• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

[NX Gamer] AC: Valhalla - XsX vs PS5 - The first REAL head to head battle.

assurdum

Banned
23 pages over... not a big difference? I never cared when One X eked out slightly more than the Pro and I certainly wouldn't care about this were I an XSX owner. I get that part of the hoopla is because of certain posters claiming there'd be a gap in performance on XSX's side but honestly who cares. Not a one of us is Nostadamus and we all know bigger number mean better.
Not true at all .
 

nowhat

Member
Folks, dial back the baiting and antagonism a little.

While "serving crow" has always been a staple in online gaming forum culture, let's not step over the line into nasty territory.

Let's have fun going into this new gen, but not completely at the expenses of others- more so at the expense of the games.

Thank you.
 

Bo_Hazem

Banned
EmspdjfUwAIfu1p.jpg

Is this real?

This is from the VG comparison video, nothing noticeably different except on XSX seems to be darker:

vlcsnap-2020-11-13-20h53m02s410-2.png


Not close enough, and not good enough for comparison.

While we are at it, this is a quick capture from internal PS5 recording, you can keep it at automatic 4K@60fps HDR recording for up to 15 minutes, or capture longer with the built-in Share Factory Studio app in PS5:




Pretty clean footage, and a very powerful editing app that will make many budget content creators stand out with quality stuff. Kingthrash Kingthrash no more 1080p when you go to PS5, my bro ;)
 
Last edited:

Lysandros

Member
https://www.pttweb.cc/bbs/PC_Shopping/M.1590679194.A.F5B

Taiwanese Aquariuszi:

PS5 did start much earlier, but the final chip revision started H2 2019.

XSX started the chip verification in H1 2019 and finished in H2 2019.

The hardware was final mid 2019 on both, and both had same time to do SDK and apis.
No, the great Dictator recently declared thus officially confirmed in Beyond3D forums that PS5 hardware was final in 2018 and was supposed to launch in 2019. Furthermore he stated very clearly that PS5's GPU lacks all the essential hardware features of RDNA2 compared to XSX and said that there is something very very wrong with the API's since it's (XSX) not performing as it should. Now he will do the DF PS5/XSX comparison of Assassin's Creed apparently.. Reality beware; you will soon be distorted by a greater power than you.
 

yurinka

Member
It is not "up to" 18%, it's supposed to be AT LEAST 18%. This has translated next to no practical change.
The 18% is also variable though. 'Up to' 18% infact.
PS5 has a whopping nearly 2X SSD i/o paper advantage, and we seen AC, a x-gen game on an old engine, gives a double figure 12% win, at best, or translates to a whopping 1.5s faster loading! 🤷‍♀️
The difference is a XSX lead of ~18% in theorical peak for some GPU tasks and a lead of ~3% in CPU frequency.

But we also know PS5 has a ~22% lead in GPU frequency (so has a lead in other GPU tasks over XSX) and a lead in several important I/O areas, which means feeding faster GPU and CPU in a more optimized way, being able to free unneeded resources faster and replacing them via streaming with new ones. Which keeps it more time 'busy' with available resources to be used soon, so being able to reach a higher % of its lower theorical peak.

This means, in practical terms the performance difference is going to be tiny, in some games or areas with a console taking a small lead, and in other cases the other console taking a small lead.

Outside DF or NX Gamer comparision videos 99% of the players won't be able to spot the performance differences outside some small bugs or optimization issues that a specific game may have in certain areas (as could be tearing or tesselation in ACV).

Then we have the loading times differences, where PS5 will have a lead for native games but in most cases only of a few seconds, at least until devs start using Oodle Texture which will enlarge the difference. For BC games PS5 seems to shrink in a bigger % the PS4 Pro loading times, but since its games took way longer to load than the XB1X due to their difference in power, in many cases Series X loads them faster than PS5.

Then there will be the Quick Resume difference: in a console only works with a few recently played games and requires a big SSD chunk but resumes exactly where you left it and works with both BC and native games. The other instead isn't limited to the few recently played games and resume to the most recent savepoint instead but allows you to have quick access to multiple specific stages or game modes skipping menus and logo screens in all native games installed in the console but not with BC games.

So at the end, we get pretty much very similar results, sometimes better in one side, sometimes better in the other one. Or are different things where your preference may be for one or the another. There isn't a clear winner both on paper and with real world results.
 
