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 da558f0e 编写于 作者: Q Qi Li 提交者： GitHub
浏览文件
操作

[ROCm] fix bfloat16 support, test=develop (#40401)

上级 60899549
隐藏空白更改
内联 并排
Showing with 1 addition and 1 deletion
paddle/fluid/imperative/gradient_accumulator.cc
浏览文件 @ da558f0e
...@@ -423,7 +423,7 @@ void TensorAdd(const VarType& src, VarType* dst) { ...@@ -423,7 +423,7 @@ void TensorAdd(const VarType& src, VarType* dst) {
} }
if (data_type == framework::proto::VarType::BF16) { if (data_type == framework::proto::VarType::BF16) {
if (platform::is_gpu_place(place)) { if (platform::is_gpu_place(place)) {
#if defined(PADDLE_WITH_CUDA) #if defined(PADDLE_WITH_CUDA) || defined(PADDLE_WITH_HIP)
return TensorAddImpl<platform::CUDADeviceContext, platform::bfloat16>( return TensorAddImpl<platform::CUDADeviceContext, platform::bfloat16>(
src_tensor, dst_tensor, place); src_tensor, dst_tensor, place);
#else #else
......
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