• Ever wanted an RSS feed of all your favorite gaming news sites? Go check out our new Gaming Headlines feed! Read more about it here.
  • We have made minor adjustments to how the search bar works on ResetEra. You can read about the changes here.

digitalrelic

Weight Loss Champion 2018: Biggest Change
Banned
Oct 25, 2017
13,124


I found this to be a pretty fascinating look at how the PS5 GPU & SSD stack up against various PC configurations with exact same settings & resolution. A great way to finally see somewhat definitely how PS5 specs stack up in the PC realm.

  • PS5 loads in 4.38 seconds vs 10.12 seconds on a PC w/ an i9 10900k & 3.5 GB/s NVME drive
  • PC w/12900k and 7.5GB/s NVME drive showed little loading time improvements on PC. Appears to hit an artificial ceiling on PC for load times.
  • Load time varies depending on what save is being loaded & what PC hardware is in place but PS5 is always significantly faster
  • Alex used scenes where the PS5 would drop below 60fps as a benchmark to test like-for-like GPU performance on PC
  • nVidia RTX 2060 Super performs at 71% of the performance of the PS5 on average
  • AMD RX 5700 performs at 75.4% of the performance of the PS5 on average
  • RX 5700XT performs at 83.5% of the performance of the PS5 on average
  • RTX 2070 Super performs at 88.5% of the performance of the PS5 on average
  • RTX 2080 performs at 96.6% of the performance of the PS5 on average.
  • PS5 vastly outpaced it's RDNA 1 cousins on the PC side
  • PS5 lands somewhere between an RTX 2080 and an RTX 2080 Super in performance

Honestly kind of insane to me, the type of GPU performance we're seeing out of a box that starts at $399.
 
Last edited:

Deleted member 110527

Mar 11, 2022
1,311
  • PS5 lands somewhere between an RTX 2080 and an RTX 2080 Super in performance

Suuuper interesting, isn't this also what they found in another PS5 vs PC comparison video? Maybe for Deathloop or the enhanced edition of Metro Exodus? Could be misremembering.
 

Spork4000

Avenger
Oct 27, 2017
8,520
So it performs right where it should based on expected raster performance? Neat to have confirmation I guess.
 
Last edited:

Lukar

Unshakable Resolve - Prophet of Truth
Member
Oct 27, 2017
23,402
Kudos to you for making the thread and having the summary / info in the OP from the start, instead of just posting the video and editing it in 15-20 minutes later. It's appreciated.
 

Wrexis

Member
Nov 4, 2017
21,247
That's not a particularly fair comparison for loading times though?

PS5 NVME is 5.5GB/s max I think, and my own computer's NVME built last year has 5.0GB/s max so using a NVME with a max throughput of 3.5GB/s surprised me.
 

maximumzero

Member
Oct 25, 2017
22,925
New Orleans, LA
Anyone know if the Director's Cut is a heavier "workload" on hardware than the original release. I managed to get around 90fps with everything on ultra at 1080p with my 5600 XT (Roughly equivalent to a stock RTX 2060) but maybe improved graphical fidelity would potentially bog that down a bit.
 

Patitoloco

Member
Oct 27, 2017
23,670
Anyone know if the Director's Cut is a heavier "workload" on hardware than the original release. I managed to get around 90fps with everything on ultra at 1080p with my 5600 XT (Roughly equivalent to a stock RTX 2060) but maybe improved graphical fidelity would potentially bog that down a bit.
No, it's the same in that sense.
 

pswii60

Member
Oct 27, 2017
26,673
The Milky Way
Suuuper interesting, isn't this also what they found in another PS5 vs PC comparison video? Maybe for Deathloop or the enhanced edition of Metro Exodus? Could be misremembering.
It's actually a little uninteresting in a sense, because it falls almost exactly in line with the raster teraflops difference between the GPUs, so it's exactly as expected with no surprises lol.

That said, as Alex points out later in the video, the bench is only a snapshot and there is another area where the 2080 outperforms the PS5 GPU, which is a little more interesting, because that would be less expected. Also keeping in mind this is raster only, obviously the tables will turn where ray-tracing is concerned on the Nvidia GPUs, not to mention AI upscaling.
Seems weird to use a slower drive for some reason.
Alex tried it with a faster drive too but it didn't make much difference.
 

ElNerdo

Member
Oct 22, 2018
2,230
That's not a particularly fair comparison for loading times though?

PS5 NVME is 5.5GB/s max I think, and my own computer's NVME built last year has 5.0GB/s max so using a NVME with a max throughput of 3.5GB/s surprised me.
They also tried one that was around 7.5GB/s and it still didn't load as fast as the PS5, but was very close.
 

Tagyhag

Member
Oct 27, 2017
12,497
This is more in line with what I was expecting from the PS5, I think the performance is supposed to be around a 2080 Super.

Not sure why DF didn't use a NVME with equal speeds on PC though. Then we could see what advantage the PS5's I/O brings.
 

