PlayStation 4 (codename Orbis) technical hardware investigation (news and rumours)

[astrograd]

We can even go so far to imagine a realistic scenerio where this was demonstrated. "Here's a scene rendered using 14 CUs. Now we add another 4 CUs to the rendering. Notice the improvement? No, we don't much either. So how's about we use those CUs to do non-graphics stuff like this...(effect showing something amazing like, I dunno, fluid dynamic volumetric smoke interacting with the environment (which of course should count as graphics!)). Better? We think so too. Of course, if you want to spend those extra 4 CUs on graphics, that's fine, but I hope you feel an incentive to use them for something else with a more dramatic impact on the game."

I consider that explanation as uniting all the evidence and soundbites, so it's the one I'm most comfortable with.

That's what I was saying as well. Ppl are way too touchy here on this issue. Ha! I wonder if they will need to use CU's for processing game audio. Sounds like their audio DSP only does compression/decompression. Maybe use CU's for synthesizing, mixing, filtering, etc instead of using the CPU for that stuff?

Sadly they weren't at all clear. He used loose language with numerous connotations (what does "round" mean? Certainly not the geometric arrangement of the hardware, making it completely unclear) and that's why people are reading it different ways. Cerny provided an ink-blot of understanding, and people are seeing it in different ways, which is to be expected.

I think he simply means that there are nuances to fully exploiting it that will take time to for devs to get used to working with.

 

[Solarus]

That's what I was saying as well. Ppl are way too touchy here on this issue. Ha! I wonder if they will need to use CU's for processing game audio. Sounds like their audio DSP only does compression/decompression. Maybe use CU's for synthesizing, mixing, filtering, etc instead of using the CPU for that stuff?



I think he simply means that there are nuances to fully exploiting it that will take time to for devs to get used to working with.

english isn't my first language but i saw the thread on txb when you said that. in the quote you showed by cerny he said its "principle" role is that compression stuff. principle (principio) doesn't equate to "only" in spanish, im assuming its different in english and can equate to only. ugh english makes no sense at all sometimes. i think rules in english are made up as you go along haha.

 

[Betanumerical]

english isn't my first language but i saw the thread on txb when you said that. in the quote you showed by cerny he said its "principle" role is that compression stuff. principle (principio) doesn't equate to "only" in spanish, im assuming its different in english and can equate to only. ugh english makes no sense at all sometimes. i think rules in english are made up as you go along haha.

Principle doesn't mean only in english either.

That's what I was saying as well. Ppl are way too touchy here on this issue. Ha! I wonder if they will need to use CU's for processing game audio. Sounds like their audio DSP only does compression/decompression. Maybe use CU's for synthesizing, mixing, filtering, etc instead of using the CPU for that stuff?

you could probably get away with 1 core for the majority of game audio, 1 out of 7, no need to use the GPU.

 

[astrograd]

This is ridiculous. As others have already said, the only bottleneck for more than 14 CU's is the code that you write. Sure you can code a game that perfectly balances 14CU's with 8 Jaguar cores and 176GB/s bandwidth but you can just as easily do the same with 18CU's, 8 Jaguar cores and 176GB/s bandwidth.

No, you're misunderstanding what I meant by handicap. I don't mean the hardware, I meant the output visuals. Clearly Cerny feels you don't get a proportional visual payoff to using 18 vs 14 for rendering.

 

[Betanumerical]

No, you're misunderstanding what I meant by handicap. I don't mean the hardware, I meant the output visuals. Clearly Cerny feels you don't get a proportional visual payoff to using 18 vs 14 for rendering.

This makes no sense, the proportional visual pay off for using a specific amount of CU's is completely up the developer and is in way fixed, this isnt the SM1 era.

 

[Solarus]

thank you beta i wasn't sure, and if cerny did say its principle role is the compression stuff, why are so many of people here saying it's only role is the compression stuff then saying that cerny said that, then linking to him not saying that? i cant speak for anyone but thats like twisting a statement to fit another wise not correct argument?

 

[astrograd]

Principle doesn't mean only in english either.



you could probably get away with 1 core for the majority of game audio, 1 out of 7, no need to use the GPU.

1 out of 6. Not 7. The OS reserves 2 of the 8 for itself.




Solarus, it's true that principle doesn't mean 'only' in English as well. However, Cerny has been asked about it a couple different times and thus far he has simply said it was designed principally for decompression/compression. If it was particularly suitable for other stuff in a meaningful fashion he would have surely mentioned it.

"For example, by having the hardware dedicated unit for audio, that means we can support audio chat without the games needing to dedicate any significant resources to them. The same thing for compression and decompression of video." The audio unit also handles decompression of "a very large number" of MP3 streams for in-game audio, Cerny added.

