Is PS4/XONE CPU a bottleneck?

[Deleted member 11852]

So any thoughts on a real world performance multiplier for the Jaguar at 1.6 Ghz over Xenon in the 360 given OoO, higher IPC and the other advantages?

Sebbbi has posted on this.

 

[Globalisateur]

So any thoughts on a real world performance multiplier for the Jaguar at 1.6 Ghz over Xenon in the 360 given OoO, higher IPC and the other advantages?

We know the new consoles have 10x as much memory (more if you consider OS allocations) and the PS4 has 10x the GPU performance so it'll be interesting to see how the CPU's stack up.

If we judge by this Ubisoft benchmark which was then only using 6 cores of PS4 jaguar, it's 2.88 (98 / 34).

 

[(((interference)))]

But it's slower than the PS3 CPU there, so it's most likely going to apply to that particular cloth sim case, where you can optimize heavily for the SPUs rather than across the board workload for, say, an FPS game where the code will be branchier.

But maybe it is only a 2-3x real world advantage over properly optimised Xenon code, and even less for properly optimised Cell code.

 

[Globalisateur]

But it's slower than the PS3 CPU there, so it's most likely going to apply to that particular cloth sim case, where you can optimize heavily for the SPUs rather than across the board workload for, say, an FPS game where the code will be branchier.

The PS3 SPEs had to process a lot of graphical tasks in real games (like post AA, because its GPU was kinda weak), so it unfair to compare PS3 CPU and PS4 CPU in this benchmark. But the comparison between X360 and PS4 CPUs is fair here, both having a quite powerful GPU and similar CPU / GPU architecture (compared to PS3 PPE / SPEs / GPU odd architecture obviously).

And as far as I know that's the only benchmark we have...

 

[(((interference)))]

Yes, I understand that.

But seeing as the Cell is beating Jaguar here, we can't take that to mean that moving to Jaguar meant CPU perf hasn't changed from Cell.
As in real world code (not just for a particular cloth sim) you would be able to get higher FPS on traditional CPU gaming workloads with Jaguar than with Cell? Assuming properly optimised code for both.

I.e. so assuming Cell doesn't have to help out RSX with graphics tasks, would it be more or less performant than Jaguar running a typical game (e.g. a FPS/TPS) workload.

 

[Arwin]

The PS4 is balanced so that the CUs take over the role of the SPEs. So you can't really see the performance of the CPUs in isolation in that respect. For straight ports of PC titles though, I wouldn't be surprised if the CPU is the biggest bottleneck and needs the most attention.

 

[Shifty Geezer]

I'm guessing not more than +$20-25 MSRP, probably less.

Moving from an APU to discrete CPU and GPU without HSA will have more costs than just the silicon though. I doubt anyone could even get close to gauging the costs of switching architectures other than the engineers on these projects.

 

[Ika]

Moving from an APU to discrete CPU and GPU without HSA will have more costs than just the silicon though.

I have to agree.

 

[Nisaaru]

Power8 with its CAPI port would have (theoretically) allowed HSA in a discrete CPU/GPU design like in these SOCs.