Support NeoGAF

[Silent Viper]

Based on comments, majority of people have not watched the video or completely missed the point of the video. Smh.

 

[Mister Wolf]

martino

Guess that throws out the theory that the cpu isn't in use.

 

[Mobile Suit Gooch]

Small victories, guys. Yay!

 

[Fredrik]

They are identical there, if that is indeed what's killing sx performance, it was 45 fps each.

With near identical CPUs and a supposed CPU bound scene that’s exactly what we should expect. The same performance. I don’t understand where you’re going with this.
And if XSX is claimed to be CPU bottlenecked for dropping from 60 to 45 fps in this scene, why isn’t PS5 claimed to be GPU bottlenecked for not climbing from 45 to 60 fps in other scenes?

 

[The Hamburglar]

Are you saying it's not?

I’ve never seen anyone in tech benchmark a still photo . Why all of sudden are DF of all people benchmarking a photomode?

‘Triggered’

 

[wachie]

I’ve never seen anyone in tech benchmark a still photo . Why all of sudden are DF of all people benchmarking a photomode?

‘Triggered’

Anything and everything to get those "wins".

 

[ethomaz]

Guess that throws out the theory that the cpu isn't in use.

I can't bold but you can read "while debugging the game myself".
I don't know who you can believe but he is the modder that did the 60fps patch for Bloodborne.

 

[Defrop]

Spoiler

[/ SPOILER]

Spoiler

 

[CrustyBritches]

The 2 spots I've been able to check against DF's measurements for PS5 land spot on, or slightly above my Ryzen 5 1600+2060S OC. *NOTE* With Lower than Low LOD Distance, and half-resolution RT reflections compared to PC.

In particular, the control room scene where Alex discusses why >=60fps results must be discarded due to the 60fps cap. In this scene PS5 = 51fps, R5 1600+2060S OC = 51fps, XSX >= 60fps. That's ~17% advantage for XSX if we take the 60fps at face value.

This is what makes this benchmark quite useful to pinpoint exact performance.

I can't bold but you can read "while debugging the game myself".

He'll need to provide screens in the same manner I did.

 

[The Hamburglar]

Riky i left ‘triggered’ there knowing how predictably you’d use the triggered emoji. I guess you have 2 types of posts.
The bullet point list you always use including vrr or the triggered emoji.

 

[ethomaz]

The 2 spots I've been able to check against DF's measurements for PS5 land spot on, or slightly above my Ryzen 5 1600+2060S OC. *NOTE* With Lower than Low LOD Distance, and half-resolution RT reflections compared to PC.

In particular, the control room scene where Alex discusses why >=60fps results must be discarded due to the 60fps cap. In this scene PS5 = 51fps, R5 1600+2060S OC = 51fps, XSX >= 60fps. That's ~17% advantage for XSX if we take the 60fps at face value.

This is what makes this benchmark quite useful to pinpoint exact performance.

He'll need to provide screens in the same manner I did.

Ask him lol

http://twitter.com/manfightdragon

 

[CrustyBritches]

Ask him lol

 

[x@3f*oo_e!]

I can't bold but you can read "while debugging the game myself".
I don't know who you can believe but he is the modder that did the 60fps patch for Bloodborne.

This is going to get interesting ..........

I can read, and I'm not seeing the name 'Lance MacDonald' on the game's credits .. https://www.mobygames.com/game/windows/control/credits

[edit] going from this post https://www.neogaf.com/threads/df-c...acing-benchmark.1590032/page-8#post-262235387

Looks like 20% CPU (up to 40) across 12 threads (I assume 6 cores hyperthreaded), but thread #11 is 80% (so basically full utilisation if hyperthreaded) ..

.. we might be looking at a single thread bottle neck (or maybe CrustyBritches forgot to turn his bitcoin miner off ??)

 

[ethomaz]

Wut?

This is going to get interesting ..........

I can read, and I'm not seeing the name 'Lance MacDonald' on the game's credits .. https://www.mobygames.com/game/windows/control/credits

Wut?

 

[isoRhythm]

Truth to be told, with Direct X layer, it's always going to be trailing behind PS5. And in Control where you have pretty advance physics, which requires tons of Object References (because that's how you do the interaction), which are expensive behind Dx, it's going to be on CPU side of things behind PS5. That's why Xbox always have higher clocked CPU, it needs it.

