diff options
author | Cedric Nugteren <web@cedricnugteren.nl> | 2017-11-02 21:47:14 +0100 |
---|---|---|
committer | Cedric Nugteren <web@cedricnugteren.nl> | 2017-11-02 21:47:14 +0100 |
commit | 9b0a435fb00b845b875590be90acffcd4f3bb009 (patch) | |
tree | 754b523789ef717619b540925c97e7167ba28f06 /CHANGELOG | |
parent | 73272ab97dbd5abe757f6558c9b89665c5ac99d0 (diff) |
Integrated the GEMM routine tuner for kernel selection; added first tuning results
Diffstat (limited to 'CHANGELOG')
-rw-r--r-- | CHANGELOG | 1 |
1 files changed, 1 insertions, 0 deletions
@@ -8,6 +8,7 @@ Development (next version) * All correctness tests and performance clients work on CUDA like they did for OpenCL - Kernels are now cached based on their tuning parameters: fits the use-case of 'OverrideParameters' - Improved performance for small GEMM problems by going from 3 to 1 optional temporary buffers +- GEMM kernel selection (direct vs in-direct) is now done automatically using a new tuner - Various minor fixes and enhancements - Added tuned parameters for various devices (see README) |