I've recently updated a system to Fedora 24 to start running this project again.
I finally got three jobs completed, but system rebooted this evening, and my system now fails to recognized the GPUs. Error message follows:

<pre>
05-Sep-2016 20:34:27 [---] OpenCL: NVIDIA GPU 0: GeForce GTX 970 (driver version 367.44, device version OpenCL 1.2 CUDA, 4036MB, 4036MB available, 1642 GFLOPS peak)
05-Sep-2016 20:34:27 [---] OpenCL: NVIDIA GPU 1: GeForce GTX 650 Ti BOOST (driver version 367.44, device version OpenCL 1.2 CUDA, 1999MB, 1999MB available, 437 GFLOPS peak)
05-Sep-2016 20:34:27 [---] App version needs CUDA but GPU doesn't support it
05-Sep-2016 20:34:27 [GPUGRID] Application uses missing NVIDIA GPU
</pre>


Seem to be missing previously seen "CUDA: " lines in output.
Is there someplace where I can download appropriate libraries?

I used synaptic for ubuntu but a similar program for fedora would be yumex, could take a look if there are any cuda packages there.

ls -al /usr/lib/libcuda.* look like this?
[root@Centos7 /]# ls -alr /usr/lib/libcuda.*
-rwxr-xr-x. 1 root root 7747600 Aug 25 20:11 /usr/lib/libcuda.so.367.44
lrwxrwxrwx. 1 root root 17 Aug 25 20:12 /usr/lib/libcuda.so.1 -> libcuda.so.367.44
lrwxrwxrwx. 1 root root 12 Aug 25 20:12 /usr/lib/libcuda.so -> libcuda.so.1

the cuda libraries and modules should have been installed as part of installer for the kernel module.

did nouveau get enabled again somehow?

Please enable coproc_debug in "event log diagnostic flags" and post log of boinc starting up.
Also the latest drivers cause issues, 367.35 are latest that work for me (arch).
Sometimes boinc looks for gpu libs in other places than they are. nVidia OpenCL requires cuda to be working so that is stange.

---