[SOLVED] FPS Drop in Warzone after 5800x upgrade from 3700x - 1440p

[warzone]

Strange but true, in Warzone I have less FPS going from 3700x to 5800x. I don't know where the problem is. While in COD Cold War I see peaks of 200 fps (multiplayer) when I barely had 125 before (so there was a big jump here). In Modern Warfare Multiplayer, slight improvement but nothing special.

What I've done so far:

• updated the BIOS of the Tuf x570 Plus to support Ryzen 5000
• activated XMP profile for the ram (3200 Mhz) CL16
• tried different "energy saving" setups, all tried without changes
• BIOS precision boost enabled
• sent the 5800x to 4.8 Ghz without improvements
• reinstalled fresh nvidia drivers
• changed core count in Modern Warfare config file to 8

The 3070 works as always, I have no problems. On YT I see benchmarks with a setup almost identical to mine with peaks of 200 FPS in 1440p, I don't even see 144 fps.

I play only Warzone / Rebird and I'm so upset about the overall Warzone situation. I can't get 144 stable fps with DLSS set on Performance Ultra. Something is going on on my build.

My complete setup:

• Ryzen 5800x
• RTX 3070 Asus TUF Gaming
• ASUS TUF X570-plus
• Crucial 3200 MHz - 2x 8GB
• Antec 650 W Gold

Warzone settings:

• 1440p @ 144 hz
• Low on all graphic settings
• DLSS Performance Ultra
• Nvidia Reflex: ON
• Nvidia GeForce Experience Filters (Sharpness / Details / Color / Contrast & Brightness -> 10 fps impact)

 

[John Chesterfield]

I'm confused. The title says FPS drops, then you say a big improvement in Cold War and then a slight improvement in Modern Warfare. You then cite YouTube videos where you don't know what settings they are using in comparison to yours, if you do know you haven't mentioned it, saying you're getting less than them.

Do you have any actual numbers from your old set up to compare? Because from where I'm sat, you've actually had a slight improvement in FPS in one game, a decent one in another and a load of YouTube videos where you've no real idea what settings they're running at.

Not possible to drop the 3700x back in, take some numbers and then do the same again with the 5800x?

 

[kurdtnz]

Can you run userbenchmark and copy and paste a link to your results please?

 

[drea.drechsler]

..
What I've done so far:

• updated the BIOS of the Tuf x570 Plus to support Ryzen 5000
• activated XMP profile for the ram (3200 Mhz) CL16
• tried different "energy saving" setups, all tried without changes
• BIOS precision boost enabled
• sent the 5800x to 4.8 Ghz without improvements
• reinstalled fresh nvidia drivers
• changed core count in Modern Warfare config file to 8

....

Did you ever reset CMOS?

....
Not possible to drop the 3700x back in, take some numbers and then do the same again with the 5800x?

Not sure how that would be all that informative since the 3700X would have to be working with a new BIOS and AGESA, introducing too many variables.

In the end, the gaming performance uplift from Ryzen 3000 to Ryzen 5000 was never all that great and game performance is still constrained more by GPU than CPU. You may be just about where it's supposed to be with a game that's more CPU limited showing greater performance uplift...and a game that's more GPU limited lesser.

Try running a CPU benchmark to test it's performance in real world processing task: Cinebench20 or 23 would be the best one that's repeatable, comparable and not easily 'gamed'. Being comparable and not easily gamed is important since you probably lack results from your system with the 3700X and older BIOS to compare to.

UseBenchmark is gamed badly and a terrible way to compare since you'll end up comparing to people with heavily overclocked systems they run multiple times to get high scores. It's only good comparing to your own system and, again, it's doubtful you have any scores from when you had the 3700X and older BIOS in it.

 

[warzone]

Thanks for your questions.

After several hours of research I tried the Passmark benchmark for the CPU (it lasts a couple of minutes, even less) and in Ryzen Master I saw that its temperatures rise to 90 degrees for a few seconds. The Passmark score is even more than average for 5800x (28.725 vs global avg. 28.492).

I am using an Enermax T50 heatsink, definitely underperforming for the 5800x which, compared to the 3700x, absorbs 220 W under stress. Could this be the cause of the problem in Warzone?

In Cold War I see an improvement in FPS but actually the two games could use the CPU in a different way.

For comparison, with the 3700x I had 120 fps on average in Rebird Island (Warzone Mode), while with the 5800x I had 110 fps on average. I seriously think this is a poor dissipation problem. Could an inadequate heatsink lead to a cut in FPS and in-game performance? The PC does not shut down in game.

 

[Phaaze88]

If you'll notice in Ryzen Master, it's not pulling that much power at all.
It's definitely NOT the cpu cooler. The Ryzen 5800X has problems under the hood that the cooler can't do anything about. It's neither yours, or the coolers fault that the heat isn't being properly transferred up through the IHS and into the cooler base plate.
There are MANY threads/videos on this cpu's notoriously high thermals.
For a number of reasons, this cpu is the hottest in the 5000 stack and has become a sort of odd-man out because of it.

 

[Phaaze88]

Silly me - there's even a thread like this over in TPU: https://www.techpowerup.com/forums/...-3-chips-pbo-settings-temperature-fix.283776/

 

[warzone]

My problem sounds like this:

View: https://www.youtube.com/watch?v=I7qDudI8Tpo


Looking for Bios update correlation, I updated my Bios from Version 4005 to the latest version (out by August 17th - 2021):

Version 4021
TUF GAMING X570-PLUS BIOS 4021
"1. Update AMD AM4 AGESA V2 PI 1.2.0.3 Patch C
2. Improve system performance"

But I had zero luck. In the meanwhile I ordered a 5900x hoping to jump over these problems. I'm not an expert but this smells like wrong Bios setting / update thing.

Did you ever reset CMOS?

No, I just discovered how to do it on my x570 Tuf Gaming. I will retry with the 5900x I just ordered.

In the end, the gaming performance uplift from Ryzen 3000 to Ryzen 5000 was never all that great and game performance is still constrained more by GPU than CPU. You may be just about where it's supposed to be with a game that's more CPU limited showing greater performance uplift...and a game that's more GPU limited lesser.

Ryzen 5000 showed great improvement over 3000 in gaming (according to YT benchmarks from trusted sources like Hardware Unboxed & Gamers Nexus).