Hey so we have this rig here that is acting strange.
MSI 890-FXAGD70 motherboard, latest bios, sempron cpu, 1GB of ram, 6x identical 5830s, 5 of them using 16x-16x ribbons, the other using a 16x-1x ribbon. 1200w (60A+60A) Thermaltake TR2 RX PSU.
Basically I install a fresh bamt to a 8GB USB. All 6 GPUs are showing up on the main screen and all 6 show up in gpumon. I set up the pools file, and enable the first card, set a conservative 888mhz core/300mhz mem clock, when its stable i will run it faster. Leaving fan @ 100%, and change nothing else. Restart mining, 1st card is mining perfectly. Proceed to the 2nd card, do the same thing, everything is going great. The bamt web interface shows up perfectly. I can get all 6 to mine @ 285 MH/s each and after about 10 shares one of the cards stops submitting shares. System will lock up, I reboot, and there's a 50/50 chance the bamt background shows up. I go into the configuration, disable all adapters except the first one, try restarting the mining and gpumon will not display at all, giving me random errors like "Invalid MIT-MAGIC-COOKIE-1 key
" and "o protocol specified
Pretty much irrelevant is my settings file since it is pretty much default, but i pasted it to pastebin here
to you can see and possibly make a suggestion. All our other systems are working just fine but this system is a fresh build with an unfamiliar psu and mobo, all of the rest of the components are known to be working. The PSU was brand new in the box and was installed just for this config.
I have a few theories in this. Brand new power supply might not be a very good one, I tested it with a voltage tester and I noticed the rails were almost where they should be but not quite ideal. It's so strange how a Rosewill 1000w can power more than 1200w Thermaltake, but I'm pretty sure the PSU is not the issue. To offset a bit of power, we tried borrowing two 6pin PCIe connectors from a PSU that was right beside this one, but exact same results. For some reason if I put a fresh copy of bamt on the USB it will work no problem right away but once it keels over then its hard to get back to a mining state.
The motherboard might be at fault, but why would it otherwise mine ok? Yesterday it was mining on all 6 cards for hours and then stopped, froze, and upon reboot it was not able to mine. SOLVED: I was using the command su - to change my user to root, and apparently that resets some environment settings that causes bamt to act a fool. Also, resetting the bios to v1.6 definitely helped our cause, everything is mining just perfectly now!