Cooking amd 6800s, must watch.

GregTombstone

Limp Gawd
Joined
Mar 6, 2004
Messages
158
My 6800xt was hitting 2500 stock out of the box at Christmas. Cut the voltage and kept the revs to 2300 as it was hitting 90c plus. Seems recent driver is not good.


 
My 6800xt was hitting 2500 stock out of the box at Christmas. Cut the voltage and kept the revs to 2300 as it was hitting 90c plus. Seems recent driver is not good.



that 90+ is junction temp or core temp?
did you experience any clock throttle?
 
People on social media and forums - AMD fanboys, maybe - I dunno who they are - keep saying it's not the drivers. According to them, it's some peeps in Germany who sent a bad batch of cards to some Germany guy KrisFix it and De8auer tested some cards too (also, in Germany) - so the problem is only isolated to Germany? :)
I don't think Jay is in Germany, though - and he's just testing his own card (and latest driver)?
 
People on social media and forums - AMD fanboys, maybe - I dunno who they are - keep saying it's not the drivers. According to them, it's some peeps in Germany who sent a bad batch of cards to some Germany guy KrisFix it and De8auer tested some cards too (also, in Germany) - so the problem is only isolated to Germany? :)
I don't think Jay is in Germany, though - and he's just testing his own card (and latest driver)?
LoL what?
I also had rx 6500 xt, it reached 95c hot spot on the default, but never experienced any throttle.
you can google and find many new gen nvidia cards have junction temp read on HWInfo and it's normal to have it aroung 80~95/100c.
MY zotac gtx 1070ti has junction temp max 94c no matter what profile fan I applied to.
 
LoL what?
I also had rx 6500 xt, it reached 95c hot spot on the default, but never experienced any throttle.
you can google and find many new gen nvidia cards have junction temp read on HWInfo and it's normal to have it aroung 80~95/100c.
MY zotac gtx 1070ti has junction temp max 94c no matter what profile fan I applied to.
Did a lot of reviewers test your cards?
 
Nvidia shills and nvidiots are all over youtube - making up stories about bad coolers. The ppl who comment below the video are paid by Nvidia as well. I like to use Linux and prefer to use an AMD card but I am paid by Nvidia so I don't. /s
 
Yeah, the cards problem in Germany - yes, miners card - reseller abused the cards - but, the 7900 XTX problem is different. The driver issue was speculation - nobody declared it was drivers.
 
To think anyone would believe a driver would do this to a card. Boggles my mind. But nvidiots are the first to raise hell.
 
I have a powercolor 6900xt that might be going bad. Lately its been crashing everyday, i checked event logs, tried different drivers, ddu, etc swapped in a 6600xt hasn't crashed since. Id like to try it in a different rig, but its to big to fit in my other builds. It can run NTLM cracking in hashcat without crashing, but it crashes very frequently in Apex.
 
I have a powercolor 6900xt that might be going bad. Lately its been crashing everyday, i checked event logs, tried different drivers, ddu, etc swapped in a 6600xt hasn't crashed since. Id like to try it in a different rig, but its to big to fit in my other builds. It can run NTLM cracking in hashcat without crashing, but it crashes very frequently in Apex.
You may be able to play around with the clock speed and get it to run stable. Maybe when it's under compute work loads the clock speed is lower which allows it to not crash? I had a Vega 56 that would crash in games(but generally not in compute workloads) but if I lowered the clock speed by 100MHz in Afterburner it never crashed. The games still performed well.
 
  • Like
Reactions: pavel
like this
You may be able to play around with the clock speed and get it to run stable. Maybe when it's under compute work loads the clock speed is lower which allows it to not crash? I had a Vega 56 that would crash in games(but generally not in compute workloads) but if I lowered the clock speed by 100MHz in Afterburner it never crashed. The games still performed well.
Thats a good idea, but you know whats crazy. Going from a 6900xt to a 6600xt had basically zero impact on my FPS at 1440p@144hz. I havent been super motivated to figure it out.
 
Last edited:
You may be able to play around with the clock speed and get it to run stable. Maybe when it's under compute work loads the clock speed is lower which allows it to not crash? I had a Vega 56 that would crash in games(but generally not in compute workloads) but if I lowered the clock speed by 100MHz in Afterburner it never crashed. The games still performed well.
What about undervolting, too? Would that be a good test for him? Also, he didn't mention what the 6600 xt is or whether the 6900 XT was new when he got it.... if it was previously, owned, someone might have messed with the power limits or mined it, who knows? If it's original, it's odd that it would suddenly have problems?

Dunno if these links would help in any way....but, maybe something will give you ideas?:

https://linustechtips.com/topic/1399479-rx-6900-xt-crashes-always/

https://community.amd.com/t5/graphics/6900-xt-mba-crashing-frequency/td-p/505146
 
I have a powercolor 6900xt that might be going bad. Lately its been crashing everyday, i checked event logs, tried different drivers, ddu, etc swapped in a 6600xt hasn't crashed since. Id like to try it in a different rig, but its to big to fit in my other builds. It can run NTLM cracking in hashcat without crashing, but it crashes very frequently in Apex.
Mine is an Asus 6900XT LC and it's doing the same thing. No mining on mine, just gaming, and it's been crap since 22.5.2 or so. DDU and rolling back is just annoying. Seems any time I approach the power target it just hard reboots the system. Doesn't matter if I set it low (289w) or high (350+) via MPT.
 
Back
Top