40%-ish performance difference from motherboard swap?

[Tea-Sir]

I have just had a really trippy thing happen.

I have an ASUS ROG STRIX B650E-F Gaming WiFi motherboard, Ryzen 7700X CPU and RTX 4080 GPU (Gigabyte Gaming OC variant).

I have been tired of ASUS bloatware for a while and decided to try and swap for an AsRock B650E Taichi Lite.

 

I did a complete re-install of Windows. When I went from Asus to Asrock.

I tested with 3 games. And the performance difference is wild. These are average observed FPS. First on ASUS then on AsRock.

R6 Siege - 300-350 average to 240-250 fps average - GPU reported about 96% usage on both boards

F1 22 - 120-150 average to 80-90 average - GPU reported 90-95% on ASUS and about 40-50% usage on Asrock,

Ratchet and Clank: Rift Apart - 120 average fps to 40-60 fps average. GPU reported about 90-95% on ASUS and 40-50% usage on Asrock

 

I quickly reinstalled my ASUS board again, without reinstalling Windows, just making sure my drivers were in order. And the FPS came right back up.

Boost clocks for the CPU is around 5.2 - 5.3 all core while gaming on both boards, and around 5 GHz all core under Cinebench R23. And the two boards are within 500 points of each other in the scores.

 

Has anyone ever experienced anything like that, or could tell me what I did not change in the BIOS?

Both BIOS are left to defaults, other than loading the memory profile. I checked and both boards were set to Auto for PBO.

 

The AsRock board is on its way to the seller again, so I will not be able to test anything. But would still like to learn, if anyone has an explanation for this.

(I tried switching to a Gigabyte AORUS B650E previously, I experienced the same FPS drop in Siege, but returned it due to poor network performance, so I didn't do any further testing.)

 

All parts but the motherboard where the exact same parts.

 

[podkall]

Did you re-enable D.O.C.P/EXPO + PBO + whatever on Asrock, you know bios settings that are saved on MB?

[Tea-Sir]

I confirmed my memory was running at the correct speed and timings and PBO was set to Auto, and I checked the same settings when I switched back to ASUS.

[Alex Atkin UK]

Does not surprise me, ASUS are known to be aggressive with default settings.  There's no doubt some difference in the CPU tuning you are missing.

 

Also not sure what you mean by ASUS bloatware, none of the ASUS software is mandatory.

[Uberwolf]

Nice to have 2 current motherboards on hand. Giving your test results one would be steered towards the Asus board for sure except there could be settings you might be overlooking, 100FPS is a huge difference and to answer your question, in my many years building systems I have never seen anything quite like those numbers, the difference that is. However the fact there is such a difference in the usage of the GPU is puzzling to me, I suspect this has more got to do with settings that actual real life, interesting. On the Asus bloat I would agree, we know it's not mandatory but that does not mean Asus do not include it or we appreciate having to police it as soon as we turn on our system.

[maartendc]

I don't know, that doesn't sound plausible at all.

 

Even if it was something related to BIOS settings (memory tuning, auto overclocking like PBO, SAM) you wouldn't see 40% difference. Any of these settings would account for like 5-10% difference in performance at most.

 

I feel like something was wrong with this Asrock board. Either overheating on the chipset / VRM's? Which could throttle the CPU under load?

 

Or something wrong with the PCI-e slot / settings? Causing the GPU to not be fully utilized?

 

Also, did both boards run latest BIOS version (same AMD AGESA version number)? It could be some kind of performance bug that AMD fixed in a newer AGESA version? Especially in the earlier days of the ZEN architecture, newer AGESA versions were known to fix memory compatibility issues, etc.

 

[Tea-Sir]

9 hours ago, Uberwolf said:

100FPS is a huge difference and to answer your question, in my many years building systems I have never seen anything quite like those numbers, the difference that is.

I have been building since 2009, many PCs. I have never seen anything like that before either. Which is why I am sure I must have missed over overlooked something.

8 hours ago, maartendc said:

I don't know, that doesn't sound plausible at all.

Yeah I had a hard time believing my eyes too.

 

8 hours ago, maartendc said:

I feel like something was wrong with this Asrock board. Either overheating on the chipset / VRM's? Which could throttle the CPU under load?

I didn't see anything overheating on any temps in HWinfo, and the CPU didn't throttle.

 

8 hours ago, maartendc said:

Also, did both boards run latest BIOS version (same AMD AGESA version number)? It could be some kind of performance bug that AMD fixed in a newer AGESA version? Especially in the earlier days of the ZEN architecture, newer AGESA versions were known to fix memory compatibility issues, etc.

They did yes.

[maartendc]

11 hours ago, Tea-Sir said:

I have been building since 2009, many PCs. I have never seen anything like that before either. Which is why I am sure I must have missed over overlooked something.

Yeah I had a hard time believing my eyes too.

 

I didn't see anything overheating on any temps in HWinfo, and the CPU didn't throttle.

 

They did yes.

Interesting on the AGESA and the throttling.

 

Perhaps we will never know, but I would say that whatever was causing this, it was not normal behavior. It must have been some kind of hardware malfuction of some kind; or some kind of serious misconfiguration that was overlooked.

 

I mean, hardware reviewers often test different motherboards with the same chipset: "what is the best motherboard for XYZ chipset". The differences you see are usually within 5% of one another. So there is absolutely no way there is a 40% difference because the Asus board is "better".

[Tea-Sir]

11 hours ago, maartendc said:

Interesting on the AGESA and the throttling.

Just to clarify. Of course when I hit it with Cinebench, the temps to go 95c instantly.

I saw around 5,0 to 5,1 GHz all core on both motherboards.

 

I had some serious stuttering (I thought for sure the PC would crash) when I alt-tabbed in and out of my games, which I have never seen before, so maybe something was not right.

 

Anyway, thanks to anyone who replied to my thread.