• 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.

Next-Gen PS5 & XSX |OT| Console tEch threaD

Status
Not open for further replies.

xool

Member
They should give us an option to stick to 1080p assets, no matter what. 4K is already ridiculous size-wise and 8K is gonna be even worse.

XB1X does not give an option for 1080p users to stick to 1080p assets. Only XB1 Slim allows you to do that. It's a simple software/OS fix.

I wonder if they will - tons of people will still be on 1080p sets when the console launches and bandwidth caps are a thing .. but I expect Sony to be traditionally slow on this, MS might offer the option .. maybe.

[edit - still even on 1080p a 4k asset will look better when close up ..]
 
Last edited:
Oh sure, I'm aware of the delta updates (which are a terrific thing). Just that unless I've missed it, I haven't had 16GB updates for Days Gone in like, ever. So for such a huge patch to arrive I'd assume you're updating from 1.00 (or whatever the gold version was). I may have worded it wrong, but that's what I was getting at.
I updated from v1.24 to v1.25 and the update size was ~15GB.

Delta updates usually take a long time to install, since they recombine old and new data (and this takes a lot of time in a 5400 RPM HDD).

I wonder if they will - tons of people will still be on 1080p sets when the console launches and bandwidth caps are a thing .. but I expect Sony to be traditionally slow on this, MS might offer the option .. maybe.

[edit - still even on 1080p a 4k asset will look better when close up ..]
Yeah, it looks better due to supersampling, but I'm not fond of 100GB games. Not to mention that loading 4K assets is slower (yet another reason about why next-gen needs SSD). 1080p loading will always be quicker.

I'd also like to have an option to remove non-English languages. Story-heavy games with tons of dialogues and a dozen language dubs take a lot of extra space these days.

People may argue that MS is the first to do this, but Sony actually had offered it back in the PS3 era for Uncharted 3 digital edition (languages, 2D/3D cutscenes, MP etc.)

ND games always use experimental tech from ICE Team. TLOU1 digital edition was a testbed for PlayGo (you download a small client and then you download the rest of the game chunks from ND's servers).
 

Imtjnotu

Member
I wonder if they will - tons of people will still be on 1080p sets when the console launches and bandwidth caps are a thing .. but I expect Sony to be traditionally slow on this, MS might offer the option .. maybe.

[edit - still even on 1080p a 4k asset will look better when close up ..]
It will be the same as now and just down sample for 1080p. They aren't going to make a game with two different size textures for the same asset. They are better off with the current solution
 

Fake

Member
16 GB?
No thx. I fine with Days Gone without this patch.
My HD is full and this patch probably don't fix shit.
Still, already beat survive difficult, get the throphy and I back to my old save for explore finally. How I miss the hud.
If they really want me to update, just add a option for dynamic hud. Until there, no go.


edit: WTF, I mistake this thread with Days Gone OT.
 
Last edited:
Nah. He's using software monitoring, a game that doesn't push the GPU(this game run 4k/30fps/X1X and averages around 1100MHz clock on RX 480). I was using software to monitor my GPU consumption and it reported 105W on DOOM, then I got a kill-a-watt and it's easily over 150W. 5700 Pro comes undervolted to .98 V and runs around 1.7GHz while pulling ~166W, while XT is 1.2V. What voltage is he running?


there are some problems with your assessment. the RX480 only shows die power without VRAM and auxiliaries. 5700 series now shows total board power on the power readout. thats apples vs. oranges. furthermore - and everyone really has to know that when comparing power: you can't directly compare power figures drawn from the wall with figures on the low voltage side (which the software power readout is). in between both figures lies the PSU efficiency as a factor. so depending on your PSU and PSU load you lose around 15% from the wall socket to the 12V rail.
 

CrustyBritches

Gold Member
I've been perusing some FS tests now that were getting more results, and I was curious about the generational uplift from Ryzen 2000 to Ryzen 3000 when paired with a RX 5700 Pro. Got some results that lined up nicely for comparison:
1. R5 3600 3.6GHz base/4.3GHz boost, RX 5700 Pro 1699MHz, 16GB RAM, SSD
2. R5 2600X 3.6GHz base/4.2GHz boost, RX 5700 Pro 1709MHz, 16GB RAM, SSD


