Overclock.net banner
3,521 - 3,540 of 9,042 Posts
  • Love
Reactions: Butanding1987
Hello!!! So repasted my 7900xtx a 2nd time after failed first repaste with cooler master cyrofuze. For 2nd repaste i used noctua nh-nt1 and getting much better temps but wanted to ask in here if its good or should be better. My room temp is about 70f-72f. Playing diablo 4 4k ultra setting after 3 to 4hrs my temps are 50c-56c cur temp/ 77c-83c junction temp. With the overclock i posted below would this be considered decent/good temps? Really appreciate the help!! I also included pic of pc with fan setup to make sure its right its my first full pc build. Side and bottom are intake and top aio and rear are exhaust at least thats what i was going for lol
Purple Red Magenta Visual effect lighting Entertainment
Font Screenshot Rectangle Technology Software

Output device Computer Font Gadget Audio equipment
 
Turn power limit back to zero and junction temp will drop down to 75 to 80. Mine did. Overclock 3350 memory 2750 and undervolt 1125. Junction temps went into mid 90s with power limit on 15 now max out at 80 with power limit back to zero.
So if I'm reading this correctly lower powerlimit to 0. Then set 3350, 2750 and undervolt to 1125 and you don't go over 80c junction? Also I thought you wanted a lower undervolt? Isn't default 1100?
 
Hello!!! So repasted my 7900xtx a 2nd time after failed first repaste with cooler master cyrofuze. For 2nd repaste i used noctua nh-nt1 and getting much better temps but wanted to ask in here if its good or should be better. My room temp is about 70f-72f. Playing diablo 4 4k ultra setting after 3 to 4hrs my temps are 50c-56c cur temp/ 77c-83c junction temp. With the overclock i posted below would this be considered decent/good temps? Really appreciate the help!! I also included pic of pc with fan setup to make sure its right its my first full pc build. Side and bottom are intake and top aio and rear are exhaust at least thats what i was going for lol View attachment 2619159 View attachment 2619158
View attachment 2619157
Hmm so your having paste issues also? I guess it's a common thing even across brands.

Your temps seem to be close to mine after a repaste. Mine are slightly higher but I used a heavily modded Skyrim @4k for testing. I suspect I am putting more load on mine slightly. Either way, I am curious to know if your paste fails again as I am testing mine in this manner. I plan on getting a water block soon but am curious if this happens on a water block as well.
 
So if I'm reading this correctly lower powerlimit to 0. Then set 3350, 2750 and undervolt to 1125 and you don't go over 80c junction? Also I thought you wanted a lower undervolt? Isn't default 1100?
I was just saying to lower the powerlimit back to zero to get the junction temps to drop back down. I was just saying the overclock and very small undervolt I could still get with the power limit back at zero, not suggesting to copy them.
 
So this is my first time getting a pretty high end card after just upgrading from a 2070s and I've been looking online for overclocking advice for the 7900xt and I seem to be getting numbers that seem to be too good to be true. However after doing 3dmark benchmarks after I've been gaming on it stably I realized how well my card ranks vs other 7900xt's. I have it paired with a 13600k and was using the speed way benchmark on 3dmark to test my card since it's quick so I can change things to see scores quicker and ended up with this score.
Rectangle Violet Font Magenta Art

Rectangle Font Magenta Screenshot Circle


These are some amd adrenalin settings and stress test results.


Colorfulness Font Technology Rectangle Screenshot

Colorfulness Rectangle Slope Font Parallel


So I'm still kind of shocked with how lucky I seemed to have gotten cause I've had very bad luck with my gpus and cpus in the past and this is super abnormal for me. Any feedback or comments would be greatly appreciated.
 
Anyone experiencing crashes with 500W bios and 15%PL? My Hotspot does not even exceed 90C, but still crashes in multiple games
No problem on two cards. check your system
 
  • Rep+
Reactions: FatalFighter
No problem on two cards. check your system
Thanks for reply.
I guess I forgot I UV'd as well during testing.
Checked CP2077 with 1150mv on GPU runs fine now.
The game which crashes no matter what is Metro Exodus, I guess not liking AMD's drivers.

