Cyberpunk 2077 Ultra performance

[Wrinkly]

There is nothing odd about it.
-You don't have an answer to it
-He refuted your claim.

That's an odd one. I can only imagine AMD are using a RT process at a lower resolution then smoothing the edges. It's not a game that RT use stands out in. I think they did well again with shadows in SOTTR? I get the feeling that Nvidia takes a belt and braces approach so far and just RTs the whole area required, hence their 'It just works' approach.

 

[SaintJimmy]

That can’t be right. My 4790k pushes my rtx 2080 to 99% usage same res. So your 7700k should be chillin.

This is with an RTX 3080, with DLSS off it's GPU bound but with it on it becomes CPU bound.

 

[gpuerrilla]

DLSS was marketed so agresively Nv and most YT influencers would have you believe it's better than native, but it is what it is, imo it's a better upscaling but it's never been better than native.

@TNA knows its better than native, now he can prove it with his 3080FE!!

 

[gpuerrilla]

Ahhh my bad and thought they were doing it via software and normal cores Thanks for the heads up.

Still thinking of going 6900XT greg?

 

[Gregster]

Still thinking of going 6900XT greg?

No bud and truthfully, New processor (Ryzen 5) with new mobo and mem will do me this round and wait it out for the next GPUs. If I do bite at all, it would be the 6800 XT or the 3080 though but doubt that will happen.

 

[Grim5]

It's a hybrid approach. There is one fixed function Ray Tracing unit per CU. It's hardware accelerated. But, it's a repurposed Texture Processor. So that Unit does the Ray tracing near the beginning of the pipeline and switches over to do the texturing at the end of Pipeline.

And there is is why it's so much slower.

With Ampere cards the RT unit does the processing in the pipe line at the same time as the texture units do their work - Concurrent RT+Shading work. The benefit of this approach is that RT does not hold up shading, this change roughly double Ampere's RT performance. With RDNA 2, if there is too much to do within the frame, the RT processing at the start of the pipeline holds up all the shading until the RT is done and then the shading kicks off - because of this frametimes spike up and framerates drop.

This is still low hanging fruit stuff, easy fixes for quick gains at the start of the ray tracing Era. I'd expect RDNA 3 will also introduce concurrent RT+Shading and therefore at least double it's own ray Tracing performance.

 

[CAT-THE-FIFTH]

it's pushing the envelope more than crysis did.

I think you really don't seem to appreciate what Crysis did,and it was not only just the graphics,but the whole interactable semi-open world environment(and the ability to destroy so many objects),the way unlike so many modern games it used multiple forms of AI models(both bipedal and flying enemies),etc and the way you could tailor gameplay in an FPS. The fact is it was essentially DX9 game,and Crytek on purpose dialed down menu graphics settings - once you edited settings and ramped them up even more,you could see how forward looking it was. In fact,it was the pinacle of what DX9 could do and could still hold its own with later DX10/DX11 games to a degree.Its just that nowadays what it did was copied by many other games after it.

I would also say Cyberpunk 2077 is also much easier to run relative to Crysis. Even with a 8800 Ultra good luck with maxing out settings on the 1080p monitors of the era and expecting decent FPS. The game actually needed more than the common 512MB framebuffers on GPUs of the era,and even the most overclocked fastest CPUs,were a bottleneck.

I had an 8800GTS 512MB and a heavily overclocked Core2 CPU,and it still wasn't enough.

This for me has nowhere near the impact Crysis had at the time WRT to graphics. Crysis was a huge jump up in many areas,and Cyberpunk is a pretty game,just not quite what Crysis did back then.

This is partly because Cyberpunk 2077 is an intergenerational game. It has to be able to run on the older consoles and older PCs,unlike Crytek who didn't give a damn on whether older systems could actually run it.

Literally if you had anything before the 8800 series which was released the year before,you might as well have not bothered. If you didn't have an overclocked Core2 then you were hitting even worse CPU limitations.Cyberpunk 2077 still runs OK on much older hardware.

Crysis will be remembered for its graphics,Cyberpunk 2077 won't be as much just like the Witcher 3 won't be. It will live or die on its RPG experience. If anything I think Atomic Heart might be closer to a real next generation graphical experience from the demos I have seen.

 

[Minato]

Something just isn't right.

RTX 3090 FE, now overclocked (core voltage % + 100, power limit % 114, Temp Limit 90, Core Clock + 130MHz, Memory Clock +300MHz)

Playing at 4K, all settings max, RTX on ultra, DLSS Quality. Getting 40ish fps in Night City during the day, dipping to 30 or even 28!!! FPS outside V's apartment. How is everyone with the exact same card and settings getting 60+fps average? It makes no sense to me.

And yes, windows power set to performance, latest drivers installed etc. Chromatic aberration, motion blur, film grain etc trash turned off. Timespy graphics score with above is 20099 (https://www.3dmark.com/3dm/54619998?)

