Topic: CGMiner on XFX 7970 (925Mhz): heat problems (Read 1341 times)
how old is your card?
It was a driver problem. Wanted to try 13.4 again, didn't install anymore. After multiple hours of trying I reinstalled Win7 with 13.4 and it is now working very well with the latest version of cgminer. The heat is still there but at least now cgminer is underclocking it when it gets to hot.
Heya,
I need some help tuning my mining setup... (BTC using CGMINER)
My computer is an i5 with 4Gb of RAM running Windows 7 Home Edition, 2 XFX Radeon HD 7970 3Gb cards and a Gigabyte GA-P55-UD4 Mobo. Cards are connected with Crossfire cable but Crossfire is disabled. I've tested with AMD Overdrive ON and OFF.
The problems I have are all temperature related. I need to underclock to really get it working but even then, it is not working very stable. My case is a midtower with a 12mm fan blowing air in the front, a 12mm fan blowing air out in the back, two 12mm fans on the side (gpu height) blowing air in and my PSU's (Antec 900Watt) fan faces the bottom GPU. My Cgminer settings:
gpu-engine -> 625-925
gpu-memdiff -> -150
intensity -> d (mostly stays on 6)
auto-gpu
auto-fan
target-temp -> 80 (is this safe to run at for longer periods of time?)
overheat-temp -> 85
cutoff-temp -> 90
The first card (gpu0) runs at 925Mhz/775Mhz, fan @ 85% and doing around 550Mh/s (I can live with this)
The second card however has more heat problems and underclocks doing around 400Mh/s (sometimes more, sometimes less)
Now, if this were running stable I could still live with it but CGMiner sometimes has trouble underclocking my second GPU. It says it is underclocking it, but nothing happens. The GPU stays at 925Mhz while memory clock stays at 1375Mhz (should be 775Mhz with the memdiff option). Ive tried to set these values in AMD Overdrive also but no joy.
So if this doesn't work the GPU is disabled after a while (REST)... It then looks like it is starting again but it doesn't, it is doing 0 Mh/s and it stays that way. At least it looks like it, maybe I have not waited long enough before manually restarting the GPU, i'll leave it for a few hours next time.
Does anybody have any idea what can cause cgminer to be unable to set the clockspeeds? Bad driver?
I've tested 13.4 and now I am running 12.8 because this seems like the most supported for cgminer?
With 12.8 I am running AMD ADL SDK 2.6. After a reboot it seems to work for a moment as I can see the gpu-engine is set to 925 and clock to 775 but after a while it looks like something is overriding this because the memclock is back to 1375Mhz and even when I try to set this manually, it won't go down. (verified in multiple tools)
Haven't tried other drivers than those because the first card works well with it...
I'm looking into better cooling options so I can overclock a bit (have reached +700Mh/s for a few minutes then crash due to heat) but until I have that I would like to know how to get this current setup stable.
So what can cause cgminer to be unable to set clockspeeds, or what could override it?
Any suggestions on heat management or experiences with the same cards are also welcome. (i.e. what are safe long term values?)
Thanks,
W
I need some help tuning my mining setup... (BTC using CGMINER)
My computer is an i5 with 4Gb of RAM running Windows 7 Home Edition, 2 XFX Radeon HD 7970 3Gb cards and a Gigabyte GA-P55-UD4 Mobo. Cards are connected with Crossfire cable but Crossfire is disabled. I've tested with AMD Overdrive ON and OFF.
The problems I have are all temperature related. I need to underclock to really get it working but even then, it is not working very stable. My case is a midtower with a 12mm fan blowing air in the front, a 12mm fan blowing air out in the back, two 12mm fans on the side (gpu height) blowing air in and my PSU's (Antec 900Watt) fan faces the bottom GPU. My Cgminer settings:
gpu-engine -> 625-925
gpu-memdiff -> -150
intensity -> d (mostly stays on 6)
auto-gpu
auto-fan
target-temp -> 80 (is this safe to run at for longer periods of time?)
overheat-temp -> 85
cutoff-temp -> 90
The first card (gpu0) runs at 925Mhz/775Mhz, fan @ 85% and doing around 550Mh/s (I can live with this)
The second card however has more heat problems and underclocks doing around 400Mh/s (sometimes more, sometimes less)
Now, if this were running stable I could still live with it but CGMiner sometimes has trouble underclocking my second GPU. It says it is underclocking it, but nothing happens. The GPU stays at 925Mhz while memory clock stays at 1375Mhz (should be 775Mhz with the memdiff option). Ive tried to set these values in AMD Overdrive also but no joy.
So if this doesn't work the GPU is disabled after a while (REST)... It then looks like it is starting again but it doesn't, it is doing 0 Mh/s and it stays that way. At least it looks like it, maybe I have not waited long enough before manually restarting the GPU, i'll leave it for a few hours next time.
Does anybody have any idea what can cause cgminer to be unable to set the clockspeeds? Bad driver?
I've tested 13.4 and now I am running 12.8 because this seems like the most supported for cgminer?
With 12.8 I am running AMD ADL SDK 2.6. After a reboot it seems to work for a moment as I can see the gpu-engine is set to 925 and clock to 775 but after a while it looks like something is overriding this because the memclock is back to 1375Mhz and even when I try to set this manually, it won't go down. (verified in multiple tools)
Haven't tried other drivers than those because the first card works well with it...
I'm looking into better cooling options so I can overclock a bit (have reached +700Mh/s for a few minutes then crash due to heat) but until I have that I would like to know how to get this current setup stable.
So what can cause cgminer to be unable to set clockspeeds, or what could override it?
Any suggestions on heat management or experiences with the same cards are also welcome. (i.e. what are safe long term values?)
Thanks,
W
Jump to: