Sony PlayStation 5 Pro

[Shifty Geezer]

That’s great in theory but that’s certainly not guaranteed these days. Unfortunately the state of modern games doesn’t inspire confidence in the ability of devs to wring every ounce of performance out of these machines.

Maybe not all, but certainly some.

Also shouldn’t we assume that Sony knows what devs want and need? If devs had a use for more FPU power it should be in the console spec.

What devs want and need is infinite processing power and infinite bandwdith. However, that's not possible and compromises need to be made with a finite budget. If more CPU FPU could be added at no cost elsewhere, it'd be in and used. Unlike the article suggesting games don't have a need for crazy maths power (?!), it could and would be used. However, given a finite budget it's important to optimise where you invest the silicon. The extra mm² dedicated to more FPUs would yield lower results overall than that same area spent elsewhere.

 

[fellix]

I was once told that x86 is slow because you can still run 16 bit code.

Pentium Pro in the 1990s had an architectural "bug" that caused slow 16-bit code execution, but otherwise there's no inherit performance penalty related to the ISA implementation.

A dual pipe FPU makes a lot of sense for the PS5, which is meant only for gaming.

Being a gaming console doesn't decrease the need for maths capabilities.

For PS3 emulation the castrated Zen2 cores in PS5 might bottleneck some games that use heavy SPU code. Otherwise most FP math in games is already in the domain of GPU compute anyways. I would rather sacrifice a second SIMD pipeline and get higher turbo boost clocks for faster single-threaded performance, still critical for many titles.

 

[Mobius1aic]

I thought the FPU downgrade (and henceforth SIMD) was kinda stupid TBH. Unless Sony and MS are actually going to leverage GPU compute for meaningful gameplay things, then keeping CPU floating point performance strong is imperative. That was a major point behind the Emotion Engine, Cell and Xenon, even if the first two were partially meant to be vertex engines. Jaguar was a general stagnation (if not regression) only made up by GPGPU potential better general performance and moar [real] cores. Modern gaming is entirely tied to CPU floating point performance.

 

[techuse]

I was once told that x86 is slow because you can still run 16 bit code.

It may be possible that x86 CPUs see such diminishing performance improvements because they have to be designed in a manner which wont break decades of compatibility and compiler support.

 

[trinibwoy]

What devs want and need is infinite processing power and infinite bandwdith. However, that's not possible and compromises need to be made with a finite budget.

Yeah let’s not veer into fantasy land. Clearly every engineering decision is made with finite resources in mind.

Unlike the article suggesting games don't have a need for crazy maths power (?!), it could and would be used.

If in reality games need a lot more scalar math throughput for the workloads that will actually run on the CPU then the author is essentially correct. Unless someone can propose a real workload that benefits from more FP/SIMD performance at the expense of ALUs or other hardware.

 

[Jmesparza05]

Does this mean is a graphical boosts?

 

[Jmesparza05]

How do I post pictures here?

 

[Charlietus]

Does this mean is a graphical boosts?

Tidux doesn't know anything. I'm surprised he is still around.

 

[digitalwanderer]

Does this mean is a graphical boosts?

It's just PR. It doesn't mean anything until we see the actual product in action.

 

[Seanspeed]

Based on?

Common sense? We can look at a 3800X or something and make some entirely reasonable assumptions about its performance were it locked to 3.5Ghz and only had like 8MB of L3 in its split CCX config.

You are the one claiming that it is somehow underperforming compared to how it would on PC and thus need to show some proof or solid reasoning on why you believe that.

 

[Seanspeed]

PS5 Pro in 2025 will run GTA6 at 30fps whereas my Nvidia RTX 5090 ti ultra triple SLI setup with my quad Intel 1600XTs and 1.5TB DDR7 Ram will run it at approximately 0 fps

So I’d say at 599 compared to 37967 dollar : PS5 Pro in my case, is much better value even at 30fps

Is playing GTAVI on day 1 an 'infinitely valuable' experience in your eyes? Cuz I assure you for most others it wont be.

There's a much better way to make the argument you're seemingly trying to make, but I'm not gonna do the work for you.

 

[Frenetic Pony]

People shouldn't forget the PS4 pro launched at the existing PS4 price point and still only accounted for, at best, 1 out of every 5 consoles sold. I think it's likely that number was actually significantly less by the end of its lifecycle. This will be launching at a higher price point than the existing PS5.

I mean that seems guaranteed. The interesting question might be "where is the price point that's interesting for the maximum number of consumers".

A 4070s is currently $549, getting that + literally the rest of the gaming box for the same price would be interesting (digital version) right? Of course it looks like RDNA4 is launching right alongside (AMD execs just started talking openly about RDNA4, can't be far from release). Still, even if AMD's equivalent is $499, $50 for everything else is still cool?

Is playing GTAVI on day 1 an 'infinitely valuable' experience in your eyes? Cuz I assure you for most others it wont be.

You know full well for a few million people it'll be worth the $. The amount of double dipping on GTAV was really high.

 

[trinibwoy]

Common sense? We can look at a 3800X or something and make some entirely reasonable assumptions about its performance were it locked to 3.5Ghz and only had like 8MB of L3 in its split CCX config.

You are the one claiming that it is somehow underperforming compared to how it would on PC and thus need to show some proof or solid reasoning on why you believe that.

You must have missed where I pointed out an 8-core Zen 2 averages about 120fps in windows. Check TPU’s latest CPU reviews. I don’t know how common sense would lead you to believe a hobbled version should struggle to hit 60fps. Quite the opposite actually.

