Skip to content

P
Paddle

BaiXuePrincess / Paddle
与 Fork 源项目一致

Fork自 PaddlePaddle / Paddle

通知 1
Star 1
Fork 0
P
Paddle
未验证 提交 79095918 编写于 作者: T taixiurong 提交者： GitHub
浏览文件
操作

fix build in xpu (#37699)

上级 feda7c1d
隐藏空白更改
内联 并排
Showing with 6 addition and 4 deletion
paddle/fluid/eager/accumulation/gradient_accumulation.cc
浏览文件 @ 79095918
......@@ -193,13 +193,14 @@ void TensorAdd(const egr::EagerTensor& src, egr::EagerTensor* dst) {
// TODO(jiabin): Support NPU here
PADDLE_TENSOR_ADD(float);
// NOTE(phlrain): xpu only support float
// NOTE(phlrain): xpu only support float
#ifndef PADDLE_WITH_XPU
PADDLE_TENSOR_ADD(double);
// NOTE(chenweihang): only support complex grad tensor accumulated,
// support selected rows if needed in the future
PADDLE_TENSOR_ADD(paddle::platform::complex<float>);
PADDLE_TENSOR_ADD(paddle::platform::complex<double>);
#endif
#undef PADDLE_TENSOR_ADD
if (data_type == paddle::framework::proto::VarType::FP16) {
......@@ -268,13 +269,14 @@ void VariableAdd(const egr::EagerTensor& src, egr::EagerTensor* dst) {
// TODO(jiabin): Support NPU here
PADDLE_TENSOR_ADD(float);
// NOTE(phlrain): xpu only support float
// NOTE(phlrain): xpu only support float
#ifndef PADDLE_WITH_XPU
PADDLE_TENSOR_ADD(double);
// NOTE(chenweihang): only support complex grad tensor accumulated,
// support selected rows if needed in the future
PADDLE_TENSOR_ADD(paddle::platform::complex<float>);
PADDLE_TENSOR_ADD(paddle::platform::complex<double>);
#endif
#undef PADDLE_TENSOR_ADD
if (data_type == paddle::framework::proto::VarType::FP16) {
......
Markdown is supported
0% 或 .
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 或