Hello, I have followed the guide installing the latest version of ATI STREAM
SDK as well as pyopencl but matrix multiply is way too slow.
The output is saying that it takes 44 seconds for the GPU method, but with
the numpy method it takes only .35 seconds. Here is the output:
*** CAL version mismatch:
This OpenCL build requires version 1.4.467, version 1.4.403 installed.
GPU (s) total: 44.1569998264
CPU (s) 0.353999853134
GPU speedup: 0.00801684567624
when I run the program, my Quad Core computer is maxed out to 100% CPU
usage. I have 2 ATI 4890's in Crossfire. I dont understand why it is so
Show replies by date