Bitcoin Forum
June 29, 2024, 02:08:23 PM *
News: Latest Bitcoin Core release: 27.0 [Torrent]
 
   Home   Help Search Login Register More  
Pages: [1]
  Print  
Author Topic: cgminer Asus HD7850 Display Driver Crashing while Litecoin Mining  (Read 6260 times)
bassminer (OP)
Newbie
*
Offline Offline

Activity: 5
Merit: 0


View Profile
December 06, 2013, 12:13:22 PM
 #1

Hi there,

Having just switched from Bitcoin mining I've been trying to get my setup to be stable enough to mine Litecoins. Its all going ok then there is the black screen and the display driver has stopped responding box.

My settings are currently running reasonably stable at 255Kh/s

ASUS ATI Radeon 7850 870MHz 2GB PCI-Express 3.0 HDMI DirectCU II
AMD Catalyst 13.4 Software Suite
AMD APP SDK v2.8

Batch file for cgminer is

setx GPU_MAX_ALLOC_PERCENT 100
setx GPU_USE_SYNC_OBJECTS 1
cgminer.exe --scrypt -o stratum+tcp://global.wemineltc.com:3335 -u xxx -p xxx -I 12

If I try increasing the intensity above 12/13 the chances of the display driver crash increase dramatically. Tried increasing/decreasing the GPU Speed, Memory Speed, GPU Voltage, Concurrancy, Shaders, Fan speed everything but the most simple command line seems to work best.

rem cgminer.exe --scrypt -o stratum+tcp://global.wemineltc.com:3335 -u xxx -p xxx --thread-concurrency 24000 --lookup-gap 0 --gpu-engine 1100 --gpu-memclock 1250 --gpu-powertune 20 -w 256 -I 13 -g 1
rem cgminer.exe --scrypt -o stratum+tcp://global.wemineltc.com:3335 -u xxx -p xxx --thread-concurrency 8192 --lookup-gap 0 --gpu-engine 1100 --gpu-memclock 1250 --gpu-powertune 20 -w 256 -I 13 -g 1
rem cgminer.exe --scrypt -o stratum+tcp://global.wemineltc.com:3335 -u xxx -p xxx --thread-concurrency 8192 --lookup-gap 0 --gpu-engine 1100 --gpu-memclock 1250 --gpu-powertune 20 -w 256 -I 18 -g 1 --shaders 1024
rem cgminer.exe --scrypt -o stratum+tcp://global.wemineltc.com:3335 -u xxx -p xxx --thread-concurrency 26624 --gpu-memclock 1250 --gpu-engine 1100 -I 13 --worksize 256 -g 1
rem cgminer.exe --scrypt -o stratum+tcp://global.wemineltc.com:3335 -u xxx -p xxx --thread-concurrency 26624 --gpu-engine 1100 --gpu-memclock 1250 --gpu-powertune 20 -w 256 -I 13 -g 1
rem cgminer.exe --scrypt -o stratum+tcp://global.wemineltc.com:3335 -u xxx -p xxx --thread-concurrency 26624 --gpu-engine 1100 --gpu-memclock 1250 -w 256 -I 13

I had some issues before with my 7850 after closing down the mining program during Bitcoin mining days. It had something to do with the 2D and 3D clock speeds. After mining it would switch to the 2D speed causing a voltage drop and I kept getting driver crashes. The solution was to setup a Catalyst profile so 2D Core 585MHz, Memory 680MHz, 3D Core 956MHz Memory 1210MHz. This kept the Voltage at 1.075V so It didn't cause the crash.

It will go up to 360Kh/s at higher intensities 18-20 so want to ideally mine at that rate. But it doesn't last long before the crash. I don't think its heat related as the fan is only on about 35% as I've tried running at 100% and it still crashes. My next plan might be a full re-install of Windows to start fresh with all the drivers as it looks like the GPU usage is cutting off dramatically then everything stops. The Clocks/Voltages/Temperature seem stable at the time so perhaps they are not related. Thanks for reading and if anyone has any comments or feedback it will be much appreciated. Of course will send a few token coins if any advice fixes the problem.
j0achim
Newbie
*
Offline Offline

Activity: 41
Merit: 0


View Profile WWW
December 06, 2013, 12:26:15 PM
 #2

Not all GFX cards are the same, for instance i have a 7970 that acts pretty much similar.

Example my 7970 has a weird sweet-spot, with stock memory frequency 1375 i have to underclock the engine to 843 (from 1000) and then i manage to pull 600Kh/s, otherwise with stock engine (1000) i only manage to pull 500~520k, my Radeon R9 280x (which are rebranded 7970 cards) they all give me a reasonable 710-720Kh/s without a single configuration.

