Bitcoin Forum

specs:

mobo: msi Z87a G-65 gaming
cpu: intel G3220 (1155)
OS: windows 8 pro
GPU: 3x Asus R9 280x DC 2 TOP version 1
PSU: Zalman 1250W 80plus platinum
Drivers: 13.11 beta 9.5 SDK v2.7
Cgminer: 3.7.2 , 3.3.1
Cgwatcher 1.3.5

config:

cgminer --scrypt -o asia.middlecoin.com:3333 -u -p x -g 2 -w 256 -d 0 -d 1 -d 2 -I 13,13,13 --thread-concurrency 8192,8192,8192 --gpu-engine 1120,1120,1120 --gpu-memclock 1800,1800,1800, --gpu-fan 100,100,100 --temp-cutoff 95,95,95 --temp-overheat 90,90,90 --temp-target 78,78,78

*cards running @ 770kh/s but after few minutes GPU 0 becomes sick/dead only reaching 80degrees temp.
*tried to interchange GPU slots but no improvement.
*always in n-1 situation where n = the number of cards I start with.
*tried to lower clock settings but still the same. still getting sick/dead GPU on 0

please help. will mine for 24hours paying out on your bitcoin address as bounty via middlecoin.

Seems like you over tuned your settings. 

Try 1080/1500 instead.

i already tried that. GPU 0 still get sick/dead after 5-10minutes

'm trying to do a fix right now and so far i am getting a positive result. running for 10minutes w/o problem. I just switched over the VGA cable to GPU 0 then reinstalled driver > reboot. will report back after an hour.





Edit - negative after 13minutes.

Is it always the same card that becomes sick, or is it any card on a particular PCI-e slot? Does it happen when only 1 card is plugged?

Are you using risers?

i'm not using risers. it's the GPU on the middle. it doesn't happen on single card.

May be it is too hot.

Did you try dummy plugs?
http://rumorscity.com/2013/12/06/how-to-create-dummy-plugs-for-your-graphics-cards/

Never tried that before. I will try to buy resistors and adapter later and see what it will do.

If the GPUs are on the motherboard, it can be high temperature of other component besides the GPU core itself, the vrm module, the ram, or even on the motherboard. Those can be at 100+ Cº by the time the core reaches 80ºC. Get speedfan and Gpu-Z softwares and see if the sensors report abnormally high temperatures (or abnormal voltages).

Use an household fan and point it to the rig. The rig is open-air, right?

this: -I 13 -g 2 -w 256 --thread-concurrency 8192 --gpu-engine 1035 --gpu-memclock 1500 --auto-fan --temp-cutoff 95 --temp-overheat 75 --auto-gpu


when a gpu gets sick it's because your HW (hardware errors) get too many, usually it's because of overclocking but I've come to believe it's mostly due to wrong thread conccurency. if that doesn't work do this:

-stop cgminer
-go to cgminer folder and delete ALL scryptXXXXXX.bin files
-go to your sh file (bat for windows) and put a simple command to start cgminer
eg: ./cgminer --scrypt -o pool -u user -p pass -I 13 -g 2 -w 256
-wait until cgminer connects to the pool and gets work
-stop it again
-go back to folder and look for the newly created scryptXXXXX.bin file, in it's name there should be tcZZZZ maybe 5 Z's, that's the correct thread concurrency you should be working at. start building around that.
-I only use 280x and all of mine work @ 8192 or 8128 so it should be one of those.

also do you use these ?

export DISPLAY=:0
export GPU_MAX_ALLOC_PERCENT=100
export GPU_USE_SYNC_OBJECTS=1

change export to SETX if you're running windows.

enjoy your newly stable rig :)

I reinstalled everything and new problem arise. Cgminer says tc is too high. Need to decrease tc or increase lg. But i am only using my old settings. Cgminer only start mining if tc is below 2000 however i only get 100 percent HW errors.


Edit: i tried to use the setx command to fix it but i am still getting same error. I guess i need to clean uninstall drivers then reinstall them.

use non-beta drivers and sdk 2.9  ;)