Last edited:

Lysandros

Member



According to Vg Tech performance analyse of AC Valhalla PS5 holds 60 FPS/16.67ms 97.52% of the time compared to XSX's 87.28%. The number of torn frames is 865 for PS5 and 4025 for XSX, so there is much less screen tearing on PS5. Thus performance is slightly higher/more consistent on PS5.

Edit: Series X/S also seem to have stuttering in cutscenes.
 
Last edited:

RJMacready73

Simps for Amouranth



According to Vg Tech performance analyse of AC Valhalla PS5 holds 60 FPS/16.67ms 97.52% of the time compared to XSX's 87.28%. The number of torn frames is 865 for PS5 and 4025 for XSX, so the is much less screen tearing on PS5. Thus performance is slightly higher/more consistent on PS5.


Jesus wept... We're down to arguing over percentage variables smaller than my dick?
 

rnlval

Member
That claim is misleading:

PS5XSXDifferenceNotes:
CPU Clock (single threaded) GHz3.53.8-8.57%
CPU Clock (dual threaded) GHz3.53.6-2.85%
GPU Clock (MHz)2233182522.35%
Shading Units23043328-44.44%
Texture Mapping Units144208-44.44%
Triangles (billion/sec)8.927.322.19%4 Primitive Units x clock speed
Triangles Culled (billion/sec)17.8414.622.19%(Each Primitive unit can cull two primitives) x clock speed
Pixel Rate (GPixel/s) via the ROPS142.9116.822.34%Clock speed x 64 ROPS (Render output unit)
Texture Rate (GTexel/s)321.6379.6-18.03%Clock speed x TMU's (Texture mapping unit)
FP32 (float) performance10.2912.15-18.07%
GPU Cache Speed (MHz)2233182522.35%Tied to clock rate.
SSD Rate (GB/sec)5.52.4229%

It's like cherry picking the specs you want, in order to claim the most powerful console.
XSX GPU has 5MB L2 cache (320-bit bus) which is higher than 4MB L2 cache(256-bit bus).
 

longdi

Banned
That claim is misleading:

PS5XSXDifferenceNotes:
CPU Clock (single threaded) GHz3.53.8-8.57%
CPU Clock (dual threaded) GHz3.53.6-2.85%
GPU Clock (MHz)2233182522.35%
Shading Units23043328-44.44%
Texture Mapping Units144208-44.44%
Triangles (billion/sec)8.927.322.19%4 Primitive Units x clock speed
Triangles Culled (billion/sec)17.8414.622.19%(Each Primitive unit can cull two primitives) x clock speed
Pixel Rate (GPixel/s) via the ROPS142.9116.822.34%Clock speed x 64 ROPS (Render output unit)
Texture Rate (GTexel/s)321.6379.6-18.03%Clock speed x TMU's (Texture mapping unit)
FP32 (float) performance10.2912.15-18.07%
GPU Cache Speed (MHz)2233182522.35%Tied to clock rate.
SSD Rate (GB/sec)5.52.4229%

It's like cherry picking the specs you want, in order to claim the most powerful console.

Ps5 uses boost clocks, variable. You can't just compare this way
 

Andodalf

Banned
320-bit cache bus? That's the external memory bus, and we don't know the size of the PS5 internal GPU cache, as that information has not be revealed.

Please provide sources.



Please provide graphs showing at what clock rate the PS5 GPU runs at in games over time.

Since you cannot provide that, I will assume maximum clocks are available when needed.

If it was max all the time it wouldn't be a boost then. It would just be the clock
 

Md Ray

Member
The difference is a XSX lead of ~18% in theorical peak for some GPU tasks and a lead of ~3% in CPU frequency.

But we also know PS5 has a ~22% lead in GPU frequency (so has a lead in other GPU tasks over XSX) and a lead in several important I/O areas, which means feeding faster GPU and CPU in a more optimized way, being able to free unneeded resources faster and replacing them via streaming with new ones. Which keeps it more time 'busy' with available resources to be used soon, so being able to reach a higher % of its lower theorical peak.

This means, in practical terms the performance difference is going to be tiny, in some games or areas with a console taking a small lead, and in other cases the other console taking a small lead.