Results:
2600x3600comp1.jpg

2600x3600comp2specs.jpg
Since the R5 3600 had 4.3Ghz boost instead of the R5 2600X's 4.2GHz boost, I checked some other scores for R5 3600 3.6GHz base/4.2Ghz boost and the difference the .1GHz makes is 2-3% Physics Score, that's within margin of error. Results showed ~14-15% generational jump in physics result, and maybe 9-10% overall score(after adjusting for .1GHz). Just from the jump to Zen 2.
I was curious about how FS handles multithreading so I ran R5 1600 6c/12t at 3.2GHz, RX 480 stock in few different configs to get a look. I used system configuration, boot, advance, # of processors to disable 1c/2t for the higher clock/lower thread count conifg. Did a control test to makes sure it was disabled by using 5c/10t @ 3.2GHz and it reports and scales correctly from 6c/12t @ 3.2GHz, with ~15% difference in Physics score.
1. R5 1600 6c/12t @ 3.2GHz, RX 480 stock, 16GB, SSD
2. R5 1600 5c/10t @ 3.8GHz, RX 480 stock, 16GB, SSD

Results(6c/12t on the left, 5c/10t on the right):
CoresCompare.jpg

CPU usage was ~10%, with FS doing most of the test on a single thread with a little help from another thread, then flat 5-10% across the rest. During the Physics/CPU Test, all threads jump to ~87% usage. Later during the Combined test it jumps again but not as high. Here's the CPU usage from Resource Monitor to demonstrate Fire Strike's multithreading:
TaskManageCPU-MT.jpg

Perfectly balanced across all threads^. You can see the thread that FS uses for the Cinematic Demo and Graphics 1&2 at CPU 8. Then it spikes, especially on that main thread up to almost full usage during the Physics Test loading screen, followed by all threads jumping for the Physics Test.

Difference in test configs was .9% total score, and 2.3% physics.
In light of FS being a synthetic, canned benchmark I'd say it's safe to assume a 3.2-3.3GHz 8c/16t Zen 2 in consoles would perform similar to a 4.2-4.3GHz R5 3600. Because of that, it's plausible Gonzalo can be 20K+ with Zen 2 8c/16t 3.2GHz CPU and RX 5700 Pro @ 1.8GHz.
20k5700.jpg
P-Plausible, I say.
giphy.gif
 
Last edited:

xool

Member
I'd also like to have an option to remove non-English languages. Story-heavy games with tons of dialogues and a dozen language dubs take a lot of extra space these days.

Just downloading the system specific language(s) (let users select multiple languages) , and maybe offer language packs as (free) DLC would work - I think the Witcher3 did something like this.


I've been perusing some FS tests now that were getting more results, and I was curious about the generational uplift from Ryzen 2000 to Ryzen 3000 when paired with a RX 5700 Pro. Got some results that lined up nicely for comparison:

Interesting the so the Zen2 has a +18% score (at similar clock) to previous Zen for CPU heavy (physics) .. Maybe this directly correlates to +18% IPC, or did we get increased SIMD too?

I think you posted the wrong image here for "Did a control test to makes sure it was disabled by using 5c/10t @ 3.2GHz and it reports and scales correctly from 6c/12t @ 3.2GHz, with ~15% difference in Physics score." https://www.upload.ee/image/10221876/CoresCompare.jpg
 

-kb-

Member


arcturus incoming.

looks like there will be three variants


I think there's a chance this is unrelated to the consoles. I mean after all, Sony didn't even merge there changes to LLVM for Jaguar optimisations until after the console was released. Why would AMD put codenames of consoles in desktop drivers?.
 

xool

Member
Interesting. Raven is the name of the APU?


Zen + Vega APU, not for us


What is your interpretation of that tweet? I don't know what to think...

Code:
/* Arcturus */
+    {0x1002, 0x738C, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ARCTURUS},
+    {0x1002, 0x7388, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ARCTURUS},
+    {0x1002, 0x738E, PCI_ANY_ID, PCI_ANY_ID, 0, 0, CHIP_ARCTURUS},

1002 is AMD's VID (Vendor ID)

the next number eg 0x738C is probably the DID (device ID) - there are 3 so three versions (gpus)

According to AMD Sends Out Linux Graphics Driver Patches For "Arcturus" As New Vega Derived GPU Arcturus is a Vega replacement (remember Vega7 just got killed) and is also not for us .. so it's a workstation/compute chip probably.
 

CrustyBritches

Gold Member
Interesting the so the Zen2 has a +18% score (at similar clock) to previous Zen for CPU heavy (physics) .. Maybe this directly correlates to +18% IPC, or did we get increased SIMD too?

