Bitcoin Forum

I'm running

Ubuntu 15.10
AMD drivers radeon-crimson-15.12-15.302-151217a-297685e
Genoil's ethminer

but I only get 10 MH/s because the GPU clocks only at core 501 MHz (normal 1000) and memory 1400 MHz (normal 1400) and sensors say 95 C / fan 100% (3500 rpm).

I've read there was a bug in 2012 if you run also a youtube video but everything is off except ethminer.

If I change the settings like this


the gpu core clocks still at 501 (not changed) and the memory now at 1450 (changed).

If I change the settings like this


the gpu core clocks now at 400 (changed) and the memory now at 1250 (changed).

Thanks for every hint!

P.S. Are there any active links to download a Stilt BIOS for the 7970 and 7950?

The temperature is too high. You may also increase  the power limit to 20%.

How can I get the gpu cooler? PC case and windows are open. Outside 0 C. My 7950 runs at 65 C and 70% fans.


How can I do that with aticonfig? Will the temps go even higher?

Fans are at 100% at 501 MHz only :(

Clocked down to 400 MHz but not really better and only 10 MH/s:

atitweak -s
0. AMD Radeon R9 200 / HD 7900 Series ()
    engine clock 400MHz, memory clock 1250MHz, core voltage 0.95VDC, performance level 1, utilization 100%
    fan speed 100% (3480 RPM) (user-defined)
    temperature 92 C
    powertune 0%

Your GPU is clocking its own speed down trying to keep itself cool. Grab HWMonitor and confirm the temperatures of that and the rest of your computers and work out a cooling solution

I'm mining with Ubuntu and atitweak and aticonfig tell the same.

I've flashed the gpu bios from version 015.012.000.004.000000 (113-AD44400-101) to 015.039.000.000.000000 (113-AD47700-100) but now the gpu doesn't boot with this bios version. Maybe the 280X bios works.

Does someone have a working link to Stilt bios versions?

You should get like 19-20MH/s with those.

If the temperatures spike as soon as you start mining then its either bad cooling or blown VRM. If you never took it apart before to change the thermal paste then most likely you have a blown VRM.

The GPU mines at 10 MH/s @500MHz (95 C) and @400MHz (85-90 C) and if I start mining the temps go up within seconds from 40 to 85/90 C. Any idea how to unblow the VRM ;D Damn!

Try changing your thermal paste, or maybe its not mounted properly but if it looks ok then its most likely a blown VRM. Looking for a burnt MOSFET or capacitor on the board.

Did you buy this 7970 new? If you bought it used, there is a 50% chance it was mined to death before.

I've bought it used on Ebay but the GPU looks pretty unused/new. The seller says the card worked fine but with Windows 10 I get about 1 MH/s at 300 MHz and with Ubuntu I get about 10 MH/s at 501/1400 MHz and 95 C or 9 MH/s at 400/1250 and 90 C. The GPU ignores completely every input from MSI Afterburner and with Ubuntu everything above 501 MHz. Changing the mem clocks with Ubuntu works fine.

If I start Furmark I get about 11 FPS and GPU-Z says the core clock stays at 300 MHz. Tomorrow I will install and try the CoD 4 demo.

Try flashing the bios, maybe he messed with it.

Go ASIC or go home!

Trying to raise your post count I see.

it must be related to the temp, the gpu is limiting itself for that or another reason check with gpu-z to see who is limiting the gpu, it should do 25mega on etheruem

I've already ordered 2 Zeus Ether ASICs by KnC and they promised to ship between Q2/Q3 2017 ;D


I had the same idea but both BIOS versions are the default one's (015.012). I've tried BIOS version 015.039 (ASUS 7970) and 015.041 (ASUS 280X) but with both versions the PC doesn't boot.


How? I'm using Windows/GPU-Z for the first time since many years. Which values are important?

For example if I start Furmark the core clock stays at 300 MHz and can't be changed with MSI Afterburner.

Did you reflash the stock bios from vbios database?

The bios can be the same version but he might of messed with some values, just reflash one from the techpowerup vbios website.

I mean I see where you're coming from but aren't you doing exactly the same thing  ;D ;D ;D

With both default BIOS versions (small lever to the right and to the left) the PC is booting fine but after flashing with techpowerup.com BIOS version 015.039 (ASUS 7970) and 015.041 (ASUS 280X) the PC didn't boot any longer.

http://www.techpowerup.com/vgabios/153444/153444.html

http://www.techpowerup.com/vgabios/157697/asus-r9280x-3072-131119-2.html

Flash that version that makes the PC boot. Or try Ubuntu, usually linux isn't picky about modified BIOSs and lets it boot.

The PC boots only with the default BIOS and I'm mining only with Ubuntu. I've downloaded Windows 10, GPU-Z and Furmark only because of this faulty GPU and at 100% load because of Furmark the engine clocks @501 and mem @1400. The GPU ignores every command through MSI Afterburner and aticonfig in Ubuntu above 501 MHz. I'll return the 7970 as soon as possible. In the meantime I'm mining @400 MHz with 10 MHs 8)

Maybe the fans are shot. What is the RPM of the fans? Not the % but the RPM?

3495 RPM = 100%

VDC = 0.95

fans are spinning fine, but your voltage is too low for regular mining clocks.

I use Vdc  0.95V for my 7990. But the clock for the mining is just 830 MHz. Otherwise, it will freeze.

How can you reach this insane temperature except in living in the Sahara ? My cards, some R9 380, run at 49C at stock clock. My fans are at about 70% I think, but I'm not sure, since it varies automatically according to the card temperature. Mine are newer, so this might explain, but running the card at a so high temperature is not healthy for it I think.

The R9 380 is more efficient than the 7970. But the OP's 7970 is too hot. That might be due to bad ventilation.

OP, reset your HSF.

Check for good contact of the HSF to the actual chip, and ensure that there is not too much thermal material used.  The application of the thermal material should be literally as thin as a credit card or thinner.

Also, try not to run your fans @ 100%.  You'll blow their brushes out.  Run at no greater RPM than say 80%.

I wanted to add some stuff for the ASUS 7970 Directcu II record:

I have just cleaned up 3 of these monsters.

1) The card(s) were taken apart to redo the thermal grease. I used a high quality paste and the ball method. An amount something on the size of a "small" pea was deposited on the chip center.

2) While the card(s) were apart, peripheral parts of the card(s) had fuzz, which was cleaned out.

3) IMHO I believe the OP's problem was an electronics issue and not a housekeeping/maintenance issue.

4) Important note: On 1 of 3 cards a fan was going bad. If a 5C or 10C degree  GPU temp increase has been observed, this means that the fan(s) are going. I purchased an exact replacement off of ebay for $23 inc shipping.

a) The replacement fans might be strung together a little differently. The main fan plug on the replacement was wired backwards as compared to the OEM setup.

b) If it is wired backwards, remove the plastic plug guide block from the gpu board and just rotate the fan plug by 180 degrees prior to insertion.

5) The most intriguing thing brought up by the poster is atitweak working under Ubuntu 15.10. I haven't seen a coherent account on how to do this.

I have the 7990 cards. I do not run them at the highest voltage and frequency. But they can get dirty easily.