Outside DF or NX Gamer comparision videos 99% of the players won't be able to spot the performance differences outside some small bugs or optimization issues that a specific game may have in certain areas (as could be tearing or tesselation in ACV).

Then we have the loading times differences, where PS5 will have a lead for native games but in most cases only of a few seconds, at least until devs start using Oodle Texture which will enlarge the difference. For BC games PS5 seems to shrink in a bigger % the PS4 Pro loading times, but since its games took way longer to load than the XB1X due to their difference in power, in many cases Series X loads them faster than PS5.

Then there will be the Quick Resume difference: in a console only works with a few recently played games and requires a big SSD chunk but resumes exactly where you left it and works with both BC and native games. The other instead isn't limited to the few recently played games and resume to the most recent savepoint instead but allows you to have quick access to multiple specific stages or game modes skipping menus and logo screens in all native games installed in the console but not with BC games.

So at the end, we get pretty much very similar results, sometimes better in one side, sometimes better in the other one. Or are different things where your preference may be for one or the another. There isn't a clear winner both on paper and with real world results.
I couldn't have said it better myself.
 

rnlval

Member
320-bit cache bus? That's the external memory bus, and we don't know the size of the PS5 internal GPU cache, as that information has not be revealed.

Please provide sources.


Please provide graphs showing at what clock rate the PS5 GPU runs at in games over time.

Since you cannot provide that, I will assume maximum clocks are available when needed.
PS5's die size is only 309mm2.
 

rnlval

Member
That claim is misleading:

PS5XSXDifferenceNotes:
CPU Clock (single threaded) GHz3.53.8-8.57%
CPU Clock (dual threaded) GHz3.53.6-2.85%
GPU Clock (MHz)2233182522.35%
Shading Units23043328-44.44%
Texture Mapping Units144208-44.44%
Triangles (billion/sec)8.927.322.19%4 Primitive Units x clock speed
Triangles Culled (billion/sec)17.8414.622.19%(Each Primitive unit can cull two primitives) x clock speed
Pixel Rate (GPixel/s) via the ROPS142.9116.822.34%Clock speed x 64 ROPS (Render output unit)
Texture Rate (GTexel/s)321.6379.6-18.03%Clock speed x TMU's (Texture mapping unit)
FP32 (float) performance10.2912.15-18.07%
GPU Cache Speed (MHz)2233182522.35%Tied to clock rate.
SSD Rate (GB/sec)5.52.4229%

It's like cherry picking the specs you want, in order to claim the most powerful console.
Theoretical numbers are nearly pointless when it's memory bandwidth bound.
 

rnlval

Member
And workloads are not maximum all the time.

Much better to have a design that down clocks to save power, then running at high clocks all the time. Don't you want to save the planet?

It's funny how PC max CPU clock, and max GPU clocks are accepted as the final rate, but PS5 is placed in it's own category of "variable" and so it's max rate is not accepted.

MS claims for their clock rate is very misleading. You think they run at maximum clocks on the desktop? Of course not, I'm sure they are down clocking also when the load is not there.

Is that a performance spec? So MS is the most powerful console, cause it has the biggest die?
Largest GPU with the largest chip area size.
 

Kerlurk

Banned
 
Last edited:

TJC

Member
Please provide proof for PS5 GPU's 2230Mhz is running 100% of the time as it's base clock speed.
How many times do the same things have to be said lol? You do realise PC's run with variable? Why would you make all 100 power available in a scene that dosent need that power? But maybe the CPU could do with a little hand that's what the variable clocks allow. Imagine a scene where GPU usage might be 50%, why use 100 when 50 is good enough?
 

rnlval

Member
How many times do the same things have to be said lol? You do realise PC's run with variable? Why would you make all 100 power available in a scene that dosent need that power? But maybe the CPU could do with a little hand that's what the variable clocks allow. Imagine a scene where GPU usage might be 50%, why use 100 when 50 is good enough?
You do realize NVIDIA's paper spec variable clock speed is very conservative?
 

longdi

Banned
What is XSX being bound by here?
vlcsnap-2020-11-13-18h56m20s419.png



He speaks about 120fps here but it applies to 60fps too.


Bound by their less than matured devkits.

Jesus of man, do you guys just conveniently ignored by what we has shown, about ps5 super fast and easy to tap development first mover advantages?
 
Top Bottom