This was honestly a really interesting video and response. Do you think smart shift in the PS5 is what allows it to have a more stable framerate?

PS5 seems to do alot better in games that utilize DRS (Valhalla) but loses in games with set resolutions (Hitman 3) which might be because of how fast smart shift can allocate resources between CPU and GPU which make DRS transitions and gameplay more smooth.

It's also ironic that PS5 will take 60fps crown (MS marketing of high framerates) and XBsx will take photo mode crown (a trend PS games started)

 

[CrustyBritches]

Wut?

 

[x@3f*oo_e!]

Ugh different distance LODs. Back to the drawing board boys.

Close the thread lol

 

[ethomaz]

Man I have no ideia what are you trying to say.

You are saying he is lying... so ask him... at least he was a way better background to say what he said than you from what we know.

 

[J_Gamer.exe]

With near identical CPUs and a supposed CPU bound scene that’s exactly what we should expect. The same performance. I don’t understand where you’re going with this.
And if XSX is claimed to be CPU bottlenecked for dropping from 60 to 45 fps in this scene, why isn’t PS5 claimed to be GPU bottlenecked for not climbing from 45 to 60 fps in other scenes?

Then if it is the cpu you have answered your own question, they aren't the same.

There's more than just 'being zen 2' and clock speed.

Customisations like caches which ps5 is rumoured to have improved.

And with the gpu are you for real, its always been expected the sx to have the gpu advantage. Cant have it both ways, either it is or it isn't.

But with just one last gen port we dont know whats going on, how much effort put in to each platform, which platform the engine prefers etc.

 

[Werewolfgrandma]

This is just almost interesting, but not.

 

[x@3f*oo_e!]

Man I have no ideia what are you trying to say.

You are saying he is lying... so ask him... at least he was a way better background to say what he said than you from what we know.

Literally CPU traces. https://www.neogaf.com/threads/df-c...acing-benchmark.1590032/page-8#post-262235387

It's lowish, except on 1 thread, but nothing like neglible. That would be 100% everything on PS4..

 

[CrustyBritches]

Looks like 20% CPU (up to 40) across 12 threads (I assume 6 cores hyperthreaded), but thread #11 is 80% (so basically full utilisation if hyperthreaded) ..

.. we might be looking at a single thread bottle neck (or maybe CrustyBritches forgot to turn his bitcoin miner off ??)

Please clarify. It's totally common for multi-threaded games to have a heavily used single thread/core for primary game usage, while the remaining cores are largely unused. This why a simple 4-core can match 8-core in heavily single-threaded titles.

Difference between the 4c/4t, 4c/8t, and 6c/12t in this game is very minimal. To double check this I'm in the process of disabling 2c/4t on my CPU and test that spot again to see what difference resolution makes in CPU thread usage.

 

[JimboJones]

I never used a photo mode to benchmark my new graphics cards

That's because the cast majority of pc games can just have their framerate uncapped. This is just an interesting quirk in the console game to get around the 30fps cap in the graphics mode.

 

[bender]

 

[x@3f*oo_e!]

Please clarify. It's totally common for multi-threaded games to have a heavily used single thread/core for primary game usage, while the remaining cores are largely unused. This why a simple 4-core can match 8-core in heavily single-threaded titles.

Difference between the 4c/4t, 4c/8t, and 6c/12t in this game is very minimal. To double check this I'm in the process of disabling 2c/4t on my CPU and test that spot again to see what difference resolution makes in CPU thread usage.

Yeah - from the original CPU traces it looked the main thread was not far off 100%ing cpu#11. So a little extra work could make that the frame rate bottle neck.

But specifically I was thinking about the "corridor of doom" (in DF foundry video )where both machines drop lowest and to the same frame rate. Idk why that particular region would be CPU heavy, but it's well known that it's common and heavy frame rate drop area.. Elsewhere I don't/didn't expect CPU bound. (Should have been clearer)

 

[TheAmazingSpiderMan]

was there any actual gameplay running or was the entire video in photo mode??

 

[ethomaz]

was there any actual gameplay running or was the entire video in photo mode??

Just Photo Mode.

 

[skneogaf]

It's great that we have a game that is confirmed to be exactly the same settings on xsx & ps5 and there is a way to unlock the fps and both are under 60fps.

It is the best benchmarking tool for consoles that we have available up to now.

Hopefully more games will have the same type of comparison ability.

 

[CrustyBritches]

Yeah - from the original CPU traces it looked the main thread was not far off 100%ing cpu#11. So a little extra work could make that the frame rate bottle neck.

Depending on how they handled multithreading, XSX could be at 3.8GHz without SMT, or 3.6GHz with SMT on Zen 2 cores without Windows overhead compared to my Ryzen 1600 @ 3.7GHz Zen 1 cores. Hmmm...I highly doubt that would be true in the majority of areas he showed.

We should test that area in few different configs. I'm only an hour or so into the game. I need to find it.

 

[Werewolfgrandma]

Spoiler

[/ SPOILER]

Spoiler

Not a single one of those pics is like for like. Could be lods? Xbox does look father back in the bottom 2. Top one is to dark to tell on my phone.

 

[martino]

Did you try "corridor of doom" yet (it's in Executive sector between Dead Letters and Communications I think) - I suspect something with in-game optimisation is wrong and it's trying to render parts of the map that it doesn't need to or something. .. On PS4 (in battle) framerate can go sub 5fps (no joke) - it's never that bad anywhere else. I don't see anything particularly complex with the scene at all compared to other places - in fact it's simpler than most

Would be interesting to see if CPU or GPU is the problem there..

i've been playing the whole area up to hotline chamber (fps being between 45-55 around 50+-1 fps most of the time on my config at 4k min settings no rt)
but I'm not sure where the corridor of doom is even looking at the video
it can also be related to scene content and rt effects.the little i played with my card (1080ti) and reflection rt effect the drops were big were i was near and looking at windows (like nearly cut in half)
of course my card do it the brute force way without the muscle

martino

i've watched cpu usage of my i7 7700k there is not a significant difference between playing an photo for me too in fight or not.
i had one core in 40-50% 2 in 30% and 4 other mostly under 20%

 

[TheAmazingSpiderMan]

Just Photo Mode.

Lol wth, that’s a weird take

 

[phil_t98]

Sorry don't get the point?

so the talk of bottlenecks but its running extremely similar to the PS5?

 

[Mister Wolf]

i've been playing the whole area up to hotline chamber (fps being between 45-55 around 50+-1 fps most of the time on my config at 4k min settings no rt)
but I'm not sure where the corridor of doom is even looking at the video
it can also be related to scene content and rt effects.the little i played with my card (1080ti) and reflection rt effect the drops were big were i was near and looking at windows (like nearly cut in half)
of course my card do it the brute force way without the muscle


i've watched cpu usage of my i7 7700k there is not a significant difference between playing an photo for me too in fight or not.
i had one core in 40-50% 2 in 30% and 4 other mostly under 20%

Apparently all of you guys are lying because some guy on Twitter said otherwise while providing zero evidence.

 

[x@3f*oo_e!]

We should test that area in few different configs. I'm only an hour or so into the game. I need to find it.

but I'm not sure where the corridor of doom is even looking at the video

iirc It's here (red line) Follow yellow brick road from starting area, through cafeteria, then dead letters - I think that should take you to a stairwell (at the bottom) - corridor is on the same level.

Occasionally you get mini-events in Communications dept. with the glowy orbs things (just kill everything) - this is typically when the frame rate totally tanks.

[edit] Right here :



@ 20sec don't go up the stairs, just down the corridor - she gets back there by 1:15

 

[FIREKNIGHT2029]

It's a valid GPU benchmark, not sure what you're laughing about to be honest.

Last time I checked, we run games in the PC world when we do benchmarks. This is the first time I've seen anyone do a photomode benchmark but I guess when you're scraping the bottom of the barrel for wins any "W" will do .

 

[CrustyBritches]

Apparently all of you guys are lying because some guy on Twitter said otherwise while providing zero evidence.

No disrespect towards him on my part. I presented ~1min CPU readout for each scenario, and I've tested combat. I'm more than open to have him explain the context of those results, but perhaps it's also possible it could be a difference in debug mode/dev kit vs retail, old version vs UE, etc.

 

[ethomaz]

Apparently all of you guys are lying because some guy on Twitter said otherwise while providing zero evidence.