Also my 7970 cannot mine with intensity above 13, with a activity of 64% Sad or the driver keeps crashing till the machine goes into bsod. But my R9 cards all manage to mine at I:18 just fine, though they pull off 710-720 at I:13 so there's no reason to push them too hard, as too high intensity in my opinion tend to make systems quite unstable.
bassminer (OP)
Newbie
*
Offline Offline

Activity: 5
Merit: 0


View Profile
December 06, 2013, 02:36:07 PM
 #3

Thanks for the comments, didn't know it was normal for some cards to not like higher intensities and hadn't considered under clocking below the stock speeds.

I've set them back to stock 870MHz GPU, 1210MHz Memory after another crash. This time it lasted a good 5-6 hours or so on slightly overclocked. Intensity is still at 12 and im getting 234Kh/s. 56C/Fan33%/GPU Load 97% 1.075V. If this is stable might increase the intensity to 13 and see how that goes or try underclocking a bit.
ilostcoins
Sr. Member
****
Offline Offline

Activity: 274
Merit: 250



View Profile
December 06, 2013, 03:17:03 PM
 #4

I think I have the same Asus model as you. I run with

-I 17 --thread-concurrency 20480 -w 256 -g 1
or
-I 13 --thread-concurrency 8192 -w 256 -g 2

Either are ok for me. I get around 300kh/s at GPU clock 860, memory clock 1150. I use powertune -15 to hopefully save some electricity as well.

LTC: LSyqwk4YbhBRtkrUy8NRdKXFoUcgVpu8Qb   NVC: 4HtynfYVyRYo6yM8BTAqyNYwqiucfoPqFW   TAG id: 4313
CMC: CAHrzqveVm9UxGm7PZtT4uj6su4suxKzZv   YAC: Y9m5S7M24sdkjdwxnA9GZpPez6k6EqUjUt
bassminer (OP)
Newbie
*
Offline Offline

Activity: 5
Merit: 0


View Profile
December 06, 2013, 04:44:23 PM
 #5

The model is HD7850-DC2-2GD5 also im using cgminer 3.5.0 which is another thing I can try and downgrade after reading around. Which version are you using btw?

Tried these ones.
--gpu-engine 860 --gpu-memclock 1150 -I 13 --thread-concurrency 8192 -w 256 -g 2
It lasted 20-30Mins @300k

--gpu-engine 860 --gpu-memclock 1150 -I 17 --thread-concurrency 20480 -w 256 -g 1
Lasted 10 mins Sad @300k

Not getting anywhere so trying to get to a known stable point and see how long that lasts. If that's no good then will try the downgrade of cgminer. With a conservative

--gpu-engine 860 --gpu-memclock 1150 -I 12 -w 256 --thread-concurrency 8192 -w 256 -g 1
ilostcoins
Sr. Member
****
Offline Offline

Activity: 274
Merit: 250



View Profile
December 07, 2013, 02:22:09 AM
 #6

My card is the same model as yours but is really stable with the settings I posted above. I've use cgminer 3.1.0, 3.2.2 and 3.7.2. My catalyst driver is version 13.4. My settings actually are on the conservative side and I generally don't like to push my hardware as hard as most people on this forum do.

It sounds like you may be unlucky with the hardware. Even cards of the same model aren't born equal and can be pushed to different degrees. Do you have problems playing demanding games? If you do and still have warranty, may be you can try to get a fix.

LTC: LSyqwk4YbhBRtkrUy8NRdKXFoUcgVpu8Qb   NVC: 4HtynfYVyRYo6yM8BTAqyNYwqiucfoPqFW   TAG id: 4313
CMC: CAHrzqveVm9UxGm7PZtT4uj6su4suxKzZv   YAC: Y9m5S7M24sdkjdwxnA9GZpPez6k6EqUjUt
j0achim
Newbie
*
Offline Offline

Activity: 41
Merit: 0


View Profile WWW
December 09, 2013, 10:48:41 PM
 #7

Oh another question what CPU do you have?


It seems there's a explanation why i get almost 200Kh/s more on my 280x vs. 7970 (stock configurations that is), my 7970 is in a PC with a i7 2600K. Now you migh think what the heck does this have to do with the GPU. its quite simple the sandy bridge CPU supports PCI-E 2.0 while the 7970 is a PCI-E 3.0 gfx card. (my 280x cards run in a brand new PC on a ivy bridge i7-4820K (PCI-E 3.0))
bassminer (OP)
Newbie
*
Offline Offline

Activity: 5
Merit: 0


View Profile
December 10, 2013, 09:50:41 PM
 #8

I had the card slightly overclocked when mining Bitcoins hopefully that has not damaged it, its out of warranty now but its paid for itself Smiley Giving a 3.7.2 a go to see if it makes a difference but saw on the wemineltc site it said "Versions higher than 3.5 do NOT support LiteCoin mining" but it seems to be working. Games work fine except when I close them the voltage dropped too low when switched back to 2D causing a crash but i've patched that more than solved it.

My CPU is an i5 3570k on a GigaByte Z77-D3H it supports PCI-E 3.0 so should give it half a chance Smiley
Pages: [1]
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.19 | SMF © 2006-2009, Simple Machines Valid XHTML 1.0! Valid CSS!