F5F Stay Refreshed Hardware Desktop Score extremely poor with a Radeon RX 6900 XT graphics card

Score extremely poor with a Radeon RX 6900 XT graphics card

Score extremely poor with a Radeon RX 6900 XT graphics card

J
jkgaga15
Member
234
03-25-2022, 04:00 AM
#1
I’m trying to improve my card’s performance but it’s not responding well. My overall score is around 7700, with a GPU score near 7366 and a CPU score of 10,342. The system specs are: 64GB Vengeance RAM, Ryzen 5950X processor, Radeon RX 6900 XT graphics card, 1250W Platinum power supply, and an MSI Meg case with 12 fans. I’ve updated the GPU drivers, checked BIOS settings, and made sure no extra features are enabled. I’ve also increased fan speeds and adjusted power limits, but the card still struggles—temperatures are normal, yet the fans stop completely during testing. I’ve attached some screenshots showing the GPU and CPU speeds and temperatures throughout the session.
J
jkgaga15
03-25-2022, 04:00 AM #1

I’m trying to improve my card’s performance but it’s not responding well. My overall score is around 7700, with a GPU score near 7366 and a CPU score of 10,342. The system specs are: 64GB Vengeance RAM, Ryzen 5950X processor, Radeon RX 6900 XT graphics card, 1250W Platinum power supply, and an MSI Meg case with 12 fans. I’ve updated the GPU drivers, checked BIOS settings, and made sure no extra features are enabled. I’ve also increased fan speeds and adjusted power limits, but the card still struggles—temperatures are normal, yet the fans stop completely during testing. I’ve attached some screenshots showing the GPU and CPU speeds and temperatures throughout the session.

T
techiseasy
Senior Member
688
03-27-2022, 03:29 PM
#2
Conducted a FurMark 1080p benchmark; temperatures were acceptable and the score was 21,186, matching expected performance.
T
techiseasy
03-27-2022, 03:29 PM #2

Conducted a FurMark 1080p benchmark; temperatures were acceptable and the score was 21,186, matching expected performance.

E
ErikJanTV
Junior Member
16
03-29-2022, 01:36 AM
#3
Downloaded Unigine SuperPosition and tested it at 1080P Extreme settings, achieving a score of 11662. At lower resolutions, similar configurations appear in the results, though some scores reach around 18k. It seems both high and low scores are possible depending on the setup. The performance boost was noticeable—GPU usage peaked at 98% while global temperature stayed at 72%.
E
ErikJanTV
03-29-2022, 01:36 AM #3

Downloaded Unigine SuperPosition and tested it at 1080P Extreme settings, achieving a score of 11662. At lower resolutions, similar configurations appear in the results, though some scores reach around 18k. It seems both high and low scores are possible depending on the setup. The performance boost was noticeable—GPU usage peaked at 98% while global temperature stayed at 72%.

K
KEA_987
Junior Member
45
03-29-2022, 06:44 AM
#4
I initially thought V Sync was active in the driver, which would lead to a very low score. However, even with FPS capped at 60FPS on a 6900 XT, it should be around 9500—based on my tests with the Red Devil 6900 XT during Time Spy graphics checks. Not that I ran a full scientific test or disabled power limits, but since it never exceeded 250W, it probably won’t matter much. There must be another factor at play. First, I’d try DDU and revert to an earlier driver version. AMD has a track record of releasing drivers that cause issues, so your current one might be faulty in 3DMark evaluations. The version I’m using (22.9.2) has been decent for performance, but if you want improvement, consider switching to a different driver revision.
K
KEA_987
03-29-2022, 06:44 AM #4

I initially thought V Sync was active in the driver, which would lead to a very low score. However, even with FPS capped at 60FPS on a 6900 XT, it should be around 9500—based on my tests with the Red Devil 6900 XT during Time Spy graphics checks. Not that I ran a full scientific test or disabled power limits, but since it never exceeded 250W, it probably won’t matter much. There must be another factor at play. First, I’d try DDU and revert to an earlier driver version. AMD has a track record of releasing drivers that cause issues, so your current one might be faulty in 3DMark evaluations. The version I’m using (22.9.2) has been decent for performance, but if you want improvement, consider switching to a different driver revision.

J
JacobLouis30
Posting Freak
856
03-29-2022, 08:15 AM
#5
Avoid pushing performance beyond the 75°C rating. Aim for 100% fan speed at 60°C. As Ronothan mentioned, constraints from monitors and drivers should be considered; verify if you can run the monitor at 144MHz with no caps during testing, then use Freesync or Enhanced Sync. Regarding the PCIe setting, is your memory module rated for 4.0 or higher? If so, enter C-States (sleep) and Eco Functions. It seems acceptable given the current situation, though AMD cards may negatively impact performance.
J
JacobLouis30
03-29-2022, 08:15 AM #5

Avoid pushing performance beyond the 75°C rating. Aim for 100% fan speed at 60°C. As Ronothan mentioned, constraints from monitors and drivers should be considered; verify if you can run the monitor at 144MHz with no caps during testing, then use Freesync or Enhanced Sync. Regarding the PCIe setting, is your memory module rated for 4.0 or higher? If so, enter C-States (sleep) and Eco Functions. It seems acceptable given the current situation, though AMD cards may negatively impact performance.

S
Supercow160
Member
71
04-14-2022, 07:39 AM
#6
Reposition the card and power connector. Start GPU-Z in the background. After Timespy completes, capture a screenshot of the monitoring tab.
S
Supercow160
04-14-2022, 07:39 AM #6

Reposition the card and power connector. Start GPU-Z in the background. After Timespy completes, capture a screenshot of the monitoring tab.

L
166
04-21-2022, 07:52 PM
#7
This result is typical under normal conditions without heavy usage. Timespy isn't ideal for your device. Consider running a Firestrike test and share the findings (resetting all PC settings, including BIOS is also suggested).
L
LifelessShadow
04-21-2022, 07:52 PM #7

This result is typical under normal conditions without heavy usage. Timespy isn't ideal for your device. Consider running a Firestrike test and share the findings (resetting all PC settings, including BIOS is also suggested).

N
nicjohn2000
Member
149
04-25-2022, 03:52 AM
#8
Alright guys, I cracked it. Cyberat gave me the boost I needed, PCIe speeds... I had this really old WinTV capture card stuck in that computer. I never managed to get it working before, so I completely missed it was still there. Probably it was pushing me to use PCIe Gen 1. Eee Gads... After giving up last night, I pulled the card out this morning. Things are better now, and the results look as expected. I’m excited to try CyberPunk again—it wasn’t running well before, which is why I checked these benchmarks to find the issue. Thanks for sharing your thoughts. I know this isn’t the best benchmark, but it clearly showed the problem, which was exactly what I needed to confirm I’d fixed it.
N
nicjohn2000
04-25-2022, 03:52 AM #8

Alright guys, I cracked it. Cyberat gave me the boost I needed, PCIe speeds... I had this really old WinTV capture card stuck in that computer. I never managed to get it working before, so I completely missed it was still there. Probably it was pushing me to use PCIe Gen 1. Eee Gads... After giving up last night, I pulled the card out this morning. Things are better now, and the results look as expected. I’m excited to try CyberPunk again—it wasn’t running well before, which is why I checked these benchmarks to find the issue. Thanks for sharing your thoughts. I know this isn’t the best benchmark, but it clearly showed the problem, which was exactly what I needed to confirm I’d fixed it.