diff options
author | 2018-08-03 10:24:13 -0700 | |
---|---|---|
committer | 2018-08-03 10:28:08 -0700 | |
commit | e36c16c6720dd64ae5d8a1f8555102a1323af9ae (patch) | |
tree | 9076cc7fa753542490bac8b1fcd2bc3a3c2ac62a /third_party/python_runtime | |
parent | 7935c176118f0e50aa657a1c68a85430b70d2245 (diff) |
[XLA:GPU] Use strided batched gemm instead of building pointer tables.
This is mostly a huge amount of plumbing just to call into the cublas functions.
blasGemmStridedBatched has been available since CUDA 8.0.
For autotuning we'd need cublasGemmStridedBatchedEx, which is new in CUDA 9.2
so I didn't wire that up yet.
PiperOrigin-RevId: 207285707
Diffstat (limited to 'third_party/python_runtime')
0 files changed, 0 insertions, 0 deletions