when tuning --gemm or --trsm
Hi,
The AMD OpenCL BLAS library tuning application clAmdBlasTune fails for me on x86_64 Linux at this line:
40a85f: 48 83 b9 d8 00 00 00 cmpq $0x0,0xd8(%rcx)
The error is SIGSEGV - segmentation fault, accessing address 0xd8. It happens when trying to tune --gemm or --trsm. The --trmm however works fine. Could you please check if this error is effective for you, and fix it if possible?
Thanks!