Official AMD Radeon 6000 Series Discussion Thread

[cadaveca]

Even though the Cypress picture makes it look like it only has 1 dispatch processor it is actually like Barts in that there are two dispatch processors. Some of the Barts reviews covered this (I think maybe Tech Report & PCPer).

Yep. but not everything is doubled...it seems that Cayman may have the final bits and pieces to make it almost a complete dualcore gpu on a single chip, except the memory interface.

 

[bear jesus]

Yep. but not everything is doubled...it seems that Cayman may have the final bits and pieces to make it almost a complete dualcore gpu on a single chip, except the memory interface.

I had kind of expected gpu's to follow in the steps of cpu's and even more so with all the work that's going into multiple gpu setups.

I see it kind of like in the past with single core cpu's the only option was to get a motherboard that had two or more sockets and with a gpu the options are get a motherboard that supports two or more or get a dual chip card but when moving onto 28nm and below I'm sure it will start getting relatively easy to fit a couple gpu cores on a single die.

If it could scale perfectly it would give some amazing power, I'm sure current high end architectures would be a bit too much size and power usage wise but imagine if something like a perfect scaling quad core 460 or 5770/6870 on something like the 22nm process could be made

But this makes me wonder if the 69xx cards are getting some of the features that will be on the 7xxx cards, could the 7xxx cards be the first true dual core gpu's or at least one of them be so?

 

[cadaveca]

The way I look at it, almost everything, including physics and AI, should be running on the gpu itself. If an entire game program could be run within the confines of a gpu, I think we might see a fundamental shift in the depth and immmersion possible. I'm thiniknig holo-deck style though...probably not what most expect.

We aren't far away from rael holodecks being possible...it's actually quite scary how close it is.


To do this, gpus must become a bit more compartmentalized ...as they were before... with rendering pipelines, although this way of doing things was killed almost 5 years ago now. With seperate "cores" doing different tasks, but each working together to render a final frame, real physics simulations will be truly possible.


I kind of envision, for physics, a complete rendering cycle that never stops during gameplay...and as you play the game, the game's world's physics will be running in simulation on the gpu, creating your triangles.

DirectX 11 kinda works towards this, but as many have seen themselves, DX11 comes with quite a severe performance impact...and I bleeive part of this is because gpus are really ready just yet.


You could look at a single SIMD as a pipline, but to me, they are not fully featured enough yet. I think we need quad gpus...like how tri-sli with a physics cards runs...but I want that physics card to take a far more primary role than it does currently.

 

[bear jesus]

I was thinking much more short term than you are, maybe not even 12 months into the future.

Long term my dream would be everything rendered in molecules to help give physics that truly apply to the laws of physics as he know them, for that i don't even know if silicone chips could/will be used.

*starts thinking about quantum computing*

 

[wahdangun]

The way I look at it, almost everything, including physics and AI, should be running on the gpu itself. If an entire game program could be run within the confines of a gpu, I think we might see a fundamental shift in the depth and immmersion possible. I'm thiniknig holo-deck style though...probably not what most expect.

We aren't far away from rael holodecks being possible...it's actually quite scary how close it is.


To do this, gpus must become a bit more compartmentalized ...as they were before... with rendering pipelines, although this way of doing things was killed almost 5 years ago now. With seperate "cores" doing different tasks, but each working together to render a final frame, real physics simulations will be truly possible.


I kind of envision, for physics, a complete rendering cycle that never stops during gameplay...and as you play the game, the game's world's physics will be running in simulation on the gpu, creating your triangles.

DirectX 11 kinda works towards this, but as many have seen themselves, DX11 comes with quite a severe performance impact...and I bleeive part of this is because gpus are really ready just yet.


