Overclock.net banner
Cancel
Create thread New Forums

[Official] AMD Radeon RX 6800 XT Owner's Club

Tags
6800 xt amd radeon rdna2
1 reading
Jump to Latest
1,561 - 1,580 of 1,702 Posts
#1,561 ·
Warmy said:
Are you suggesting the card is still thermal throttled?

Putting the PC outside or even near a window would be impractical for me, with the current weather at my place.
Click to expand...
What do you mean by "impractical"? You can't put your pc under cold air for ten minutes to run a benchmark?
 
Save
Reply Quote Rep+ Rep+
#1,562 ·
PJVol said:
What do you mean by "impractical"? You can't put your pc under cold air for ten minutes to run a benchmark?
Click to expand...
It's currently -20 here, and I haven't got any smaller windows to put the PC near. So yeah it's sadly impractical for me :(

With a delta T at 20c and a hotspot at about 90c, it doesn't seem like a thermal throttle to me, considering it didn't have any issues going to 108 c on hotspot prior to the install. What indicates that it's still thermals to you?

I used MPT to change the powerlimit and TDC close to their limits, and got a slightly higher score, but I was unable to complete any TS with an Fmax set above 2570. Delta T rose to 24 and hotspot got up to 100c, but I was able to do consecutive runs without any issues. See attached screenshot.

Naturally I think such a small performance increase at almost 100w increased power consumption isn't worth it, so I've changed it back to a PPT limit of 272W and changed the TDC limit to 275.
 

Attachments

#1,563 ·
Can you compare the results in TS at 272W with your current settings and with the MPT voltage tweaks I suggested before?
And btw, do memory clock calm down in idle if you switch refresh rate in windows to 120Hz ?
 
Save
Reply Quote Rep+ Rep+
#1,564 ·
PJVol said:
Can you compare the results in TS at 272W with your current settings and with the MPT voltage tweaks I suggested before?
And btw, do memory clock calm down in idle if you switch refresh rate in windows to 120Hz ?
Click to expand...
Sadly I saw a score decline, just did fresh reruns of TS. See attached Screenshots.

Fmax: 2550, 1050 mV, 2112 MHz VRAM, Fast timing, 272W PPT, 275A TDC: 20176 TS score.

Fmax: 2550, 1050 mV, 2112 MHz VRAM, Fast timing, 272W PPT, 275A TDC + your settings: 19747 TS score

Changing monitor refresh rate didn't lower the idle memory clock, see screenshot.
 

Attachments

#1,565 · (Edited)
Warmy said:
Sadly I saw a score decline, just did fresh reruns of TS. See attached Screenshots.

Fmax: 2550, 1050 mV, 2112 MHz VRAM, Fast timing, 272W PPT, 275A TDC: 20176 TS score.

Fmax: 2550, 1050 mV, 2112 MHz VRAM, Fast timing, 272W PPT, 275A TDC + your settings: 19747 TS score
Click to expand...
I see something's clearly wrong with these tests, because the results don't fit the GPU operating frequency.

With the average GPU clock @ 2363mhz you scored 20,176, while @ 2419 mhz - only 19,747. That could only be explained if your GPU just stalls in the second run for some reason, i.e. despite GFX Core is capable of running at a higher frequency, it still waiting for other components for the tasks to feed.

Could it be that some workload was running in the background during the second test, i.e. some services or 3rd party apps? Can you check the idle load in a task manager, including the GPU load? Is there any game launchers in autorun? Have you installed some 3rd party intivirus soft?

For example, just turning on logging to file in HWInfo64 during a test decreases the graphics score by about 200 pts.

PS: I checked on my PC and confirm that VRAM running at max basically has no impact on the TS score. Aside from that, I attached the screens of two TS runs to compare at the same settings, i.e. 1063mv 2650mhz 2112mhz - in Adrenalin, with the 2nd - voltage tuned in MPT
 

Attachments

Save
Reply Quote Rep+ Rep+
#1,566 ·
Forgot to add, these load discrepancy issues have to be sorted out to apply the final touch
 

Attachments

Save
Reply Quote Rep+ Rep+
#1,567 ·
PJVol said:
I see something's clearly wrong with these tests, because the results don't fit the GPU operating frequency.