Digital Foundry: Is there dedicated audio processing hardware within the PlayStation 4? What can it do?

Mark Cerny: There's dedicated audio hardware. The principal thing that it does is that it compresses and decompresses audio streams, various formats. So some of that is for the games - you'll have many, many audio streams in MP3 or another format and the hardware will take care of that for you. Or, on the system side for example, audio chat - the compression and decompression of that.

 

[Solarus]

but isnt that still just assumption on your part? your saying because he didnt mention must mean it doesnt do it. and because he didnt say it then it only does compression/decompression. thats not good logic is it? how many streams do audio chips normally do? maybe he talked about the stream stuff because normally audio chips cant handle many?

 

[astrograd]

but isnt that still just assumption on your part? your saying because he didnt mention must mean it doesnt do it. and because he didnt say it then it only does compression/decompression. thats not good logic is it? how many streams do audio chips normally do? maybe he talked about the stream stuff because normally audio chips cant handle many?

Yes, it's an assumption. It's a rather well informed one though. Note that he was specifically asked by a technical website "what can it do?". His response was to say it principally handles compression/decompression. A you can see, he gave the same response to GI.biz in their interview with him where he was tasked with outlining the functionality of the PS4's additional hardware (first quote I posted there).

We don't need rigorous logic to derive reasonable conclusions. We just can't be certain about them, but they can still be reasonable.

 

[Solarus]

i see, i guess its because i don't understand all this tech talk. but to me i don't see how you guys can come to the conlusion that it only does compression stuff. it still seams to me that arugment has holes in it.

 

[astrograd]

This makes no sense, the proportional visual pay off for using a specific amount of CU's is completely up the developer and is in way fixed, this isnt the SM1 era.

Cerny and Sony's dev doc (as per VGleaks) strongly suggest that if you were going only for graphics performance you wouldn't want to spend all 18 there. The dev documentation supposedly says it is "balanced" at 14 with 4 being the extra ALU resources Cerny has mentioned several times for compute purposes.

 

[warb]

thank you beta i wasn't sure, and if cerny did say its principle role is the compression stuff, why are so many of people here saying it's only role is the compression stuff then saying that cerny said that, then linking to him not saying that? i cant speak for anyone but thats like twisting a statement to fit another wise not correct argument?

Because of the comparison with other audio hardware to which this is not the principle function.

 

[Betanumerical]

Cerny and Sony's dev doc (as per VGleaks) strongly suggest that if you were going only for graphics performance you wouldn't want to spend all 18 there. The dev documentation supposedly says it is "balanced" at 14 with 4 being the extra ALU resources Cerny has mentioned several times for compute purposes.

The VGLeaks documentation was for a example on using both GPGPU and graphics concurrently on the same system, there is nothing to suggest that is required or that it is not worthwhile to go with 18CU's for graphics.

This has been completely debunked more then once. There is nothing stopping you from using all 18CU's for graphics and obviously depending on what your doing with them getting a decent result. There is also nothing to suggest that a 'balance' exists in the system that somehow makes the other 4CU not worthwhile to graphics.

The only people who keep harping on about only '14 CU's' being used for graphics have a obvious agenda and also do not seem to realise that you can use GPGPU FOR GRAPHICs. So even if the split exists (which it doesn't, as been mentioned multiple times on the last page) it wouldn't even matter as the other ALU's could still contribute towards the VFX of the scene.

From the previous page, where it seems you failed to read the replies to your own godamn post.

This is ridiculous. As others have already said, the only bottleneck for more than 14 CU's is the code that you write. Sure you can code a game that perfectly balances 14CU's with 8 Jaguar cores and 176GB/s bandwidth but you can just as easily do the same with 18CU's, 8 Jaguar cores and 176GB/s bandwidth.

Cerny's angle is clearly that he believes there's a greater return for using ~22% of the GPU's resources on GPGPU than if those resources were used to incrementally increase the graphics by an additional ~29%.

There's no mysterious handicap that prevents all 18 CU's from being efficiently used for graphics. There's just (arguably) more effective work you can put ~22% of your CU's to. The same argument could possibly be made for the XB1 in the same percentages as well.

For fucks sake, no. They are all identical CUs.

No, as Brit mentioned they are all the same.

If a developer wanted to use all 18 CUs for compute, they could do so.

If a developer wanted to use all 18 CUs for graphics related work, they could do so.

But he's suggesting that perhaps the best use is to use some number of them for compute and some number of them for graphics. Which is where I'm going to guess the whole misunderstanding about the 14+4 came from. It was likely just an example he threw out at some meeting as an example of what he meant.