Deleted member 110527

Mar 11, 2022
1,311
It's actually a little uninteresting in a sense, because it falls almost exactly in line with the raster teraflops difference between the GPUs, so it's exactly as expected with no surprises lol.

That said, as Alex points out later in the video, the bench is only a snapshot and there is another area where the 2080 outperforms the PS5 GPU, which is a little more interesting, because that would be less expected. Also keeping in mind this is raster only, obviously the tables will turn where ray-tracing is concerned on the Nvidia GPUs, not to mention AI upscaling.

Fair point, I haven't looked into the hard numbers between the two, so it was news to me. Just summoned that half-remembered memory that maybe it had been mentioned in another video of theirs.
 

zaitsu

Banned
Jan 27, 2022
276
So when RTX and DLSS isn't involved PS5 is on par with 2080=3060 ti.
That's wild for 399$
 

exodus

Member
Oct 25, 2017
9,949
Current gen consoles are absolute bargains when it comes to price/performance. You need a 2080 Super or 3060 Ti to roughly match a PS5. That'll run you $740-800 CAD if you're lucky enough to find one today. Compared to a $500 CAD PS5, consoles have a monumental price advantage today.

I really hope PC component prices come back to the glory days of the early-mid 2010's.

I paid more than 3x the price of my PS5 for a 3080 Ti FE. ; _ ;
 

Tora

The Enlightened Wise Ones
Member
Jun 17, 2018
8,640
Not to be a hater, but surely most people are dropping 499 on the PS5 (which is still a great price anyway), not 399 on the non-existent digital sku that ends up costing you a shit ton more in the long run.

Just a nitpick :P I know people own the digital edition

Not sure why DF didn't use a NVME with equal speeds on PC though. Then we could see what advantage the PS5's I/O brings.
Because the game wouldn't even be able to take advantage of it on PC without directstorage I suppose. You saw Will's PC with a 7.5GB/s NVMe produced no additional time savings in loading
 

JoJo'sDentCo

Unshakable Resolve
Member
Oct 25, 2017
8,538
dgrdsv This is pretty interesting.

"PS5 loads in 4.38 seconds vs 10.12 seconds on a PC w/ an i9 10900k & 3.5 GB/s NVME drive"
 
OP
OP
digitalrelic

digitalrelic

Weight Loss Champion 2018: Biggest Change
Banned
Oct 25, 2017
13,124
That's not a particularly fair comparison for loading times though?

PS5 NVME is 5.5GB/s max I think, and my own computer's NVME built last year has 5.0GB/s max so using a NVME with a max throughput of 3.5GB/s surprised me.
Seems weird to use a slower drive for some reason.
This is more in line with what I was expecting from the PS5, I think the performance is supposed to be around a 2080 Super.

Not sure why DF didn't use a NVME with equal speeds on PC though. Then we could see what advantage the PS5's I/O brings.

I'll add it to the OP, but Alex also tested a PC with 7.5GB/s max throughput and the PS5 still outperformed it.
 

Firmus_Anguis

Member
Oct 30, 2017
6,117
This is why I'm baffled when people are 'worried' about this generation...

The new consoles are quite beastly. Devs have hardly utilised them properly.
 

zaitsu

Banned
Jan 27, 2022
276
Current gen consoles are absolute bargains when it comes to price/performance. You need a 2080 Super or 3060 Ti to roughly match a PS5. That'll run you $740-800 CAD if you're lucky enough to find one today. Compared to a $500 CAD PS5, consoles have a monumental price advantage today.

I really hope PC component prices come back to the glory days of the early-mid 2010's.

I paid more than 3x the price of my PS5 for a 3080 Ti FE. ; _ ;
PS5 contrary to PS4 has great capabilities for it's price.
 

Tagyhag

Member
Oct 27, 2017
12,497
Because the game wouldn't even be able to take advantage of it on PC without directstorage I suppose. You saw Will's PC with a 7.5GB/s NVMe produced no additional time savings in loading

Yeah that makes sense, either way it's really cool to see the hardware finally being utilized correctly.
 

Tora

The Enlightened Wise Ones
Member
Jun 17, 2018
8,640
Yeah that makes sense, either way it's really cool to see the hardware finally being utilized correctly.
I'm so excited for the future. Lots more to come before the consoles have truly been utilised to the full extent of their capabilities.
 

Temperance

Member
Oct 25, 2017
5,808
[NO 2FA]
That's not a particularly fair comparison for loading times though?

PS5 NVME is 5.5GB/s max I think, and my own computer's NVME built last year has 5.0GB/s max so using a NVME with a max throughput of 3.5GB/s surprised me.
The way current loading on PC is done you are more likely to be bottlenecked by the CPU then by the read speed of current day drives.
 

Expy

Member
Oct 26, 2017
9,862
PS5 version holds up damn well. Already bought it, just haven't gotten to it yet since I was focusing on HFW and ER. Excited to pick it back up though.
 

