search

clMathLibraries / clBLAS

a software library containing BLAS functions written in OpenCL
Apache License 2.0
839 stars 240 forks source link

Build Error #20

Closed ashbachj closed 10 years ago

ashbachj commented 10 years ago

Hi, I am attempting to run some examples and the tests before I incorporate clBLAS into my project. Pretty much everything I run results in the following error

AN INTERNAL KERNEL BUILD ERROR OCCURRED! device name = Quadro FX 5800 error = -11 memory pattern = Cached global memory based block gemm, computing kernel generator Subproblem dimensions: dims[0].itemY = 64, dims[0].itemX = 32, dims[0].y = 64, dims[0].x = 32, dims[0].bwidth = 4; ; dims[1].itemY = 8, dims[1].itemX = 4, dims[1].y = 8, dims[1].x = 4, dims[1].bwidth = 4; ; Parallelism granularity: pgran->wgDim = 2, pgran->wgSize[0] = 8, pgran->wgSize[1] = 8, pgran->wfSize = 64 Kernel extra flags: 939554576 Source: source dump cut

Build log:

Error in processing command line: Don't understand command line argument "-g"!

Would you be able to assist me with this problem? Thanks

kknox commented 10 years ago

This is a shot in the dark, but try commenting out the '-g' compiler option at: clMathLibraries\clBLAS\src\library\blas\generic\common.c:527

The error message sounds like your installed opencl runtime is not accepting the '-g' flag.

kknox commented 10 years ago

Closing this issue for lack of activity

DonkeyDragon commented 10 years ago

Just wanted to confirm that commenting out line 529 of clMathLibraries\clBLAS\src\library\blas\generic\common.c fixed the error for me. Tested on various Nvidia configurations.

jonathantompson commented 9 years ago

I also wanted to confirm that this was also a problem (using OpenCL.lib from NVidia's CUDA 6.5 SDK). Commenting out the -g flag line also worked for me.