On top of the PS5 being 4-5 years old by the time this product comes out and having newer technology powering it.Despite it being from MLiD.
Why can't this be more powerful than the PS5?
16 Zen5 cores and 40 CU RDNA3.5 chiplet APU surly should be more powerful than the 8 Zen2 cores and 36 CU RDNA2 PS5.
New AMD CPU’s integrated graphics reportedly more powerful than PS5
That CPU is way better than PS5.
Bojji
Member
I don't know how ps5 operates in this aspects but wouldn't it be super inefficient if it dropped to half the speed? Cpu doesn't really need more 40GB/s (it can't) to operate so why 1/2 448GB/s BW for that?.
Xbox has separated memory pools and it performs more or less the same most of the time.
rodrigolfp
Haptic Gamepads 4 Life
Nvidia doesn't have a x86 APU.
Loxus
Member
I don't believe bandwidth is an issue.
The 7600 has a 128-bit interface with 32 MB Infinity Cache and 288 GB/s bandwidth for an effective bandwidth of 477 GB/s.
If I'm not mistaken 256-bit LPDDR5X @ 8.5 Gbps gives 272 GB/s.
Strix Halo is rumored to have 32 MB of Infinity Cache as well, so the effective bandwidth should be in the 7600 range.
playsaves3
Banned
Also shows how much room there is for a ps5 pro to improveDespite it being from MLiD.
Why can't this be more powerful than the PS5?
16 Zen5 cores and 40 CU RDNA3.5 chiplet APU surly should be more powerful than the 8 Zen2 cores and 36 CU RDNA2 PS5.
playsaves3
Banned
Hopefully the ps5 pro fixes it
Bojji
Member
If it has that memory BW it has to be shared with CPU and Zen 5 will for sure take more than Zen 2. Other thing is power consumption, 120W for the whole thing when 7600 alone has 165W TDP.
Last edited:
SportsFan581
Member
It's just the nature of how memory is configured. It's a stripped array across the chips. The typical 64bit controllers that AMD uses support a single data stream at full speed or two data streams at half speed.
Storage is much the same way. The feeling that you can write and read multiple files at once is just an illusion on standard setups, everything really happens one bit at a time. It just all happens so fast and different jobs can be intertwined, giving the impression of multiple things being done at once.
Last edited:
SportsFan581
Member
This game doesn't work the CPU too hard, I doubt that is the primary bottleneck on this one. Plus, these results can be replicated in probably a good 90% of games if you are actually comparing apples to apples. Many times people just pick a general setting preset and resolution and try to compare that way, which will often result in the PC working harder on each frame. If you really go setting by setting and choose the options that are going to give you the same final image, the 7600 and 4060 are largely comparable to the XSX and PS5. There are some outliers of course.
PS5 and XSX are still great values though since you get everything.
Last edited:
Senua
Member
Is ND still working to bring performance improvements or have they stopped now? It's a shit ton better but it's still too heavyIn theory 7600 is close to PS5 GPU in power but it has 288.0 GB/s BW vs. 448 on PS5. 4060 has even less so it looks like Remedy incompetence here.
PC version can run just like on console hardware or better or worse depending on how good optimization is on console version AND pc version. TLOU1 requires much more powerful GPU to match PS5 version in performance when usually 2070S is not far off in raw power (now we can assume that PS5 is more like 2080 in raw raster power after 3+ years of testing):
256-bit requires octo-channel on PC, which doesn't exist in consumer motherboards. AMD would need to launch a whole new platform for the APU. Or, it's in a console of some kind.
Bojji
Member
Fully patched game is still behind where it should be, 4070S is just 36% better than PS5 here:
While in most games it's ~2x
Last edited:
SportsFan581
Member
Agreed. Probably not a socketed processor but something that will be embedded on custom designed boards for OEM laptops/PCs and/or gaming devices.
Bry0
Member
Yes it's for laptops.
SportsFan581
Member
HP and Dell are known to use laptop chips in some of their budget PCs. We might see it in desktops also if it is something that works well and can be sold as a low-end gaming box.
Might be interesting as a mini PC type deal, depending on how small the system can be and still keep the thing cooled.
How MLID isn't a banned source of information from PC websites is crazy to me. This jackass has been wrong on nearly everything for like 3 years now. Remember how he said AM5 was going to destroy Intel? Only for AM5 to be mediocre at best (until 7800x3D came out) and Raptor Lake almost completely outperformed it? Or how Intel ARC was going to be dead by Celestial? Only for Intel recently to straight out say Battlemage is coming and Celestial is already starting early engineering? Or how RDNA3 GPUs were going to spell big trouble for Nvidia? Only for the 4090 to outperform everything else by an insane margin?