You could look at a single SIMD as a pipline, but to me, they are not fully featured enough yet. I think we need quad gpus...like how tri-sli with a physics cards runs...but I want that physics card to take a far more promary role than it does currently.

no...no..and no...., the physix better of running on the cpu because lets face it, what the point of > 6 core if we can't use them all, and reserved the GPU power for pure graphic so we can have all the power to maximize the eyecandy like maybe a ray traced games?

we don't even close to photo realistic picture

EDIT : i think what AMD did to cayman was a god think, especially if they can make it modular enough like bulldozer, so they can speed up the development of new GPU and cutting the cost

 

[cadaveca]

no...no..and no...., the physix better of running on the cpu because lets face it, what the point of > 6 core if we can't use them all, and reserved the GPU power for pure graphic so we can have all the power to maximize the eyecandy like maybe a ray traced games?

we don't even close to photo realistic picture

Yes, I kinda agree with you, but I think that physics calculations are much better run on the gpu, due to the type of work done. I do not think a cpu has enough cores or bandwidth to do the type of physics I want to see in games. And I think the gpu manufacturer's agree with me.

6-8-12core cpus are kinda useless, to me. "Fusion" chips, to me, are a far different beast, though.

 

[wahdangun]

Yes, I kinda agree with you, but I think that physics calculations are much better run on the gpu, due to the type of work done. I do not think a cpu has enough cores or bandwidth to do the type of physics I want to see in games. And I think the gpu manufacturer's agree with me.

6-8-12core cpus are kinda useless, to me. "Fusion" chips, to me, are a far different beast, though.

hmm i think we don't know that for sure, take a look at BF:BC2 its have more stunning phisyx a fully destructible enviorment and yet its just run on CPU, imagine if the game can fully support 6 core CPU it will be more stunning and it will be easier for developer to code for it (maybe intel will push the havoc and make it support more than one core ) and not use X87 instruction crap like nvdia use for their CPU phisyx

and i fully agre on fusion chip, it will be great if AMD can make the GPU on it to use for phisyx, but there is the problem especially the different architecture from intel, it will be hard for developer to code it to be compatible for SB

 

[cadaveca]

I've put well over 200 hours into Bad Company 2. The physics, while visually appealling, are far from realistic, nor is anything actually "Fully Destrucible".

Every animation is pre-baked, and you can look at everthing in the game as a "player" with hitpoints, that when reduced, initiate the pre-decided animation for destruction.

For example, for the environments to be "fully destrucible", buildings would not always fall in the same way, walls would not break in the exact same place, and the game would be near impossible to play online.

It's not even impressive, to me. I've seen more realistic physics in many other apps.

 

[wahdangun]

I've put well over 200 hours into Bad Company 2. The physics, while visually appealling, are far from realistic, nor is anything actually "Fully Destrucible".

Every animation is pre-baked, and you can look at everthing in the game as a "player" with hitpoints, that when reduced, initiate the pre-decided animation for destruction.

For example, for the environments to be "fully destrucible", buildings would not always fall in the same way, walls would not break in the exact same place, and the game would be near impossible to play online.

It's not even impressive, to me. I've seen more realistic physics in many other apps.

yeah i realize that, but the point is, it was integrated into the gameplay, and i'm sick seeing lame physix effect on metro, Mirror edge, or even batman, what the point of physix if we just seeing lame cloth or smoke effect.

and btw its not hard to implement it online, i think its just need more data to be uploaded and sync the game condition every second

 

[CDdude55]

I've put well over 200 hours into Bad Company 2. The physics, while visually appealling, are far from realistic, nor is anything actually "Fully Destrucible".

Every animation is pre-baked, and you can look at everthing in the game as a "player" with hitpoints, that when reduced, initiate the pre-decided animation for destruction.

For example, for the environments to be "fully destrucible", buildings would not always fall in the same way, walls would not break in the exact same place, and the game would be near impossible to play online.

It's not even impressive, to me. I've seen more realistic physics in many other apps.

You're really just arguing that the level of Physics you want to see isn't yet implementing in games yet.

Physics rendering on the GPU should help get to that point of realistic physics, but as time goes on better CPU's with more cores, speed, cache etc will come into the light making it possible for developers have more room to work to create the kind of physics we'd love to see in game. I would think GPU physics would bog down overall rendering, as the GPU is the most important factor in gaming.

