This repository was archived by the owner on Aug 11, 2023. It is now read-only.
This repository was archived by the owner on Aug 11, 2023. It is now read-only.
Memory issues in matrix_multiply example #7
Description
Reproduced this on both Hawaii and Fiji GPUs. When given matrix size is not a power of two, it fails.
Steps to reproduce:
$ git clone https://github.com/codeplaysoftware/computecpp-sdk $ cd computecpp-sdk/samples $ cmake . -DCOMPUTECPP_PACKAGE_ROOT_DIR=/usr/local $ cd matrix_multiply $ ./matrix_multiply 32 Input matrix C++: Time: 0 GFLOPs: inf The Device Max Work Group Size is : 256 The order is : 32 The blockSize is : 8 SYCL: Time: 24 GFLOPs: 0.00273067 Output Success $ ./matrix_multiply 33 Input matrix C++: Time: 1 GFLOPs: 0.071874 *** Error in `./matrix_multiply': free(): invalid next size (normal): 0x00000000015a7d40 *** Aborted (core dumped) I am hoping this is only a problem in the sample?