 

[Grim5]

The game runs like total ass on base consoles

PS4 = 720p resolution, all settings on Low and drops down to 15fps

https://wccftech.com/cyberpunk-2077-base-ps4-xbox-one-bad-graphics-day-1-patch-no-fix/

 

[Besty]

Something just isn't right.

RTX 3090 FE, now overclocked (core voltage % + 100, power limit % 114, Temp Limit 90, Core Clock + 130MHz, Memory Clock +300MHz)

Playing at 4K, all settings max, RTX on ultra, DLSS Quality. Getting 40ish fps in Night City during the day, dipping to 30 or even 28!!! FPS outside V's apartment. How is everyone with the exact same card and settings getting 60+fps average? It makes no sense to me.

as it’s an FE are you getting heat soak?

 

[Minato]

as it’s an FE are you getting heat soak?

I'm not sure what that is, but according to MSI afterburner temps never exceed 73 degrees (even tho temp limit set to 90)

 

[StevoIRL]

Something just isn't right.

RTX 3090 FE, now overclocked (core voltage % + 100, power limit % 114, Temp Limit 90, Core Clock + 130MHz, Memory Clock +300MHz)

Playing at 4K, all settings max, RTX on ultra, DLSS Quality. Getting 40ish fps in Night City during the day, dipping to 30 or even 28!!! FPS outside V's apartment. How is everyone with the exact same card and settings getting 60+fps average? It makes no sense to me.

And yes, windows power set to performance, latest drivers installed etc. Timespy graphics score with above is 19749

What CPU have you got? The feeling is that the game is more CPU bound than we thought.

 

[opethdisciple]

How is a Ryzen 3700x faring in this game?

 

[TNA]

Something just isn't right.

RTX 3090 FE, now overclocked (core voltage % + 100, power limit % 114, Temp Limit 90, Core Clock + 130MHz, Memory Clock +300MHz)

Playing at 4K, all settings max, RTX on ultra, DLSS Quality. Getting 40ish fps in Night City during the day, dipping to 30 or even 28!!! FPS outside V's apartment. How is everyone with the exact same card and settings getting 60+fps average? It makes no sense to me.

And yes, windows power set to performance, latest drivers installed etc. Timespy graphics score with above is 19749

Turn off all the basic settings for starters. You image quality will get better at the very least.

 

[Minato]

What CPU have you got? The feeling is that the game is more CPU bound than we thought.

Ryzen 3600. I don't think I should be CPU bound at 4K. MSI afterburner shows GPU usage at 100%, CPU all cores usage not exceeding 70%

Turn off all the basic settings for starters. You image quality will get better at the very least.

Sorry yes obviously disabled trash like motion blur, chromatic aberration etc - goes without saying.

 

[TNA]

Ryzen 3600. I don't think I should be CPU bound at 4K. MSI afterburner shows GPU usage at 100%, CPU all cores usage not exceeding 70%



Sorry yes obviously disabled trash like motion blur, chromatic aberration etc - goes without saying.

Maybe try dlss on performance. There actually is not as much of a difference between performance and quality which I am surprised to be honest.

 

[CAT-THE-FIFTH]

How is a Ryzen 3700x faring in this game?

I have a GTX1080FE and a Ryzen 7 3700X with 32GB of 3600C18 DDR4.

At mostly ultra settings(also enabled a 60FPS framecap),I am getting between 35~60FPS(top end is inside corridors,but hovers in the 40 to 50FPS range outside) at 1080p. Its closer to 30FPS at qHD with the same settings outside,and closer to 40FPS at medium settings. It's still quite smooth either way.

 

[gpuerrilla]

No bud and truthfully, New processor (Ryzen 5) with new mobo and mem will do me this round and wait it out for the next GPUs. If I do bite at all, it would be the 6800 XT or the 3080 though but doubt that will happen.

Well it may drop to some tempting prices next year so probably worth delaying as your current card is decent for a while.

 

[gpuerrilla]

Something just isn't right.

RTX 3090 FE, now overclocked (core voltage % + 100, power limit % 114, Temp Limit 90, Core Clock + 130MHz, Memory Clock +300MHz)

Playing at 4K, all settings max, RTX on ultra, DLSS Quality. Getting 40ish fps in Night City during the day, dipping to 30 or even 28!!! FPS outside V's apartment. How is everyone with the exact same card and settings getting 60+fps average? It makes no sense to me.

And yes, windows power set to performance, latest drivers installed etc. Chromatic aberration, motion blur, film grain etc trash turned off. Timespy graphics score with above is 20099 (https://www.3dmark.com/3dm/54619998?)

Heard latest drivers arent good. I havnt installed them so couldnt input.

 

[StevoIRL]

I have no idea how those DLSS stats are indicating 50% performance increases, im seeing at most 5 FPS gains with it