Actually he is not somebody.

 

[Mister Wolf]

Actually he is not somebody.

Who is he? He work for Remedy? Is the monitoring software lying? Seeing is believing.

 

[Larlight]

Good job Xbox, you won this round. I guess if Playstation has interactive movies, Xbox has slideshows.

 

[ethomaz]

Who is he? He work for Remedy? Is the monitoring software lying? Seeing is believing.

He is a modder.
His more recent mod is Bloodborne 60fps patch.
He has works on others Souls games like Sekiro and Dark Souls 3.
He has some work in NiER Automata if I’m not mistaking too.

You can support him on Pateon if you wish: https://www.patreon.com/lancemcdonald

So do you remember Durante? He does basically what Durante did in the past... Durante is most know for fixing the framepace issues in Dark Souls.

 

[DeaDPo0L84]

I've always owned both consoles every gen so these threads served as a way to figure out which console to buy a 3rd party game on if the differences were drastic enough.

Now that I've moved mostly to pc these threads are pure entertainment, keep it up everyone.

 

[Mister Wolf]

He is a modder.
His more recent mod is Bloodborne 60fps patch.
He has works on others Souls games like Sekiro.
He has some work in NiER Automata.

So he is a hobbyist. When he can back up his statement with evidence I will believe him over hardware monitoring software just about everyone uses. You seem pretty pretty knowledgeable. Can you explain why the software isn't seeing a difference in performance demand on the CPU?

 

[ethomaz]

So he is a hobbyist. When he can back up his statement with evidence I will believe him over hardware monitoring software just about everyone uses. You seem pretty pretty knowledgeable. Can you explain why the software isn't seeing a difference in performance demand on the CPU?

Yeap it is better to believe he is lying lol
Random posters calling somebody that works with performance enhancement in games to be a liar lol

Let’s I get my notebook to take some Windows task manager screens with 100% CPU... maybe you will believe me over him.

 

[Larlight]

This was honestly a really interesting video and response. Do you think smart shift in the PS5 is what allows it to have a more stable framerate?

PS5 seems to do alot better in games that utilize DRS (Valhalla) but loses in games with set resolutions (Hitman 3) which might be because of how fast smart shift can allocate resources between CPU and GPU which make DRS transitions and gameplay more smooth.

It's also ironic that PS5 will take 60fps crown (MS marketing of high framerates) and XBsx will take photo mode crown (a trend PS games started)

For one game lol. If it become more consistent, then sure.

 

[x@3f*oo_e!]

He is a modder.
His more recent mod is Bloodborne 60fps patch.
He has works on others Souls games like Sekiro and Dark Souls 3.
He has some work in NiER Automata if I’m not mistaking too.

You can support him on Pateon if you wish: https://www.patreon.com/lancemcdonald

So do you remember Durante? He does basically what Durante did in the past... Durante is most know for fixing the framepace issues in Dark Souls.

I guess he was just exaggerating for effect in the tweet. I believe that when destruction happens CPU (or GPU load) must increase substantially.

 

[Bo_Hazem]

Xbox Series X is the most powerful in Control... Only in photo mode.

In gameplay? Well, performance dips to 40's fps. Well, when it went to 45fps DF cut the clip out:

This only shows that the system has more bottlenecks that kill the potential of the GPU. Both in Graphics and Performance modes PS5 has the upper hand, as always.

 

[ethomaz]

I guess he was just exaggerating for effect in the tweet. I believe that when destruction happens CPU (or GPU load) must increase substantially.

I don’t know the comment was on YouTube btw.
On the video linked in the OP.

 

[x@3f*oo_e!]

This was honestly a really interesting video and response. Do you think smart shift in the PS5 is what allows it to have a more stable framerate?

Thing is - Series X CPU is at least as good as PS5 (100MHz higher?), and afaik Smartshirt doesn't allow temporary upclock..

If smartshift allowed temporary overclock for CPU I would buy this. But afaik it doesn't

Maybe it does upclock ???

 

[JimboJones]

Last time I checked, we run games in the PC world when we do benchmarks. This is the first time I've seen anyone do a photomode benchmark but I guess when you're scraping the bottom of the barrel for wins any "W" will do .

Are you being dense on purpose?