But nothing about the 18 CUs screams out don't use more than 14 for graphics related work or [pick your choice] happens.

Regards,
SB

Bringing back 14+4 again is an act of desperation.
You can see it as some kind of death throes, just let it flow it will end soon.

 

[astrograd]

i see, i guess its because i don't understand all this tech talk. but to me i don't see how you guys can come to the conlusion that it only does compression stuff. it still seams to me that arugment has holes in it.

Sure, any argument premised on an unknown will, but not big holes. Certainly rather small ones compared to other info that is reasonably taken as reliable info. Sure, all of it could be wrong, but it's unlikely. There has been lots of discussion on the topic of compariong PS4's audio chip to X1's and the consensus there seems to be that it's highly unlikely that PS4's audio chip does more than decompression/compression. There's certainly no evidence that I'm aware of to suggest the contrary as of yet.

The thread I referenced is here: http://beyond3d.com/showthread.php?t=63677&page=10

 

[Solarus]

Cerny and Sony's dev doc (as per VGleaks) strongly suggest that if you were going only for graphics performance you wouldn't want to spend all 18 there. The dev documentation supposedly says it is "balanced" at 14 with 4 being the extra ALU resources Cerny has mentioned several times for compute purposes.

on teamxbox you said something about ps4 aces having 4 queues for 14 of the cus and 2 queues for the last 4, does that somehow play a role? im not sure what an alu is, but if its related to aces and queues and stuff wouldn't it make more sense for the last 4 to have more queues than the 14 not less?

Because of the comparison with other audio hardware to which this is not the principle function.

so xbox's audio dsp principle function isnt compression then that means ps4's audio dsp only does compression? i dont see how that can work as an argument

 

[Betanumerical]

on teamxbox you said something about ps4 aces having 4 queues for 14 of the cus and 2 queues for the last 4, does that somehow play a role? im not sure what an alu is, but if its related to aces and queues and stuff wouldn't it make more sense for the last 4 to have more queues than the 14 not less?



so xbox's audio dsp principle function isnt compression then that means ps4's audio dsp only does compression? i dont see how that can work as an argument

Considering theres only 1 graphics pipe in the GCN cards and multiple compute I don't really see how this could in any way shape or form play a role. He is making stuff up at best, again.

 

[astrograd]

on teamxbox you said something about ps4 aces having 4 queues for 14 of the cus and 2 queues for the last 4, does that somehow play a role? im not sure what an alu is, but if its related to aces and queues and stuff wouldn't it make more sense for the last 4 to have more queues than the 14 not less?

There are 64 queues for 18 CU's. I was wondering how that was being split up in the TXB thread. One way to split that is 14 of them get 4 queues and the other 4 get a pair each. That's just playing with numbers though, it wasn't an actual argument I made or anything.

Here's a pic that others here can try explaining (likely already discussed in the thread):
http://www.vgleaks.com/wp-content/uploads/2013/02/gpu_queues.jpg

As for the audio stuff, read the link in my previous post (I edited it in for ya). I won't rephrase what was said there already as folks like Shifty did a good job expressing the sentiment I agree with.

 

[Betanumerical]

There are 64 queues for 18 CU's. I was wondering how that was being split up in the TXB thread. One way to split that is 14 of them get 4 queues and the other 4 get a pair each. That's just playing with numbers though, it wasn't an actual argument I made or anything.

As for the audio stuff, read the link in my previous post (I edited it in for ya). I won't rephrase what was said there already as folks like Shifty did a good job expressing the sentiment I agree with.

Its the same odd split for the XBONE, 16 queues for 14 CU's. This leads me to believe that the queues are in some sort of big pool that any CU can take them from.

 

[astrograd]

Its the same odd split for the XBONE, 16 queues for 14 CU's. This leads me to believe that the queues are in some sort of big pool that any CU can take them from.

12 CU's on X1. But yeah, another strange setup. I assume that PS4 diagram has already been thoroughly deciphered since it was posted in February?

 

[Betanumerical]

12 CU's on X1. But yeah, another strange setup. I assume that PS4 diagram has already been thoroughly deciphered since it was posted in February?

Actually change that. Not 16 queues for XBONE nor 64 queues for PS4.

12 CU's / 24 queues (XBONE) [16 Compute + 8 Graphic] 2 Queues / CU
18 CU's / 72 queues (PS4) [64 Compute + 8 Graphic] 4 Queues / CU

That should put the queue argument to rest, and it seems that it is no coincidence that they both line up perfectly for the amount of CU's they have.