Help with thread concurrency failure after testing a new card
Hello all, I need some help from you expert miners.
I've been running a modest rig with 2 x 5870's which have been happily running at ~420KH/s each. They were kind of free and I don't have a lot of spare cash. My rig is Win 7 x64 based with a 700W PSU. Neither card is running on risers but both stay below 75C most of the time. I'm using CGminer (3.1.0 at the moment).
Both were running more than happily on a thread concurrency of 8000, undervolted to 1.075V.
I picked up a cheap 7950 from ebay and thought I would swap out one of the 5870s. All three cards use 2 x 6pin PCIe power.
I managed to get the 5870 and 7950 running together but whenever I tried to increase the intensity over 14 on both cards, the system hangs and reboots. My guess is the PSU (cheap brand, again, inherited) isn't putting out enough power if the 7950 is more power hungry. I've got a power meter so I'll check this out over the following week.
My real problem came when I put the original 5870 back in and took the 7950 out, so I was effectively back to my original configuration. When I ran CGminer with my normal config file, CGminer crashed. I couldn't get it to work at all.
Eventually, I had to delete all the binaries associated with the cards and start with a batch file with almost nothing in it. Since then, I cannot get my cards to go above 6016 thread concurrency which has knocked about 50KH/s off both cards. I cannot understand why. Half the time, CGminer now just hangs on startup and won't start mining, however many times I reboot and I can't get it to work from a config file anymore.
The only other thing that's changed is I installed MSI afterburner (as the 7950 was an MSI) so I could undervolt the 7950 because cgminer was unable to. I've subsequently uninstalled it though since I took the 7950 out.
Any help or suggestions from anyone would be greatly welcome.