 

[Subtlesnake]

You must have missed where I pointed out an 8-core Zen 2 averages about 120fps in windows. Check TPU’s latest CPU reviews. I don’t know how common sense would lead you to believe a hobbled version should struggle to hit 60fps. Quite the opposite actually.

This discussion is confusing to me. We already have benchmarks from the XSX CPU on PC in the form of the 4800S Desktop Kit and it performs roughly where we would expect.

"Ultimately we end with one or two surprises in our gaming tests - and a disappointment with Metro Exodus - but for the most part, an Xbox Series X CPU in PC form tends to deliver the same class of performance as it does on the Xbox Series X running on its own streamlined version of Windows. Just remember, we do have the advantage of no memory contention with the GPU, an extra core plus 11 percent of core clock. It may be time to put a bit more time into delivering absolute console parity in terms of settings with a few more PC titles, as this could be interesting for further testing - not just with this board but in our GPU reviews too."

Page 4 | AMD 4800S Desktop Kit review: playing PC games on the Xbox Series X CPU

What if you transplant the CPU from the Xbox Series X into a Windows gaming PC? AMD’s 4800S Desktop Kit allows just that - and we've tested it extensively!

www.eurogamer.net

The titles that struggle to hit 60 FPS due to CPU limitations (eg. Starfied, BG3, Dragon's Dogma 2) also don't hit 60 FPS on PC Zen 2 parts. So I don't see where the mystery is.

 

[XboxKING]

Is playing GTAVI on day 1 an 'infinitely valuable' experience in your eyes? Cuz I assure you for most others it wont be.

There's a much better way to make the argument you're seemingly trying to make, but I'm not gonna do the work for you.

if you want to play Microsoft Excel 2022 edition then even a PS5 at 99 dollar would be worthless. Would you agree on that?

 

[Shifty Geezer]

This is a tech thread. Discussion the software value of the platforms is OT.

 

[trinibwoy]

This discussion is confusing to me. We already have benchmarks from the XSX CPU on PC in the form of the 4800S Desktop Kit and it performs roughly where we would expect.

"Ultimately we end with one or two surprises in our gaming tests - and a disappointment with Metro Exodus - but for the most part, an Xbox Series X CPU in PC form tends to deliver the same class of performance as it does on the Xbox Series X running on its own streamlined version of Windows. Just remember, we do have the advantage of no memory contention with the GPU, an extra core plus 11 percent of core clock. It may be time to put a bit more time into delivering absolute console parity in terms of settings with a few more PC titles, as this could be interesting for further testing - not just with this board but in our GPU reviews too."

Page 4 | AMD 4800S Desktop Kit review: playing PC games on the Xbox Series X CPU

What if you transplant the CPU from the Xbox Series X into a Windows gaming PC? AMD’s 4800S Desktop Kit allows just that - and we've tested it extensively!

www.eurogamer.net

The titles that struggle to hit 60 FPS due to CPU limitations (eg. Starfied, BG3, Dragon's Dogma 2) also don't hit 60 FPS on PC Zen 2 parts. So I don't see where the mystery is.

Exactly. We should expect console Zen 2 to perform similarly (or better) than a similarly configured Zen 2 CPU on windows. So if mainstream Zen 2 on windows is averaging ~120 fps across a variety of games we should expect even a cut down version on consoles to average north of 60 fps.

The edge cases where PC Zen 2 struggles to hit 60fps also underperform on other architectures. Starfield and Dragons Dogma don’t refute the above hypothesis because they also struggle on much faster CPUs. They are edge cases where lack of software optimization is a likely culprit and shouldn’t dictate our baseline expectation.

i.e. the mantra that we shouldn’t expect 60fps from consoles doesn’t make sense to me.

 

[Subtlesnake]

Exactly. We should expect console Zen 2 to perform similarly (or better) than a similarly configured Zen 2 CPU on windows. So if mainstream Zen 2 on windows is averaging ~120 fps across a variety of games we should expect even a cut down version on consoles to average north of 60 fps.

The edge cases where PC Zen 2 struggles to hit 60fps also underperform on other architectures. Starfield and Dragons Dogma don’t refute the above hypothesis because they also struggle on much faster CPUs. They are edge cases where lack of software optimization is a likely culprit and shouldn’t dictate our baseline expectation.

i.e. the mantra that we shouldn’t expect 60fps from consoles doesn’t make sense to me.

Your original comment which started this discussion stated that: "It’s wild to me that an 8-core 3.5Ghz Zen 2 struggles to hit 60fps in any game". If your point is really about average performance, then how do we know that Zen 2 on the consoles isn't averaging north of 60 FPS? Since most 60 FPS modes seem to be GPU limited.

It's hard to make comparisons with the PC when most console games ship with locked framerates. Even if there are dips below 60 FPS due to the CPU, those same dips may occur on PC as well with Zen 2 CPUs. (Since TPU isn't averaging minimum FPS).

 

[trinibwoy]

Your original comment which started this discussion stated that: "It’s wild to me that an 8-core 3.5Ghz Zen 2 struggles to hit 60fps in any game". If your point is really about average performance, then how do we know that Zen 2 on the consoles isn't averaging north of 60 FPS? Since most 60 FPS modes seem to be GPU limited.

My comment was in response to Dictator stating that the PS5 pro won’t fare much better on titles that are struggling to hit 60fps on the PS5.

If the number of titles with that problem is low then it’s not really a problem for the PS5 pro either.

 

[Subtlesnake]

My comment was in response to Dictator stating that the PS5 pro won’t fare much better on titles that are struggling to hit 60fps on the PS5.

If the number of titles with that problem is low then it’s not really a problem for the PS5 pro either.

Ah, ok. It seems we made the same point then.