 

[cadaveca]

The way I see it, there is simply NOT ENOUGH CORES in a traditional cpu design to do realistic physics in real-time. Sure, we don't see it in games now..because it's not possible, at all.

I've been playing GT5 the last couple of days, and even that game, which is so focused on how the car's physics is simulated, does a poor job of it. IT even seems that the yhave focused only on how the wheels interface wit hthe road, but not with how the car reacts to teh entire environment, including wind, dirt, other cars, walls...


That doesn't mean I'm not enjoying it. It's just that what I want is really decades beyond what we have now.

 

[CDdude55]

That doesn't mean I'm not enjoying it. It's just that what I want is really decades beyond what we have now.

So do the rest of us, in due time it should happen if developers really push for it and see it as an important part of the gaming experience.

I agree that right now the CPU's aren't nearly as powerful enough to push the kind of physics of the future we're talking about and while processing everything else at a decent rate.

 

[wahdangun]

The way I see it, there is simply NOT ENOUGH CORES in a traditional cpu design to do realistic physics in real-time. Sure, we don't see it in games now..because it's not possible, at all.

I've been playing GT5 the last couple of days, and even that game, which is so focused on how the car's physics is simulated, does a poor job of it. IT even seems that the yhave focused only on how the wheels interface wit hthe road, but not with how the car reacts to teh entire environment, including wind, dirt, other cars, walls...


That doesn't mean I'm not enjoying it. It's just that what I want is really decades beyond what we have now.

can you proof that ?
even right now there are no games that can use more than quad core except BF:BC2

and i think there are not enough time to develop a physix that can 100% mimic real world, even the GT 5 took nearly 4 years to make the best racing sim, so if they want to add all the environment then PD need at least 10 years to develop it.

 

[CDdude55]

can you proof that ?
even right now there are no games that can use more than quad core except BF:BC2

and i think there are not enough time to develop a physix that can 100% mimic real world, even the GT 5 took nearly 4 years to make the best racing sim, so if they want to add all the environment then PD need at least 10 years to develop it.

No gaming physics will be able to mimic the real world 100%, but the more you try and get close to it the more it will require more processing power, something that 4 or 6 cores can't currently do, and if games get more multi-threaded then what room would be left for physics currently?, you'd need a CPU with many more cores, faster speed, more cache etc. to get everything done on time.

 

[cadaveca]

can you proof that ?
even right now there are no games that can use more than quad core except BF:BC2

and i think there are not enough time to develop a physix that can 100% mimic real world, even the GT 5 took nearly 4 years to make the best racing sim, so if they want to add all the environment then PD need at least 10 years to develop it.

As CDDude says, yes, I think I can prove that. If you want hard numbers as to the math power required for accurate real-time physics, take a look at any distributed computing program. F@H runs simulations of protein folding, and to render the physics required for the few bits of a second it takes for this folding to happen, requires many hours on any of today's cpus.

 

[wahdangun]

i know that in raw power GPU wins hands dow, but remember in game the GPU must also render the game and if game is quite demanding like crysis or metro then GPU will runs out the resource for physix

and take a look at this conclusion from tomshardware :

CPU-Based PhysX summary

To summarize the headlines of the last few months and summarize the test results, we can conclude the following:
The CPU-based PhysX mode mostly uses only the older x87 instruction set instead of SSE2.
Testing other compilations in the Bullet benchmark shows only a maximum performance increase of 10% to 20% when using SSE2.
The optimization performance gains would thus only be marginal in a purely single-core application.
Contrary to many reports, CPU-based PhysX supports multi-threading.
There are scenarios in which PhysX is better on the CPU than the GPU.
A game like Metro 2033 shows that CPU-based PhysX could be quite competitive.