Yudoken

Member
Jun 7, 2019
812
The depth of field is just a toggle and really needs a higher resolution option. It looks extremely low res on 4k with dlss quality.
 

Piggus

Member
Oct 27, 2017
4,699
Oregon
This is cool, but it's one game, and y'all making the 3060 Ti comparison can chill lol. The 3060 Ti is a >16 TF card and more like a 2080 Super (not counting the RT performance or DLSS). When I had my 3060 Ti it performed a lot better than my PS5 (or Series X in DF videos) in the exact same games (including those without RT).

That said, it's obvious at this point that the PS5 punches above its weight.
 

LCGeek

Member
Oct 28, 2017
5,857
Impressive that the PS5 is around a 2080's power level, especially at $399.

On a subsidized console not like consumers got the discounts sony got for their chips.

Yeah it's clear the game uses PS5's IO hardware, as most third party games still load faster on PC. Will be interesting to see how much of a different DirectStorage will make in future.

Not enough based on tests with it.

MS needs to change it IO stack, which is a fat chance in hell for it to really benefit. GPU decompression potential makes me happy but ideas that aren't due to hit for years aren't helping in the mean time.
 

RivalGT

Member
Dec 13, 2017
6,397
The tflop comparison is interesting, since AMD has never really been in the same ball park as Nvidia when it comes to 1 to 1 comparison IE 10tf vs 10 tf. In this specific game PS5 is closer to nvidia, vs AMD. I would have like to have seen the newer 6000 series cards in the benchmarks.
 

jroc74

Member
Oct 27, 2017
28,995
Agree with those saying the loading is interesting.

DirectStorage is coming out just in time for PC.

I am curious tho why arent comparisons being done against an RDNA 2 gpu. Early on it was because they weren't out yet. But now...

Honestly kind of insane to me, the type of GPU performance we're seeing out of a box that starts at $399.
Basically.

I get it that new technology comes out and ppl want consoles to take better advantage of it, but I am more than content with these consoles. I would probably skip any mid gen refreshes...maybe....lol.

The tflop comparison is interesting, since AMD has never really been in the same ball park as Nvidia when it comes to 1 to 1 comparison IE 10tf vs 10 tf. In this specific game PS5 is closer to nvidia, vs AMD. I would have like to have seen the newer 6000 series cards in the benchmarks.
Yes, I am not alone, lol.
 
Last edited:

LCGeek

Member
Oct 28, 2017
5,857
They also tried one that was around 7.5GB/s and it still didn't load as fast as the PS5, but was very close.

They could try a ramdrive or a decked out raid setup the result would be the same.

Windows IO Stack sucks balls. Someone mentioned it in other topic great for servers not for gamers.
 
Oct 25, 2017
9,872
Didn't expect them to do a video on this but I appreciate it.

It's cool that we're confirming that the PS5 GPU really is performing almost exactly as we expected before it was even launched- around a 2080 Super.

Edit: I noticed the DLSS differences from the original release when I played the intro last night. Glad to see it confirmed
 
Last edited:

vixolus

Prophet of Truth
Member
Sep 22, 2020
54,483
Sony knew this video was releasing today to show the PS5 off in a good light right before announcing their acquisition of Kojima Productions. It's all coming together! Greg Miller did it again.

2080 Super perf is crazy tho, very well optimized on PS5.
 

McFly

Member
Nov 26, 2017
2,742
A small but relevant correction on SmartShift usage on PS5

PS5 never shifts power based on how stressed or not stressed the GPU is. Both CPU and GPU have enough power to run at their max simultaneously. Because of the nature of games and software in general, Smartshift is used to move unused CPU power to the GPU if the CPU does not need it.

Cerny explains in his presentation. "While we're at it, we also use AMD's SmartShift technology and send any unused power from the CPU to the GPU so it can squeeze out a few more pixels."

In short, the idea is that developers may learn to optimise in a different way, by achieving identical results from the GPU but doing it faster via increased clocks delivered by optimising for power consumption. "The CPU and GPU each have a power budget, of course the GPU power budget is the larger of the two," adds Cerny. "If the CPU doesn't use its power budget - for example, if it is capped at 3.5GHz - then the unused portion of the budget goes to the GPU. That's what AMD calls SmartShift. There's enough power that both CPU and GPU can potentially run at their limits of 3.5GHz and 2.23GHz, it isn't the case that the developer has to choose to run one of them slower."
 

pswii60

Member
Oct 27, 2017
26,673
The Milky Way
Not enough based on tests with it.

MS needs to change it IO stack, which is a fat chance in hell for it to really benefit. GPU decompression potential makes me happy but ideas that aren't due to hit for years aren't helping in the mean time.
I haven't seen those DirectStorage tests yet, please can you link me? Will be interesting to see. Although I believe it's not using the direct to GPU and GPU decompression stuff yet, so not sure what it even has at this stage.