I think you posted the wrong image here for "Did a control test to makes sure it was disabled by using 5c/10t @ 3.2GHz and it reports and scales correctly from 6c/12t @ 3.2GHz, with ~15% difference in Physics score." https://www.upload.ee/image/10221876/CoresCompare.jpg
Yeah more or less it was 17.9%, but I looked up other scores with R5 3600 @ 3.6GHz base/4.2GHz boost to match 2600X clocks and the difference in Physics Test was 2-3%, so let's say 15-18% on the Physics.

It was just a massive clusterfuck of a post, and I was trying to trim it down by quoting/embed/spoiler-ing. Here's the control results and the corresponding image I omitted before:
It was 16% difference in Physics Score.
 

-kb-

Member
Just downloading the system specific language(s) (let users select multiple languages) , and maybe offer language packs as (free) DLC would work - I think the Witcher3 did something like this.




Interesting the so the Zen2 has a +18% score (at similar clock) to previous Zen for CPU heavy (physics) .. Maybe this directly correlates to +18% IPC, or did we get increased SIMD too?

I think you posted the wrong image here for "Did a control test to makes sure it was disabled by using 5c/10t @ 3.2GHz and it reports and scales correctly from 6c/12t @ 3.2GHz, with ~15% difference in Physics score." https://www.upload.ee/image/10221876/CoresCompare.jpg

IIRC, Zen 2 should have ~2x the FLOPS of Zen1 without any other performance improvements because they doubled the width of the AVX units from 128bit to 256bit.
 

R600

Banned
You claimed X1X performance reflected a RX 580(200W avg consumption). I showed why it reflected a RX 480 underclocked/undervolt in the same FH4 with 1120MHz avg, ~150W RX 480, 4k/30fps/near ultra. If I end up with ~150W at 1120MHz with a GPU that starts at 166W at 1266MHz, then there is a problem with your math. That's all. 5700 Pro is your underclocked/undervolted 5700 XT, but less 4CUs and it still pulls ~166W avg, just like RX 480. I bet a good tuning would yield around ~150W just like my results. Anandtech's results match this as well.
Nah, its very easy to see X1X GPU is more akin to 580 then 480. It has more grunt actually compared to 580, but thats beside the point.

Comparing X1X version with 580 and Intel or Zen+ CPUs is very hard since Jaguars are incredibly limited, even bottlenecking GPU tasks.
 

xool

Member
Don't know why it's being discussed, but re RX 480 vs RX 580 - they are literally the same chip, -the same 14nmLPP 5.7billion transistor 232mm2 die common to RX470, RX480, RX570, RX580, and RX590 (lower parts have broken/disabled CUs) .. the difference between the 5 (Polaris20) and 4 series (Polaris10) is 5 series are up clocked.

Xbox One X graphics part has 40CUs (more than the 36 limit on Polaris10/20) but runs at nearer to 4 series clocks
 
Last edited:
I've been perusing some FS tests now that were getting more results, and I was curious about the generational uplift from Ryzen 2000 to Ryzen 3000 when paired with a RX 5700 Pro. Got some results that lined up nicely for comparison:

I was curious about how FS handles multithreading so I ran R5 1600 6c/12t at 3.2GHz, RX 480 stock in few different configs to get a look. I used system configuration, boot, advance, # of processors to disable 1c/2t for the higher clock/lower thread count conifg. Did a control test to makes sure it was disabled by using 5c/10t @ 3.2GHz and it reports and scales correctly from 6c/12t @ 3.2GHz, with ~15% difference in Physics score.

In light of FS being a synthetic, canned benchmark I'd say it's safe to assume a 3.2-3.3GHz 8c/16t Zen 2 in consoles would perform similar to a 4.2-4.3GHz R5 3600. Because of that, it's plausible Gonzalo can be 20K+ with Zen 2 8c/16t 3.2GHz CPU and RX 5700 Pro @ 1.8GHz.
20k5700.jpg
P-Plausible, I say.
giphy.gif
And you can see how small difference in 3.3GHZ and 4GHZ is in actual score.

Bodes well for Gonzalo being 8-8.5TF as 5700 + 3600X (6core) score ~20K in FS.

B...but songoku said...
 
And you can see how small difference in 3.3GHZ and 4GHZ is in actual score.

Bodes well for Gonzalo being 8-8.5TF as 5700 + 3600X (6core) score ~20K in FS.
Makes sense to me. Optimistically I think we will be more along the lines of 8.5-9TF. No science or math behind that thought. All gut feeling.
 
Last edited:

SonGoku

Member
And you can see how small difference in 3.3GHZ and 4GHZ is in actual score.