Then why is the performance picture so dreary right now?
With CPU-based PhysX, the game developers are largely responsible for fixing thread allocation and management, while GPU-based PhysX handles this automatically.
This is a time and money issue for the game developers.
The current situation is also architected to help promote GPU-based PhysX over CPU-based PhysX.
With SSE2 optimizations and good threading management for the CPU, modern quad-core processors would be highly competitive compared to GPU PhysX. Predictably, Nvidia’s interest in this is lackluster.

so if the developer want to utilize better multi core support and nvdia use better instruction set like SSE 2 - 4 and not use X87 crap, maybe it will make the developer to use physix more into the game

 

[Benetanegia]

i know that in raw power GPU wins hands dow, but remember in game the GPU must also render the game and if game is quite demanding like crysis or metro then GPU will runs out the resource for physix

and take a look at this conclusion from tomshardware :


so if the developer want to utilize better multi core support and nvdia use better instruction set like SSE 2 - 4 and not use X87 crap, maybe it will make the developer to use physix more into the game

You should read better into that article. In Metro 2033 all 6 cores of the test bed are being maxed out, and the CPU is clocked at 4 Ghz. It's still far behind the GPU and Metro does not have a lot of physics goign on anyway. Difinately not compared to the level of physics that Cadaveca is talking about, which I guess is the same I want. If you add SSE2 you'd get 10-20% more performance which would change nothing.

Physics are better run on the GPU because that kind of task is much better run on a lot of cores with superfast local memory and that's what a GPU is. You say you want your 6/12/24 cores being used, but as I see it that's your problem, maybe you should start thinking about spending less on the CPU and more on the GPU as that is the future. Future battles will occur between integrated GPU (Fusion and fusion-like architectures) and discreet GPU and not GPU vs CPU, because CPU as we know it will move to a secondary/auxiliary role. There's so very little tasks that require big conventional CPUs (multi-core) even today, that cannot be done much better on the GPU...

Bottom line is, we need fast CPUs, but current CPUs are not fast per se, the actual cores are only barely faster than what 2005 CPUs were, we just have 4/6/12 of them crammed together and that's not fast, that's parallel, and if a task is parallel enough to be adecuate to max out a current 6 core CPU, it's most probably adecuate for GPU computing too and 90% of the times the GPU will be orders of magnitude faster, more power efficient and significantly cheaper to produce.

 

[20mmrain]

Nevermind answered my own question

 

[motasim]

You should read better into that article. In Metro 2033 all 6 cores of the test bed are being maxed out, and the CPU is clocked at 4 Ghz. It's still far behind the GPU and Metro does not have a lot of physics goign on anyway. Difinately not compared to the level of physics that Cadaveca is talking about, which I guess is the same I want. If you add SSE2 you'd get 10-20% more performance which would change nothing.

Physics are better run on the GPU because that kind of task is much better run on a lot of cores with superfast local memory and that's what a GPU is. You say you want your 6/12/24 cores being used, but as I see it that's your problem, maybe you should start thinking about spending less on the CPU and more on the GPU as that is the future. Future battles will occur between integrated GPU (Fusion and fusion-like architectures) and discreet GPU and not GPU vs CPU, because CPU as we know it will move to a secondary/auxiliary role. There's so very little tasks that require big conventional CPUs (multi-core) even today, that cannot be done much better on the GPU...

Bottom line is, we need fast CPUs, but current CPUs are not fast per se, the actual cores are only barely faster than what 2005 CPUs were, we just have 4/6/12 of them crammed together and that's not fast, that's parallel, and if a task is parallel enough to be adecuate to max out a current 6 core CPU, it's most probably adecuate for GPU computing too and 90% of the times the GPU will be orders of magnitude faster, more power efficient and significantly cheaper to produce.

... maybe the solution is a high-end APU? ...

 

[pantherx12]

hmm i think we don't know that for sure, take a look at BF:BC2 its have more stunning phisyx a fully destructible enviorment and yet its just run on CPU, imagine if the game can fully support 6 core CPU it will be more stunning and it will be easier for developer to code for it (maybe intel will push the havoc and make it support more than one core ) and not use X87 instruction crap like nvdia use for their CPU phisyx

and i fully agre on fusion chip, it will be great if AMD can make the GPU on it to use for phisyx, but there is the problem especially the different architecture from intel, it will be hard for developer to code it to be compatible for SB

Stunning my foot, go to snowy level where the tank comes are starts blowing up buildings/your cover, just ignore the tank and keep grabbing grenades for launcher and take out I think it's 6 walls of a building, watch all the tiles magically vanish and the building break into the same parts every-time with only actually a few "bits" to render ( individual objects)

In my opinion that's cheap and crappy and is not physics.


