6990 Mining Blues

[jamal]

Hey Everybody =)

So pretty new to mining, just picked up a 6990 so mining with that in my desktop as a secondary GPU while I have a 7770 driving my display.
However I am having some trouble as the GPU cores keep dropping to 0% load and then back to 100 then back to zero.

I decided to put the bios switch on the card to switch 1, and the problem has disappeared on one gpu core.
On the core in which the problem still occurs, the voltage has increased to 1.175 but drops to 1.0 frequently also dropping the gpu usage to 0% and reducing the clock speed. On the other core which is now working the voltage 1.12. Any ideas what the problem might be? Is there a second bios switch I am missing

I have attached two screen shots for reference, notice how the GPU with 1.12 voltage shows the cores continuously staying at 100%, while the other shows it as 1.175 but is fluctuating in voltage and memory clock while the usage fluctuates from 0-100% as per the spikey chart by GPU usage.


http://s11.postimage.org/oz3enyd0f/pic_1.jpg

http://s11.postimage.org/fsl40o7s3/pic_2.gif
anonymous image hosting

[thepwnorbpwnd]

Dear god! Drop the clocks! Your VRM's are within 8 degrees of frying!! 

[jamal]

Woah I didn't even notice that (hence why I am a newbie LOL!). Downclocked back to 830. Guessing a lack of airflow with the two cards increases heat so much that the core keeps dropping due to high heat?

[kwoody]

Typically temps that get into 90C+ range will cause the card to throttle; it'll continue mining but hashrate will drop to about half every few seconds. Since your 6990 isn't being used to drive any displays, you should lower the memory clocks of those cores. Like 400MHz memory clocks at the most, as low as is stable pretty much. This will reduce heat/electricity use. I use 5830's mainly, 950MHz GPU clock and 325MHz memory clock. Also, fan is only at 30% in that screenshot? Run your fan at like 75%. It'll be loud, but it'll save your card from fiery death.

[thepwnorbpwnd]

^^ +1 Was just about to suggest that you also ramp your fan to ~60-80%

[jamal]

Thanks for all the info! I did try to downclock the memory, however when I do it shows in GPUz to still be running at the stock settings (ie I am unable to underclock the memory). Also I am using the twin turbo accellerro cooler and even though I have the fan set to 100 percent it shows it as 30 from gpuz for one core but 100 for the other core. Guessing there is some problemm with reading the fan settings? Because I'm quite sure at 30 percent the temperature would be much higher! Any suggestions on why I am unable to lower my memory clock?


Oh also I should mention in gpuz it shows that the mem clock has been reduced but on the real-time stats page it shows its still running at default speeds. Thanks!

[jamal]

Thanks for the suggestion still no luck thugh. Wont even downclock 1 mhz. Tried to put it at the same speed as gpu, then lower, then higher, and only higher worked. Even going down 1mhz would make it run at stock. Will update with pics to show what I mean. Perhaps there is some patch download or something?

[c4n10]

Thanks for the suggestion still no luck thugh. Wont even downclock 1 mhz. Tried to put it at the same speed as gpu, then lower, then higher, and only higher worked. Even going down 1mhz would make it run at stock. Will update with pics to show what I mean. Perhaps there is some patch download or something?

Also, click on the drop-downs next to each reading and make sure "Show Current Reading" (or something similar, dunno, not looking at the program right now) as opposed to "Show Highest, Show Lowest or Show Average".

[jamal]

Yup it is set to "Show Current Reading"  I can't even set the memory clocks 1 mhz lower then stock. Tried to set it same as gpu too. Nothing doing. Can only overclock it not underclock

http://s9.postimage.org/58h26u0gr/Core_1_gc.gif

http://s9.postimage.org/tq95uq317/core_1.gif

http://s9.postimage.org/7flaur5qz/core_2_gc.gif

http://s9.postimage.org/u5p11hecb/core_2.gif

[c4n10]

Hmmm, strange... I wish I had more knowledge about that specific graphics card...

[Kaliecious]

what brand vid card you running?

[jamal]

Yea I found it pretty weird too, and its an XFX 6990. And thanks for all the support and suggestion guys! Much appreciated

[c4n10]

Check this out: http://hardforum.com/showthread.php?t=1647728

Your answer just might be in there...

[Kaliecious]

are you using msi afterburner?
what driver set you have?

and as to  the heat, I called msi, this cards have been rated up to 99c.. ive tried undervolting but it acts up. my hotest gpu is about 96-98c im over clocked to 915 ram at 625 and with a fan mine mine 24/7 no problems.

[c4n10]

Also, what is the wattage on your PSU...? It may be possible that the card is drawing more power than your PSU has to give...

[c4n10]

And last try messing with power control settings in CCC. Try between the range of 0 to +20%...

I am about 90% sure this will solve your problem.

[Shadow383]

are you using msi afterburner?
what driver set you have?

and as to  the heat, I called msi, this cards have been rated up to 99c.. ive tried undervolting but it acts up. my hotest gpu is about 96-98c im over clocked to 915 ram at 625 and with a fan mine mine 24/7 no problems.

...Call us in a month and let us know if they're broken 

[jamal]

Hey so my PSU is 850 watt OCZ Z series 80%+Gold Cert

I am using the lastest drivers, MSI, etc. Played around with the power slides still nothing =/. Weird =/. Oh well, running it at stock (830 mhz/1250) ... each cores pulling about 370 Mhash/s...temperatures after 4 hours at 96-100% load are ~70 Celsius at cores and vram is at 106.

I'm going to try other things when I have more time, if anybody else has any other solutions I am all ears. If I do get it to work I will update

Thanks to everyone for their help and support! You guys rock =)

[c4n10]

Hey so my PSU is 850 watt OCZ Z series 80%+Gold Cert

I am using the lastest drivers, MSI, etc. Played around with the power slides still nothing =/. Weird =/. Oh well, running it at stock (830 mhz/1250) ... each cores pulling about 370 Mhash/s...temperatures after 4 hours at 96-100% load are ~70 Celsius at cores and vram is at 106.

I'm going to try other things when I have more time, if anybody else has any other solutions I am all ears. If I do get it to work I will update

Thanks to everyone for their help and support! You guys rock =)

Hmm, well the minimum required for your card is 650w, recommended is 1200w for optimal performance...

Do you have additional video cards or any other after-market hardware that might be using a lot of power...?

Also, make sure you have ULPS disabled...

[c4n10]

Another thought, if you are using software to overclock, make sure you turn off all your turbo and throttling settings in the BIOS as you may be over or underclocking your FSB through the BIOS which will affect the gpu.