in getting a game to run stably and would most likely massively increase the length in developement. All when they can simply make games for a highly programable gpu core that everyone has lots of experience with which is on par with the competition power wise.

Someone tell me where the extra cost and problems make it even remotely worthwhile? you can't, they can't, therefore they won't do it.
 
In the end, they simply DO NOT NEED to do it, its complicated, means learning a whole new code set, brings about dozens if not hundreds of more issues in getting a game to run stably and would most likely massively increase the length in developement. All when they can simply make games for a highly programable gpu core that everyone has lots of experience with which is on par with the competition power wise.

Someone tell me where the extra cost and problems make it even remotely worthwhile? you can't, they can't, therefore they won't do it.

insomniac games already are doing it, and other developers are going to follow lead. im not sure what you are basing your opinions on, but they are of course going to seek the fastest way around any problem on a fixed platform. that has and always will be the way:)
 
RSX specs have never been officially announced, but beyond3d seem to have a pretty good idea...

beyond3d said:
Based on our analysis and information, it seems that the RSX has 28 pixel shaders internally, with 24 active. G71, on the other hand, has only 24 and NVIDIA sells the defective parts as the 20 pixel shader GeForce 7900 GS. Such an opportunity for multiple SKUs based on the same die does not exist with consoles, and therefore higher redundancy is desirable. It's also likely that other parts of the RSX are therefore disabled for yeilding, such as the vertex shaders and possibly the raster operations units (ROPs). This is similar to the strategy used by Sony with the CELL, where 1 of the 8 SPEs is disabled.

While the RSX might have some parts removed compared to the G71, such as its PureVideo technology, transistor count is likely balanced back by having more redundancy and larger caches: indeed, we believe that most of the RSX's internal caches are larger compared to any G7x part, including the post-vertex transform cache and compression caches. In addition to its normal 48KiB of texture cache to its local memory, it also has an extra 96KiB of cache dedicated to communication with the XDR memory pool, in order to improve bandwidth utilization and average latency, and make it possible to use that memory pool for texturing operations - resulting in greater overall system bandwidth utilization.

RSX also has a FlexIO bus connecting it directly to main memory and allowing access to both memory pools, likewise Cell is able to write directly into RSX's memory.
 
Last edited:
insomniac games already are doing it, and other developers are going to follow lead. im not sure what you are basing your opinions on, but they are of course going to seek the fastest way around any problem on a fixed platform. that has and always will be the way:)

you're confusing something, the fastest way around a problem and the fastest computing are two very different things. programming for a single gpu will be infinitely FASTER in developement than programming for SPU's and a gpu to share the load, they aren't equal in any way at all so balancing, coding, cross communication will all lead to massively more complex developement. even if there is more power there, its simply stupid to use it as again they don't need to, at all. Their base games of which most will be cross platform have to work at the same settings on the 360, so why make a game that uses lots of extra power that will take longer to develope, which will then only lead to even longer time and more money to port to other platforms.
 
not really, they see everything as a challage/problem. they will always look to optimise and expand what they can do with the ps3. they will try to increasingly push it to the limits. the very nature of the cell, regardless of the rsx and the spu's, means they have to program in an almost completely new way. i think im safe in saying even sony themselves wont be getting the best from the platform for some time to come, purely because of the complex architecture if nothing else.

these programmers are all learning a completely new way of doing things anyway, so its a good time to experiment.

apparently so the rumour says, its far easier to make the ps3 the lead platform and then port to the 360. why i dont know, i dont think ive seen that explained. but if that is the case - ie:develope for the ps3 and program specifically for it, and then port to the ps3 - then its not really going to make much of a difference at the end of the day.

whatever the case, using the spu's to streamline code and increase thruput is already happening and they will keep doing it as it obviously has many benefits:)
 
insomniac games already are doing it, and other developers are going to follow lead. im not sure what you are basing your opinions on, but they are of course going to seek the fastest way around any problem on a fixed platform. that has and always will be the way:)

infact on actually reading some info from insomniac it seems they are, unintentionally massively misleading in their terms of the use of "shader".

Shader power in gpu's is essentially, well every shader is like an SPU of the main core, doing prework almost before it goes into final rendering. its essentially a massively paralel unit of small work loads being doing spread out as much as possible.

Insomniacs powerpoint presentation on it pretty much says, "why do we call them shaders" because we do small work units quickly and offload them from the main ppu cell. There are some very good things about it, its teaching coders to basically try and separate AI, physics and every other bit of game data into massively paralel small workloads rather than what we have now. IE a quad core having the main work load done on one core at 70% and the 2nd core doing sound for 5%, and the 3rd core doing driver stuff using 5% and the 4th core doing physics at 25%. The Cell might be a nice quick way to get games working to be better on multiple cores quickly. essentially spread out work now so when we get 8 cored cpu's there should already be cross platform games that will be able to spread the workload very well. Unfortunately that will only really matter on a few RTS's like now, SUp commander and so on, the 3rd and 4th core barely do any work on sup commander, mostly because the workload is certainly not split up evenly.