With the average GPU clock @ 2363mhz you scored 20,176, while @ 2419 mhz - only 19,747. That could only be explained if your GPU just stalls in the second run for some reason, i.e. despite GFX Core is capable of running at a higher frequency, it still waiting for other components for the tasks to feed.

Could it be that some workload was running in the background during the second test, i.e. some services or 3rd party apps? Can you check the idle load in a task manager, including the GPU load? Is there any game launchers in autorun? Have you installed some 3rd party intivirus soft?

For example, just turning on logging to file in HWInfo64 during a test decreases the graphics score by about 200 pts.

PS: I checked on my PC and confirm that VRAM running at max basically has no impact on the TS score. Aside from that, I attached the screens of two TS runs to compare at the same settings, i.e. 1063mv 2650mhz 2112mhz - in Adrenalin, with the 2nd - voltage tuned in MPT
Click to expand...
I tried doing a rerun of TS with your settings, as in 1063mv and 2650mhz Fmax, and the MPT settings you've posted.

With this TS run i closed down any unnecessary process, as in only having HWinfo (without any logging), Taskmanager and TS open.

According to HWinfo and Taskmanager the GPU utilization stayed at 100% throughout the test (whenever the GPU actually had to render)

I'm using the latest drivers. 23.12.1

What am I missing?
 

Attachments

#1,568 · (Edited)
Warmy said:
What am I missing?
Click to expand...
1) It's hard to say without more data, at least the screens of task manager in idle (attached an example)
2) I'd also suggest to uninstall 23.12.1 driver with the AMD Cleanup Utility (from my personal experience, the prev. version is more stable) which can be found in a driver package installation folder or from this link:
then install the minimal setup of the 23.11.1 driver. Save the MPT profile before clean up - the driver installation procedure removes SPPT from the registry - so you can used it with the newly installed 23.11 driver.
And now check, if the minimal setup of 23.11.1 improve the performance in TS
 

Attachments

Save
Reply Quote Rep+ Rep+
#1,569 · (Edited)
PJVol said:
1) It's hard to say without more data, at least the screens of task manager in idle (attached an example)
2) I'd also suggest to uninstall 23.12.1 driver with the AMD Cleanup Utility (from my personal experience, the prev. version is more stable) which can be found in a driver package installation folder or from this link:
then install the minimal setup of the 23.11.1 driver. Save the MPT profile before clean up - the driver installation procedure removes SPPT from the registry - so you can used it with the newly installed 23.11 driver.
And now check, if the minimal setup of 23.11.1 improve the performance in TS
Click to expand...
I did as you suggested and used the AMD cleanup utility, reinstalled version 23.11.1.

Had to install the full version to use a manual profile to set the clock speeds etc. MPT can probably do this as well, but I'm not confident enough to use that as of yet.

I have attached screenshots of the TS run and the idle task manager.
 

Attachments

#1,570 · (Edited)
Warmy said:
I have attached screenshots of the TS run and the idle task manager.
Click to expand...
I've asked for the TM details sorted by CPU load in descending order - you need to click on table header - exactly as on the screen I posted. Or post the screens of "processes" tab with the corresponding column clicked (see screens).
Do you have any 3rd party soft installes, like RGB things or iCue?

Anyway, at this point the only thing that comes to mind is to find another PC (friends, colleagues, parents, children, PC store... :giggle:) where you can test your GPU to rule out the motherboard as the possible reason. Judging by what I see from the Red Dragon 6800xt owners in this thread, something is wrong with yours.
In all honesty, AB350M is the worst choice for the PC motherboard with the 5800X3D and 6800XT. At some point I had to get rid of my previous X370 Taichi board due to problems with 5000 series support.
 

Attachments

Save
Reply Quote Rep+ Rep+
#1,571 ·
PJVol said:
I've asked for the TM details sorted by CPU load in descending order - you need to click on table header - exactly as on the screen I posted. Or post the screens of "processes" tab with the corresponding column clicked (see screens).
Do you have any 3rd party soft installes, like RGB things or iCue?

Anyway, at this point the only thing that comes to mind is to find another PC (friends, colleagues, parents, children, PC store... :giggle:) where you can test your GPU to rule out the motherboard as the possible reason. Judging by what I see from the Red Dragon 6800xt owners in this thread, something is wrong with yours.
In all honesty, AB350M is the worst choice for the PC motherboard with the 5800X3D and 6800XT. At some point I had to get rid of my previous X370 Taichi board due to problems with 5000 series support.
Click to expand...
No RGB software of any kind, and I doubt it would have a greater impact.