By the way the reason people want to use gpus for phsics is because the amount of cores they have, for example 1920 ( supposedly) in thee 6970 vs the 2 cores in most peoples computers , whilst not as powerful or as adaptable as a cpu having that many of them is really beneficial put it this way.

A cpu could realistic simulate a small structure collapsing with maby 8 "chunks" quite nicely, although there would still be some hiccups.

The gpu could render individual bricks falling in the same render time ( literally thousands and thousands and thousands of falling objects behaving realistically)
And hell if it's not using it's it resources then it can actually use other cores to speed it up or to double check the calculations so the simulation is perfect.

Very simplified but get the idea now?

gpus for the win!



I too want full on physics in games, where as many materials/particles are simulated at once ( billions and billions if not more) combined with procedurally(not sure if that's right use) generated animations so you could shoot at one brick at a wall with a big gun and that brick by it's self would smash in you could play jenga with the wall taking out one by one til it actually falls. or you start slicing up a guy and leave him with wounds that actually reflect what you had just done.

 

[the54thvoid]

I wont be happy with physics until you can pick up individual pebbles on a beach in a game or even better, watch grains of sand trickle through your fingers. Just thought i'd push the boundaries there. Such simple things that our £2000/$3000 rigs can't come close to mimicking.

I'll be drawing my pension before i see those days.

 

[HalfAHertz]

Actually one reason why GPU physics sounds more attractive to me is because of the lower latencies you'd get. If all the calculations don't have to leave the GPU, then you wouldn't have to worry about slowdowns related to the CPU.
E.g. moving them around from the main memory to the CPU cache, calculating them, then sending them to the GPU vmem, then adding the effects to the render pipeline, etc.

Pretty much just like when Nvidia first introduced T&L on the GPU and we got a nice boost in performance.

To be honest tho, I don't need phyics. I've done just fine without it for the last 15 years. What I'd like to see is higher dpi monitors and GPUs powerful enough to feed them

 

[the54thvoid]

Two weeks till NDA.....

Two weeks of not buying gfx products...

I have a growing idea that the 6970 might well exceed the 580.

I know people are allowed views and that is mine but my basis is centered around the marketing of the 580 release.

1) No pomp and ceremony - NV just released the fixed Fermi. No bragging, no massive campaign.
2) It's in limited* (very in UK) quantites and pricing is very varied - hints at low volume - hints at rushed out to jump the gun on the 6970.

Those two things alone suggest a marketing ploy to steal share 'while you can'. Nought wrong with it either but i think it points to NV fear over the potential of the 6970.

Now before the red and green go to war on this statement, it's my opinion, thats all and when push comes to shove - I'll buy a 6970, 580 or a custom designed 480 - whichever suits my quiet & powerful gaming pc best.

Roll on two weeks.

*Edit: 29/11 (11/29 for US) Stock over all these e-tailers: Overclockers UK (ZERO), Scan (1 Model) , Ebuyer (16 in stock), Novatech (6 in stock), Aria (ZERO), CCL (ZERO), Microdirect (1 Brand -KFA), Dabs (ZERO)

 

[Goodman]

Two weeks till NDA.....

Two weeks of not buying gfx products...

I have a growing idea that the 6970 might well exceed the 580

In performance?
I think not , what i think is that the 6970 will be more or less equal to the 580 & most likely cheaper to buy , the AMD 6990 (dual GPU) will definitely beat the crap out of the 580 & 5970 but wont be available before January or February 2011

Only what i think & no way base on facts

 

[finndrummer]

I think not , what i think is that the 6970 will be more or less equal to the 580 & most likely cheaper to buy

+1

the AMD 6990 (dual GPU) will definitely beat the crap out of the 580

You can't compare a dual gpu card with a single gpu one. And from rumors, nvidia is preparing a dual card based on the GTX 570.