提交 40450401 编写于 作者: T tensor-tang

change macro, can use omp when paddle use mklml

上级 00a51557
......@@ -30,7 +30,7 @@ void sgdUpdateCpu(real learningRate,
const real* grad,
real* momentumVec) {
decayRate *= learningRate;
#ifdef PADDLE_USE_MKLDNN
#ifdef PADDLE_USE_MKLML
#pragma omp parallel for
#endif
for (size_t i = 0; i < size; ++i) {
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册