The motherboard is the leftovers from a budget build i did back in 2017 (1500x). When i built the current system earlier this year, I kinda liked the fact that I could hang on to the motherboard and get a completely new PC in terms of performance, and that still holds true IMO.

The GPU might not be able to perform as well as others, but I believe it's still performing well within reason, as I've seen other 6800 XT's perform even worse.

But it obviously bothers me that a successful repaste (from a temperature standpoint) hasn't improved my benchmark scores even slightly.

I'll wait for someone else to maybe chime, and maybe find more sources elsewhere to improve things or locate my issue.

Regardless, thanks for your time and effort!
 
#1,572 · (Edited)
Warmy said:
No RGB software of any kind, and I doubt it would have a greater impact.

The motherboard is the leftovers from a budget build i did back in 2017 (1500x). When i built the current system earlier this year, I kinda liked the fact that I could hang on to the motherboard and get a completely new PC in terms of performance, and that still holds true IMO.

The GPU might not be able to perform as well as others, but I believe it's still performing well within reason, as I've seen other 6800 XT's perform even worse.

But it obviously bothers me that a successful repaste (from a temperature standpoint) hasn't improved my benchmark scores even slightly.

I'll wait for someone else to maybe chime, and maybe find more sources elsewhere to improve things or locate my issue.

Regardless, thanks for your time and effort!
Click to expand...
You might want to go legit (registered) ,stand alone version and run as administrator.
I was having inconsistent results also ..had to dig out my old 3DMark key :D
 
#1,573 ·
coolhandluke41 said:
You might want to go legit (registered) ,stand alone version and run as administrator.
I was having inconsistent results also ..had to dig out my old 3DMark key :D
Click to expand...
I've got a legit copy of 3DMark bought through the Epic games launcher, because it was cheapest at the time.

It doesn't seem to matter which platform you buy it on, as it just launches the 3DMark stand alone client.
 
#1,574 · (Edited)
Hope this is ok , bought this card for casual gaming
 

Attachments

#1,575 · (Edited)
coolhandluke41 said:
Hope this is ok , bought this card for casual gaming
Click to expand...
I'd be more concerned about your cpu being able to feed it with tasks.

ed: haven't run FireStrike for a while. Here is mine in Extreme at stock PL just in case
 

Attachments

  • Rep+
Reactions: coolhandluke41
Save
Reply Quote Rep+ Rep+
#1,578 ·
Crazyonion said:
hey im new to this place just wondering if my setup all on air is getting okay results i mostly game at 1440p
Click to expand...
Hi, I believe they are okay as long as you're comfortable with the fan noise and have an overall positive gaming experience.
Everything else, including benchmarks, is for perfectionists/enthusiasts.
 
Save
Reply Quote Rep+ Rep+
#1,579 ·
Hi guys, i`ve got the strix 6900XT and buy me the 6800XT. I love PC's and build them and OC, Gaming and so on ...

I worked with MPT (More Power Tool) and More Clock Tool.

Why i bought the 6800XT? Because i know; its the best choice for money (415€). Now it cost 399€ Speedster on alza.de

This new card have no heat problems and easy OC to 2600MHz. And they are efficient!

The Thing is: You can make your own voltage curves AND active FT2 (Fast Timing 2 with MPT and MCT)
This is crazy ****tt ;-)

On Gamers Nexus, we can see the 6800XT on stock is better, as the 7800XT. :cool:
AMD Radeon RX 7900 GRE GPU Review & Benchmarks vs. RX 7900 XT, 7800 XT, RTX 4070 Super (youtube.com)
 

Attachments

1,561 - 1,580 of 1,702 Posts
About this Discussion
1.7K Replies
218 Participants
Last post:
Travis Scott
Overclock.net
posts
28M
members
562K
Since
2004
The premier forum for overclocking experts and enthusiasts. Discuss hardware optimization, custom builds, benchmarking, cooling solutions, and pushing the boundaries of computing performance. From beginner guides to extreme overclocking, join our technical community to master system tuning.
Full Forum Listing
Explore Our Forums
Intel CPUs Hardware News NVIDIA PC Gaming Water Cooling