Description of problem: Launched the program without access to the GPU, setting CUDA_VISIBLE_DEVICES=''. It immediately crashed. Version-Release number of selected component: clblast-tuners-1.5.0-3.fc31 Additional info: reporter: libreport-2.11.3 backtrace_rating: 4 cmdline: clblast_tuner_xgemm crash_function: clblast::CLCudaAPIError::Check executable: /usr/bin/clblast_tuner_xgemm journald_cursor: s=d1a52455c6e14670938f25393295ed7b;i=40bcc;b=03ca2d63652a45f9a5fd1b2871c6981c;m=4141cff210;t=59e38e430f476;x=3646678d7e3aa038 kernel: 5.4.17-200.fc31.x86_64 rootdir: / runlevel: N 5 type: CCpp uid: 1000 Truncated backtrace: Thread no. 1 (5 frames) #6 clblast::CLCudaAPIError::Check at /usr/src/debug/clblast-1.5.0-3.fc31.x86_64/src/clpp11.hpp:82 #7 clblast::Platform::NumDevices at /usr/src/debug/clblast-1.5.0-3.fc31.x86_64/src/clpp11.hpp:191 #8 clblast::Device::Device at /usr/src/debug/clblast-1.5.0-3.fc31.x86_64/src/clpp11.hpp:237 #9 clblast::Tuner<float>(int, char**, int, std::function<clblast::TunerDefaults (int)>, std::function<clblast::TunerSettings (int, clblast::Arguments<float> const&)>, std::function<void (int, clblast::Arguments<float> const&)>, std::function<std::vector<clblast::Constraint, std::allocator<clblast::Constraint> > (int)>, std::function<clblast::LocalMemSizeInfo (int)>, std::function<void (int, clblast::Kernel&, clblast::Arguments<float> const&, std::vector<clblast::Buffer<float>, std::allocator<clblast::Buffer<float> > >&)>) at /usr/include/c++/9/bits/std_function.h:685 #10 StartVariation<1> at /usr/include/c++/9/new:174
Created attachment 1662177 [details] File: backtrace
Created attachment 1662178 [details] File: cgroup
Created attachment 1662179 [details] File: core_backtrace
Created attachment 1662180 [details] File: cpuinfo
Created attachment 1662181 [details] File: dso_list
Created attachment 1662182 [details] File: environ
Created attachment 1662183 [details] File: limits
Created attachment 1662184 [details] File: machineid
Created attachment 1662185 [details] File: maps
Created attachment 1662186 [details] File: mountinfo
Created attachment 1662187 [details] File: open_fds
Created attachment 1662188 [details] File: proc_pid_status
It looks like the tuning code tried to count the number of CUDA/OpenCL devices on the system. When the count came out zero, it threw an exception, CLCudaAPIError. Nothing catches that exception, so you got the default C++ abort due to an uncaught exception. I have asked upstream to please catch that exception. In the meantime, all I can suggest is that you don't try to tune 0 devices. :-) Thanks for the report.
FEDORA-2020-ea0c27ac1b has been submitted as an update to Fedora 31. https://bodhi.fedoraproject.org/updates/FEDORA-2020-ea0c27ac1b
clblast-1.5.1-1.fc31 has been pushed to the Fedora 31 testing repository. If problems still persist, please make note of it in this bug report. See https://fedoraproject.org/wiki/QA:Updates_Testing for instructions on how to install test updates. You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-2020-ea0c27ac1b
clblast-1.5.1-1.fc31 has been pushed to the Fedora 31 stable repository. If problems still persist, please make note of it in this bug report.