But their own presentation says their SPU shaders are just doing some AI, Physics, effects, animation but they are working in the same style that shaders do on a GPU, and coded to be used in a similar manner. They could have called it something else to avoid the confusion, but a little tiny hint of misleading there to "create" the effect they'll have more gpu power than other consoles, i think thats highly possible ;)

They suggest they MIGHT in the future be able to do some basic calculations, IE precondition some data for the GPU, which would probably mean do a couple of basic steps before hitting the GPU, which could be good, but they certainly aren't saying they are doing that now or will be, they are "experimenting" with doing some vertex shader work, which would all then need to go to the GPU. If they ever got it working, which doesn't seem likely, we're talking about offloading 5-10% of the gpu workload, which well, any boost is good. but we're not talking about SPU's matching and doubling GPU performance at all, it can't do all the stuff a GPU can do, just a couple basic parts, and it can't load balance, as i suggested latency would simply not allow that. If it can do ALL the first 1-2 calculations before passing on to the gpu theres a clear flow of work that can be maintained which would eliminate any lod balancing that needs to occur.

I highly doubt we will ever see gpu work done on an SPU, as i said its basically ludicrous. The SHADERS insomniac are using are so called because of the similar style of splitting the work load that gpu's use shaders for, ridiculous naming as its obviously caused confusion.

The Cell is pretty sweet, but in the end, theres no real reason you can't just use multiple identical cores.
 
i know all about these 'shaders', i said this in an earlier post:
james.miller said:
because shader routines can be written on the spu's. although 'shader routine' isnt only for graphics work, they can be used for culling, streamlining and general optimisation of the graphics which obviously leads to faster performance overall. trust me, it has quite a lot to do with it

again, i have NEVER claimed they will suddenly trasform the performance of the ps3.only that it'l improve the efficiancy somewhat:

james.miller said:
guys im not saying this approach will magically transform it into an all conquering graphics monster, im just saying using the psu's moves the goalposts a little. enough not to be able to comapre anything from the pc world side-by-side. that is all i m saying
 
whatever the case, using the spu's to streamline code and increase thruput is already happening and they will keep doing it as it obviously has many benefits:)

hehe, well my post after yours pretty much backs up what i said which can be confirmed reading through insomniacs info(theres a link to the R&D stuff they have on their frontpage, i'm really impressed by how openly they discuss stuff).

At the end of the day, the gpu on the 360 is designed to be more programmable than the more standard older 7800/7900 style gpu's. THe 360 has more of a R600 scaled back to the power the x1900 had, fully programmable shaders. In all honesty, both consoles are simply limited by what they can do from a graphics point of view power wise. The 360 would seem to be easier to program for, and possibly more powerful. Therefore it would make sense to program for the PS3 first, why? Because if somethings actually slightly more limited and more narrow in its design its easier to aim for that in the beginning then port to the slightly more programmable gpu after.

Trying to think of a good analogy. hmm, i guess this as basic as it is. The nvidia core is more defined in how its to be used, its a round peg, you design the program to fit in the round hole fine, the 360 gpu is the square hole thats bigger and more forgiving and that round peg fits through it easily.

But if you had a square peg and tried to fit it in the round hole it won't fit and will take a lot more work to get in there. Also if there is a little less power, then if the game works on the nvidia core, theres no real worry it should run on the ATi core. Where you might program a game for the 360, and find it needs further optimizing or even dropping settings to run on the PS3.
 
i know all about these 'shaders', i said this in an earlier post:


again, i have NEVER claimed they will suddenly trasform the performance of the ps3.only that it'l improve the efficiancy somewhat:

sorry but thats wrong, the shaders on the SPU at the moment aren't for graphics work AT ALL thats the incorrect part of the statement. ALL they have done is split up normal stuff you do on the cpu into the spu's in "shader style". at the moment it has NOTHING to do with graphical power and NO games are using SPU's for graphical power, and its incredibly unlikely they ever will. the ONLY reason people think it might be possible is the use of the term shader, which they only did to let you know the rough idea behind how they work, IE spreading the work across multiple massively paralel cores. it has nothing to do with shaders in a GPU and the GPU power of consoles can be directly compared.

This is exactly what i'm getting at. Its more than possible that something liek FlightsimX and sup commmander could run significantly faster on the Cell, but all the other games are gpu based, the Cell doesn't and with 99.99999% probability will never hell with gpu power. You can do all the physics and AI you want as fast as you want, but the gpu still draws it and thats the limit.

Look at crysis, it barely put a 50% load on my Q6600, the Cell isn't doing anything faster CPU wise, because its not going to get fully loaded. Current physics, AI, animation and effects simply don't load a multicored cpu fully anyway.
 
Back
Top Bottom