It seems that a lot of GPU problems revolve around specific versions of drivers. Though NVidia has their own support structure, you can often learn from information reported by others who fold.
It was a lot simpler when GPUs only needed to be supported on Windows. Everybody could use Microsoft's method of enumerating hardware. V7 was aimed at working also on Linux and (eventually) OS-X so FAH had to adopt a non-proprietary methodology.
There are several open tickets that it only works right randomly when lspci happens to match the Windows sequence.
If you run FAHBench and look at how it chooses which GPU to test, it will help you understand many of the underlying details. One order is how the GPUs appear in the V7 system config. Then, CUDA and OpenCL drivers may or may not agree depending on what each of them finds.
Got them going, and thanks everyone who took the time to post for their input.
Uninstalling the client w/ data, and reinstalling the cards with drivers and fresh client install never worked. It did not "auto discover" properly out of the 7 or 8 times I tried it.
Was able to tweak the slot indexes manually and eventually found a working approach, more trial and error than anything else.
Tweaking the gpu-index makes sense but when to change the opencl-index and cuda-index, and what to change them to, still confuses me.
Hardware configuration: Intel i7-4770K @ 4.5 GHz, 16 GB DDR3-2133 Corsair Vengence (black/red), EVGA GTX 760 @ 1200 MHz, on an Asus Maximus VI Hero MB (black/red), in a blacked out Antec P280 Tower, with a Xigmatek Night Hawk (black) HSF, Seasonic 760w Platinum (black case, sleeves, wires), 4 SilenX 120mm Case fans with silicon fan gaskets and silicon mounts (all black), a 512GB Samsung SSD (black), and a 2TB Black Western Digital HD (silver/black).
Even if there is no OC'ing going on it is worth using one of the OC'ing utilities just so you can continuously monitor the state of clocks, temps, and % GPU usage in the systray.
Hey PantherX, thanks for the detailed steps. One question, I can see how you get the gpu index, but how do you get the opencl and cuda indicies? Thanks again for the help!
cordis wrote:Hey PantherX, thanks for the detailed steps. One question, I can see how you get the gpu index, but how do you get the opencl and cuda indicies? Thanks again for the help!
Run FAHBench. It will show you which devices are identified as an OpenCL device and in what order and which devices are identified as a CUDA device in what order, even if you only get as far as deciding how to select the device you want to test.
What a week!
Thanks for the suggestions.
Had to reinstall everything several times to figure out how to get the 2 cards working again, and of course the heat was a major issue.
For the time being I just took the side of the case off and am blowing a fan directly onto the video cards. Just going to have to blow them off more often.
When I turn the fan off I can see the temperatures start climbing immediately with gpu-z.
Running fine for 72 hours now.