It said 2-3 days and now this...

Rectangle Font Parallel Fashion accessory Transparency
 
Thanks for reply.
I guess I forgot I UV'd as well during testing.
Checked CP2077 with 1150mv on GPU runs fine now.
The game which crashes no matter what is Metro Exodus, I guess not liking AMD's drivers.

It said 2-3 days and now this...

View attachment 2619236
Yeah UV is the cause of most crashes. If you're still crashing at 1150mv then probably thermals at 550W are to blame given you're still on air.

What are the temps on your voltage regulators?

Install hwinfo64 and log during Metro Exodus to see what the temps are like across the card when it crashes.
 
So I'm still kind of shocked with how lucky I seemed to have gotten cause I've had very bad luck with my gpus and cpus in the past and this is super abnormal for me. Any feedback or comments would be greatly appreciated.
I'd run OCCT vram test to see if memory is actually stable. My 7900 XT didn't give much hoot about running 2750 until OCCT showed all of the errors generated within like 10 seconds.
 
So this is my first time getting a pretty high end card after just upgrading from a 2070s and I've been looking online for overclocking advice for the 7900xt and I seem to be getting numbers that seem to be too good to be true.
I'd run OCCT vram test to see if memory is actually stable. My 7900 XT didn't give much hoot about running 2750 until OCCT showed all of the errors generated within like 10 seconds.
Another good VRAM stability test is GitHub - GpuZelenograd/memtest_vulkan: Vulkan compute tool for testing video memory stability
 
  • Rep+
Reactions: RedF
I'd run OCCT vram test to see if memory is actually stable. My 7900 XT didn't give much hoot about running 2750 until OCCT showed all of the errors generated within like 10 seconds.
I ran OCCT with my vram at 2700 at first then at 2800 and no errors were produced. I've crashed a test once on 2800 (3dmark speedway stresstest) and haven't crashed at all on 2700. Also I've been able to undervolt to 950 with an occasional crash so I've just stuck to 975 as it's been very stable for me so far. I could probably fine tune it to find exactly what my cards limits are but I'd rather not deal with the constant crashing of finding that out and I'm perfectly happy with what seem to be these ridiculously great results as I've been able to game on them for the last day with almost no crashing. The only game that crashed much was star citizen and I'm pretty sure there's not a computer in the world that runs that game well.
 
Yeah UV is the cause of most crashes. If you're still crashing at 1150mv then probably thermals at 550W are to blame given you're still on air.

What are the temps on your voltage regulators?

Install hwinfo64 and log during Metro Exodus to see what the temps are like across the card when it crashes.
I run HWInfo in the background all the time practically
I tried playing Metro rn and it crashed almost instantly. Temps are not perfect but I believe should not cause crashes?
Font Screenshot Menu Electric blue Darkness
 
Anyone who re-pasted the 7900xtx/xt had to do it again after a while?
I re-repasted PowerColor Radeon RX 7900 XTX Red Devil Limited Edition and after three weeks it started hiting 110C hotspot again.
A friend with PowerColor Radeon RX 7900 XT Red Devil has the same issue.
I recived two days ago Asrock rx 7900 XTX Aqua. In first day I tested it in 3DMark Time Spy with Default bios and default settings and it hits 91C hotspot . With Extreme bios and default settings in Time Spy in second GPU test it hits 110C hotspot.
The next day I tested it again in Time Spy with Default bios and default settings, the hotspot reached 101C. With Extreme bios and default settings it almost instantly hits 110C hotspot.
The temperature of hotspot it's degrading 10C in one day.
Today I tested with Default bios and default settings and PL@-10, the hotspot temperature reached 90C.
 
I run HWInfo in the background all the time practically
I tried playing Metro rn and it crashed almost instantly. Temps are not perfect but I believe should not cause crashes?
View attachment 2619296
You've got core clock spikes over 3.6Ghz, but perhaps that just after it crashed?

What have you got your clockspeed set to? Try something like 500Mhz min and 2800, 2900 or 3000Mhz max and see if it will run.
 
3,521 - 3,540 of 9,042 Posts