Question Ryzen 3600 Underperforming Heavily after system changes

Apr 4, 2020
9
1
15
Hi all,

My setup is as follows:
Ryzen 3600
MSI B450m Mortar MAX
2x8 GB 3600Mhz Corsair RGB Pro
RX 580
Corsair H100i
Cooler Master MWE Gold 650W

Recently I've purchased an RX 5700 XT. After installing its drivers I decided to run some benchmarks, and it turned out that my CPU suddenly underperforms heavily. Cinebench R15 score dropped by 40%, and userbenchmark says the CPU is in the 1st percentile of lows (was 70th percentile previously). In the mean time i've also relocated my radiator from the top to the front of the case, but while doing so the cooling block stayed attached.

To do some testing, I've switched my GPU back to an RX 580 with which the previous benchmarks were ran. However, problems still persist. After that I tried reseating the CPU, but this also didn't improve anything. I've never overclocked my CPU. Any ideas what might be causing this?

Temperatures have gone from 75C while benchmarking to 50C with the radiator change, don't know if that has anything to do with it..

Thanks a lot in advance!
 
Just ran it again, runs steady at 2.8GHz with a maximum temp of 48C. In idle it runs at about 38C.

Edit: If i remember correctly the standard clock is 3.6GHz. Could it be that the new AMD driver clocked it down?
 
....

Recently I've purchased an RX 5700 XT. After installing its drivers I decided to run some benchmarks, and it turned out that my CPU suddenly underperforms heavily.
...
Have you updated motherboard BIOS to latest?

Also, update GPU drivers for the new GPU. Even though moving from an RX580 still do a clean install - uninstalling the old drivers and use DDU to clean up - before installing the latest AMD drivers.

Then re-install the latest AMD chipset drivers and run the Ryzen balanced plan.

Also: have you made any BIOS settings changes?
 
Have you updated motherboard BIOS to latest?

Also, update GPU drivers for the new GPU. Even though moving from an RX580 still do a clean install - uninstalling the old drivers and use DDU to clean up - before installing the latest AMD drivers.

Then re-install the latest AMD chipset drivers and run the Ryzen balanced plan.

Also: have you made any BIOS settings changes?

The motherboard BIOS was already updated to the latest version before the swap. I did not use DDU, but did install the latest drivers ( I think they're the exact same). Have not made any BIOS changes. Will try using DDU and re-install the drivers. Thanks!
 
The motherboard BIOS was already updated to the latest version before the swap. I did not use DDU, but did install the latest drivers ( I think they're the exact same). Have not made any BIOS changes. Will try using DDU and re-install the drivers. Thanks!
Be sure to re-install the chipset drivers after DDU because it removes them.

The reason I like to use DDU is to clean up old settings in the registry that may affect how the new gpu operates.
 
Just used DDU to uninstall and re-installed the drivers, but the problem still persists. Still wondering about the low clock-speed, could that have anything to do with it? It also stays at 2.8GHz all the time, doesn't clock down either.
 
Just used DDU to uninstall and re-installed the drivers, but the problem still persists. Still wondering about the low clock-speed, could that have anything to do with it? It also stays at 2.8GHz all the time, doesn't clock down either.
Being stuck at 2800 really sounds like something configured strange in BIOS. But at least you know you've got a clean GPU and chipset driver install.

So the next thing is do a CMOS reset. Save a profile if you like what you got and want to come back to it. If it works, you can go back and write down fan profiles, memory timings and whatever else is important...or do that first and don't go back.

As for CPU settings: leave mult and vcore on AUTO, both of them. Set these to enabled: CoolnQuiet, Global C States, Processor CPPC and CPPC Preferred Cores.
 
Being stuck at 2800 really sounds like something configured strange in BIOS. But at least you know you've got a clean GPU and chipset driver install.

At least one thing to exclude! Just did a CMOS reset and left the previous RAM timings alone for now, so running on factory default 2666MHz. Made the changes you proposed in BIOS, but it's still stuck at 2.8GHz (even though BIOS thinks it's at 3.6MHz).
 
...but it's still stuck at 2.8GHz (even though BIOS thinks it's at 3.6MHz).

I've had that happen with my B450m Mortar if I manually overclock (setting the multiplier off AUTO) and then turn on CoolnQuiet. CnQ would automatically disable whenever I took the multiplier out of AUTO so I'd have to go and turn it back on to do that. But that's a BIOS thing...if you've reset CMOS it should all be at defaults now.

Are you running any utilities in Windows? like iQue or MSI's Dragon software? I'd uninstall all of those if possible. Many of them do strange things.
 
Hmm... So after uninstalling MSI's Dragon Center / iCue & doing a CMOS reset it seems to have worked! Cinebench scores are back to previous levels again, with hardwaremonitor showing variable clock speeds again (2.2GHz lows going up to 4.1 boost while benchmarking). Userbenchmark still thinks it's <Mod Edit>, but trusting Cinebench & numbers more I'd say it's fixed!

Concluding, probably Dragon Center that was messing around? With iCue you are unable to see / set clock speeds.
 
Last edited by a moderator:
Hmm... So after uninstalling MSI's Dragon Center / iCue & doing a CMOS reset it seems to have worked! Cinebench scores are back to previous levels again, with hardwaremonitor showing variable clock speeds again (2.2GHz lows going up to 4.1 boost while benchmarking). Userbenchmark still thinks it's <Mod Edit>, but trusting Cinebench & numbers more I'd say it's fixed!

Concluding, probably Dragon Center that was messing around? With iCue you are unable to see / set clock speeds.
I think yes, Dragon Center was messing with that. But iCue has also been strange in the past it's prevented the processor from coming back to idle clocks, which adversely affected power consumption.

Userbenchmark is completely unreliable for comparison. Both because they're biased and because the database used for comparison is heavily skewed by overclockers making multiple runs with really high performing systems to try and take...and re-take....top spot in competitions.

At any rate...it's good to see you've found the problem!
 
Last edited by a moderator: