projects / platform / upstream / openblas.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 6adc4b7) | patch
Improve performance of GEMM for small matrices when SMP is defined.
authorCraig Donner <cdonner@google.com>
Thu, 7 Jun 2018 13:54:42 +0000 (14:54 +0100)
committerCraig Donner <cdonner@google.com>
Thu, 7 Jun 2018 14:29:13 +0000 (15:29 +0100)
commit66316b9f4c8c7c48eed8b29e86f64581c02d45b0
tree7d4b51bc4ad344d93671bebb9c1d3f4b634448d1tree | snapshot
parent6adc4b7b363bf85c53b83968a146f69fb0601943commit | diff
Improve performance of GEMM for small matrices when SMP is defined.

Always checking num_cpu_avail() regardless of whether threading will actually
be used adds noticeable overhead for small matrices.  Most other uses of
num_cpu_avail() do so only if threading will be used, so do the same here.
interface/gemm.c diff | blob | history
interface/trsm.c diff | blob | history
Domain: Machine Learning / ML Framework;