Sorry to butt into this amazing round-and-round discussion (last few pages really are going round in circles, boys!), but isn't the point here not really PS4's fabled technical capabilities, but the fact that Sony will support (and finance?) projects that might not use traditional rendering methods, like Dreams? Just like they seem to have supported other seemingly crazy ideas in the past, to various degrees of success?
Could PlayStation 4 breathe new life into Software Based Rendering?
- Thread starter onQ
- Start date
- Status
Just like the PPU handed off work to the SPU's, Jaguar is handing off work to the 8 compute pipelines.
If the PS4 8 compute pipelines isn't MIMD, what is it?
Interesting.
Reading up on it, their concept of MIMD is if you were on a PC and had multiple programs executing on the GPU at the same time. Each thread would be a different application or game. AMD seems to have an SMT architecture, where wavefront is made up of SIMD instructions. You can occupy the GPU with multiple wavefronts, but it's not a MIMD approach.
All of that said, every GCN GPU is essentially equipped the same way, with ACEs and CUs etc. The programming model for all of them is the same, whether it's PS4, Xbox One, or PC. The hardware is exposed with different APIs. In terms of ACEs and wavefronts, there is nothing particularly different about the PS4 GPU.
I guess you could say the hardware is powerful enough to give decent visual results at an affordable price. Xbox One may as well, and maybe PC as well. I'd more likely give a lot of credit to Sony for actually funding a studio to do research on the subject, rather than enabling it through hardware.
The PS4 is 8x8 MIMD ( 64 threads).
ACEs don't execute compute tasks but the CU do.
You use terminology you don't understand. 3dcgi is an AMD employee...
ACE aren't execution units, they are scheduler.. The 18 CU are execution unit.
Ok ...
So is Xbox one 2x8 MIMD (16 threads)? I don't think threads is the correct term to be used here.
What does seem fairly clear is that the execution model for GCN is SMT or SIMT, not MIMD. Each CU has 4 vector SIMD units that share instruction fetch/decode. It does have MIMD features, according to some slides, but that seems to be the case where multiple applications are sharing the GPU, say if you were browsing a website that used GPU resources to render 3D-graphics on a web-page at the same time that you were using a modelling tool or playing a video game. The SMT or SIMT model is multiple wavefronts, each wavefront being a list of SIMD instructions for the vectors units the wavefront is dispatched to. That's how games run on GCN. That's pretty clear from this forum, reading real devs posts. Search wavefront.
They use the terminology MIMD for CU not ACE...
I'm not using it for ACE I'm using it because the PS4 APU can run 8 compute pipelines, I never seen them AMD slides before.
Last edited:
You keep mixing up your points. You said you're not talking about HSA but GPGPU. Then you say PS4's APU is 8x8 MIMD because there's 8 CPU cores feeding 8 ACEs. One core can feed 8 ACEs. They are command buffers. It's all irrelevant because whether PS4's APU is described as MIMD or not, whether it has one core feeding 8 ACEs or eight cores feeding 2 ACEs, it's still not doing anything any other system can't do regards novel rendering techniques.
You restarted this thread to a claim recent evidence proves you right. What specific techniques is it in games like Dreams that are enabled by PS4's architecture and are not possible on XB1 or PC? How does 8 cores (six available to devs) feeding 8 ACEs enable SDF splat based rendering on PS4 but XB1's 6+ cores feeding 2 ACEs can't use that technique?
GCN is all 3 MIMD, SIMD, & SMT
I called the PS4 APU a 8x8 MIMD because the CPU can basically use the GPGPU as a co-processor with the 8x8 command buffers. & yes that's HSA I already told you that the APU is a product of HSA.
Who said that it can't be done on a Xbox One or PC? what does that even have to do with anything? This is another one of your loopholes.
No, it's your thread title and opening question. You talk exclusively about what PS4 brings to software rendering, not modern rendering architectures. And go on to discuss PS4's architecture as being beneficial to software rendering. If PS4's architecture doesn't matter to your ideas, why are you raising PS4 as being 8x8 MIMD? If you accept it makes no odds and XB1 and PC are just as capable at novel rendering solutions thanks to GPU compute, why'd you ever raise PS4's APU architecture?
I repeat, if you just want to know what software rendering solutions will arise this gen, you need to ask the right (platform agnostic) question, probably in the Algorithms forum (although the console ecosystems will generally be more favourable to experimentation thanks to 1st/2nd parties, so there's some merit to basing the discussion in Console Tech).
What is that even in answer to?!?! Are you talking about HSA compute? Is this thread "can HSA compute breathe new life into software rendering"? If so, 1) HSA has nothing to do with what devs are currently achieving with compute (eg. Dreams). 2) What do you mean here...?
In one breath you say it's nothing to do with HSA, and the next you say it's all about the HSA. And then you say it's nothing about HSA again. And then you're talking about PS4 specifically. And then you say you're not excluding other platforms!
onQ:
Let's start over because honestly your terms are mixed to me and I'm still struggling to find your argument beyond trying to explain "how" or "what".
Let me start you off:
GPGPU: general purpose computing completed by the GPU. Traditional GPGPU has code which copies the values from System RAM to graphics RAM and sends code over and the pointers to the data to the GPU to run it against. When the GPU is complete it needs to copy the value back to system ram.
Software Rendering: drawing graphics entirely on the CPU.
HSA: Integrated based architecture in which the CPU and GPU can access the same memory locations. Saves the Step of copying values back and forth for processing. in PC space for graphics this is often one way only. For GPGPU this will save 2 copies, one from DDR to GDDR and back.
Media molecule: looks like graphics algorithms performed entirely by compute shaders.
Okay, I think I nailed down the anchor points. Could you tell us
A) why or what advantages software rendering has over fixed function programmable pipeline ones? There are some if I recall but often not worth the trade off
B) using that advantage that software has over hardware, what aspects of PS4 HSA will help accomplish or exploit these advantages that previously were not possible.
Let's start over because honestly your terms are mixed to me and I'm still struggling to find your argument beyond trying to explain "how" or "what".
Let me start you off:
GPGPU: general purpose computing completed by the GPU. Traditional GPGPU has code which copies the values from System RAM to graphics RAM and sends code over and the pointers to the data to the GPU to run it against. When the GPU is complete it needs to copy the value back to system ram.
Software Rendering: drawing graphics entirely on the CPU.
HSA: Integrated based architecture in which the CPU and GPU can access the same memory locations. Saves the Step of copying values back and forth for processing. in PC space for graphics this is often one way only. For GPGPU this will save 2 copies, one from DDR to GDDR and back.
Media molecule: looks like graphics algorithms performed entirely by compute shaders.
Okay, I think I nailed down the anchor points. Could you tell us
A) why or what advantages software rendering has over fixed function programmable pipeline ones? There are some if I recall but often not worth the trade off
B) using that advantage that software has over hardware, what aspects of PS4 HSA will help accomplish or exploit these advantages that previously were not possible.
- Status
Similar threads