Sethbacca
Member
I wouldn't be surprised if we get a MiniITX board from a Minisforum or similar. They've kind of started moving into that space in the last few months.
Last edited:
TheThreadsThatBindUs
Member
TheRedRiders
Member
Let's not pretend like variability in PC hardware doesn't play a major role in this. A great example of this was the use of Mesh/Primtive Shaders on Avatar FOP, the consoles benefited from this thanks to next-gen exclusivity but not the PC version and the developers cited variability and lack of hardware support as the main reason. The game does run well on PC, but it would have run much better had it leveraged these next gen features.
Last edited:
twilo99
Member
Valid point.
playsaves3
Banned
It would be even worse if they used a ps5 equivalent cpu in that comparison and not one strongerFully patched game is still behind where it should be, 4070S is just 36% better than PS5 here:
While in most games it's ~2x
Bojji
Member
In all those moments PS5 is GPU limited so it's irrelevant.
playsaves3
Banned
Not in the performance mode
That tells you more about their PS5 optimization than it does anything else.Fully patched game is still behind where it should be, 4070S is just 36% better than PS5 here:
While in most games it's ~2x
Zathalus
Member
No, it's clearly a subpar PC port. The game engine was built with zero consideration in porting it to PC eventually and it shows. The PC version doesn't even use async compute for example.
Bojji
Member
It doesn't work like that, "console optimization" was a myth from the beginning. With dx12 and Vulcan there shouldn't be any difference between consoles and pc other than stuff I/O related in case of ps5 (thanks to dedicated hardware it doesn't hammer cpu that much with streaming in game like spider man for example).
They used quality modes.
Last edited:
smbu2000
Member
Does nvidia have a competing product? The one in the switch? Possibly switch 2?
Otherwise the only competition is from intel with their integrated arc graphics.
Think about what you're saying here, you're saying there is no difference between a console and PC optimisation yet you're saying the PC port is bad because the difference between PC and console fps isn't as large as other games. So why can't it work the other way? If there was no difference then wouldn't the delta always be proportional?
What I'm saying is 50fps@4k for a game that looks as good as it does is in fact good performance. it running at a higher 36fps on PS5 than a plagues tale doesn't take away from that, it shows that they possibly optimised the PS5 version of TLOU better than a plagues tale. The PS5 version of a Plagues tale running at 35fps@1440p and not really beating it in graphics isn't necessarily showing a sign of better PC performance.
Last edited:
Bojji
Member
What I should have said is: pc developers have ABILITY to make games run just like on consoles without much API overhead.
Its all on developers in the end and games can (and are) unoptimized. Most pc games run as they should but there are always few bad releases that should perform better.
In this comparison you have:
Aw2 - 2x performance
Plague tale - 2x performance
Tlous - only 36% better
What game is the outlier? This is first naughty dog pc game so I guess it can be excused. It doesn't show magic optimization on PS5 but lack of optimization on pc.
In the end ps5 is just 10TF RDNA2 GPU and should perform like that, there is no magic attached to it other than decompression hardware.
My point is that it's an outlier in the sample of 3 because it's the PS first party game and it's been optimised better on PS5. That it running at 4k36fps is in fact more impressive than A plagues tale running at 1440p35fps.
Bojji
Member
Out of those 3 games TLoU1 was a shit port from the beginning and had many problems that were or weren't fixed in several patches, you can't use it to draw any fair comparisons. It can be greatly optimized for PS5 but it certainly isn't (still) for PC.
It was rough at launch but it certainly isn't now and not based on its performance from those screens which you're using.
4k50fps is still more impressive in terms of performance and with DLSS Quality it even beats A Plagues tale. 82fps in TLOU vs 66fps on A Plagues tale. Using one delta with PS5 doesn't show you anything. Just because A Plagues tale has low PS5 framerate doesn't mean the PC version is great. The PS5 version of a plagues tale could be poorly optimised in comparison. It was too.
XesqueVara
Member
Strix Halo can Compete favorably even againt the Ps5 pro tbh, like the CPU alone is a Antonishing Gap between the two APUs, some games which are CPU bottlenecked at PS5 PRO and runs at 30 FPS would run at 60 FPS with ease on the Halo.
Last edited:
Rocco Schiavone
Member
So you know the exact PS5 Pro Specs?? Tell me more.
XesqueVara
Member
Of course not just basing myself over the rumours that the CPU of the PS5 PRO it's still the same as the PS5 per leakers like Kepler and RGT.
playsaves3
Banned
The strix would easily outperform the pro if they stick to zen 2 (which I don't believe) but not if they update it to zen 5
playsaves3
Banned
If the cpu is still the