Bodes well for Gonzalo being 8-8.5TF as 5700 + 3600X (6core) score ~20K in FS.
Bodes well for consoles in general, regardless of gonzalo being related to PS5 or Subor or it being real at all
B...but songoku said...
Did i say anything about gonzalo FS score? comon man try harder
Optimistically I think we will be more along the lines of 8.5-9TF.
If that's optimistic, I dread to think what's pessimist for you
5.8TF? :messenger_astonished:
there are some problems with your assessment. the RX480 only shows die power without VRAM and auxiliaries. 5700 series now shows total board power on the power readout. thats apples vs. oranges. furthermore - and everyone really has to know that when comparing power: you can't directly compare power figures drawn from the wall with figures on the low voltage side (which the software power readout is). in between both figures lies the PSU efficiency as a factor. so depending on your PSU and PSU load you lose around 15% from the wall socket to the 12V rail.
Interesting and even if the readings are lower than reported it still shaves nearly 100W from uv and downclocking to 1686.
Gives me confidence a 7nm DUV APU can hit 11TF (2080 SUPER territory)
On EUV 13TF is quite likely a baseline expectation (2080Ti territory)

11TF provides a decent next gen jump but if Sony/MS really intend to push the power envelope without breaking the bank, I believe EUV was in their plans all along.
Sony said:
We are just as dedicated to delivering that immersive experience on the next generation console we are currently developing. We plan to do that by further improving the computational power of the console, measured in TFLOPS (Tera Floating-point Operations Per Second), and by dramatically increasing the graphics rendering speed through the employment of a customized ultra-fast, broadband SSD.
 
Last edited:
Did i say anything about gonzalo FS score? comon man try harder

you discarded completely Gonzalo, because the name doesn't suites Sony

SonGoku represents the best-case scenario, so he's like the Ghost of Christmas Present...
NuKu.gif


I'm the Ghost of Christmas Future...
future8.gif

i told him numerous times

based on Gonzalo :
36 CU's @1800mhz = 8.3 Tf RDNA 1
40 CU's @1800mhz = 9.2 Tf RDNA 1
if the clocks are too high for a console

46 CU's @ 1550mhz = 9.1 Tf RDNA 1
44 CU's @ 1550mhz = 8.7 Tf RDNA 1 (i'm leaning more on this one tbh)

worst case is :

* 44 CU @ 1550 Mhz = 8,7 Tf Rdna which is 10,8 Tf GCN
im pretty sure next gen is better than google stadia

more optimistic are :

* 46 CU @ 1550 Mhz = 9,1 Tf Rdna
* 48 CU @ 1550 Mhz = 9,5 Tf Rdna

dream scenario would be
* 52 CU @ 1550 Mhz = 10.3 Tf Rdna
 

FranXico

Member
We are just as dedicated to delivering that immersive experience on the next generation console we are currently developing. We plan to do that by further improving the computational power of the console, measured in TFLOPS (Tera Floating-point Operations Per Second), and by dramatically increasing the graphics rendering speed through the employment of a customized ultra-fast, broadband SSD.

I'm afraid this doesn't tell us anything at all. Just standard corporate vague PR speak.
 

FranXico

Member
It hints they may talk about floating point perfomance during the PS5 reveal, why would they prop up TF increase if its anything less than double digits?
They would always prop up any amount of increase, no matter how seemingly minor. From what I can tell, they are managing expectations the exact same way as the competition.
It gives me the feeling, in terms of power, that it could go either way...
 

SonGoku

Member
They would always prop up any amount of increase, no matter how seemingly minor. From what I can tell, they are managing expectations the exact same way as the competition.
It gives me the feeling, in terms of power, that it could go either way...
Fair enough
I brought it up, because earlier people were speculating MS/Sony not mentioning FP metrics somehow related to consoles being under powered (8-9TF). There hasnt been a full reveal yet which is why they havent mentioned specifics
 

Mass Shift

Member

Zen + Vega APU, not for us

Well I knew Raven Ridge couldn't have possibly been it. Just thought it was another Raven code name.

But thanks for the clear up 😊
 

vpance

Member
Makes sense to me. Optimistically I think we will be more along the lines of 8.5-9TF. No science or math behind that thought. All gut feeling.

$399 if that happens.

I don't think it's worth it for Sony or MS to go that low. Too much bad PR spec wise, even if it is cheap.
 
Last edited:
That was one of two reasons!
The other being that 5700 performance is not enough for next gen gaming at 4K
If the target resolution was 1080p/1440p it would be fine.
Anyhow what does fs scores posted by CrustyBritches CrustyBritches have to do with anything I've said?


Ghost of Tsushima demo was running on PS4 Pro btw

i don't know what 4k next gen gaming experience your hoping for
Ghost of tsushima DF :
But as the end credit tile reveals, the demo runs on hardware you may already own.
Rumours persist that Tsushima is a cross-gen game that may also appear on PlayStation 5
just compare physics simulation on this game and Just Cause 4 (lower graphics) running on Ps4Pro.
you should know by now, all that means is it will be on Ps4Pro
the witcher3 trailer, watch dog trailer, crackdown3 trailer and so and so... (hype)

Crazy ass particle physics score cancelled, I'm afraid




NOOOOOOO :messenger_loudly_crying:
 

SonGoku

Member
i don't know what 4k next gen gaming experience your hoping for
Better graphics than current gen lol you know... a next gen leap
Rumours persist that Tsushima is a cross-gen game that may also appear on PlayStation 5
Rumors are just that rumors... we have a detailed analysis that points to the demo running on PS4 Pro and the CB resolution matches Second Son Pro patch
Its just a testament to devs talent and pushing the limits on exclusives
the witcher3 trailer, watch dog trailer, crackdown3 trailer and so and so... (hype)
lol i could understand your concern last gen with KZ2 E3 demo, but this gen Sony 1st party haven't bullshoted with trailers
 
Better graphics than current gen lol you know... a next gen leap

we already discussed this, why you still bringing it up ?
we talked about how 4k could run on next gen based on realistic next gen specs

we have a detailed analysis that points to the demo running on PS4 Pro and the CB resolution matches Second Son Pro patch
Its just a testament to devs talent and pushing the limits on exclusives

are these sceneries possible on a Ps4 pro ?

 

SonGoku

Member
we already discussed this, why you still bringing it up ?
Because you brought up my reasons for dismissing Gonzalo
5700 not being up to the task is one of them :)
we talked about how 4k could run on next gen based on realistic next gen specs
I haven't seen anything of the sort, you mentioned PCs settings of current gen games
Im talking next gen games at 4K, 5700 doesn't have enough juice for that.
are these sceneries possible on a Ps4 pro ?
Sure why not? TLOU2 demo looks similarly impressive, and again its 1800p CB
Late PS4 exclusives are getting the most out of fine grain async compute features implemented on PS4 and absent on xboneS
 
Last edited:
Last edited:

xool

Member
It hints they may talk about floating point perfomance during the PS5 reveal, why would they prop up TF increase if its anything less than double digits?
But they're talking about "computational power" ie CPU, not "graphics power" - that's something that can brag about because Zen2 >> Jaguar .

..what I think they can't brag about so much is GPU power - hence the focus on SSD speed.

I expect promotional material to heavily push :
  • CPU FLOPS increase over Jaguar
  • SSD speed increases over previous
  • 4k ability / VRR / more 60fps capable
But the GPU FLOPS figure won't be prominent because though bigger, it won't look as great a relative improvement
 

SonGoku

Member
But the GPU FLOPS figure won't be prominent because though bigger, it won't look as great a relative improvement
He explicitly mentioned TFLOPS how many TF does the 3700x even have?
If the GPU is on the double digits TF, they'll prop it up especially if its on the 12-13TF range
 
here you go:


hope that's the right one. not familiar with geekbench.

overclocking results:


doesn't seem to do to much, but strangly when i turned off one CCX on the CPU it bosted the score as much as all GPU and CPU OC combined :p
 

LordOfChaos

Member
Last edited:

Mass Shift

Member
Sony said:
We are just as dedicated to delivering that immersive experience on the next generation console we are currently developing. We plan to do that by further improving the computational power of the console, measured in TFLOPS (Tera Floating-point Operations Per Second), and by dramatically increasing the graphics rendering speed through the employment of a customized ultra-fast, broadband SSD.

Further improving. Plenty self explanatory right there.
 

mckmas8808

Banned
i don't know what 4k next gen gaming experience your hoping for
Ghost of tsushima DF :
But as the end credit tile reveals, the demo runs on hardware you may already own.
Rumours persist that Tsushima is a cross-gen game that may also appear on PlayStation 5
just compare physics simulation on this game and Just Cause 4 (lower graphics) running on Ps4Pro.
you should know by now, all that means is it will be on Ps4Pro

the witcher3 trailer, watch dog trailer, crackdown3 trailer and so and so... (hype)

Just Cause 4? You know Sony's internal teams are better than most.
 
Status
Not open for further replies.
Top Bottom