fix gpu to gpu copy bug (#72792)

wanghuancoder · web-flow · commit b94ea2194a9c · 2025-05-20T10:22:32.000+08:00
diff --git a/paddle/phi/core/tensor_utils.cc b/paddle/phi/core/tensor_utils.cc
@@ -169,6 +169,7 @@ void Copy(const Context& dev_ctx,
     auto src_gpu_place = src_place;
     auto dst_gpu_place = dst_place;
     auto ctx_place = dev_ctx.GetPlace();
+
     PADDLE_ENFORCE_EQ(
         ctx_place.GetType() == AllocationType::GPU,
         true,
@@ -178,18 +179,19 @@ void Copy(const Context& dev_ctx,
     auto stream =
         blocking ? nullptr
                  : reinterpret_cast<const phi::GPUContext&>(dev_ctx).stream();
-    if (src_place.GetType() == dst_place.GetType()) {
+    if (src_place.GetDeviceId() == dst_place.GetDeviceId()) {
       memory_utils::Copy(
           dst_gpu_place, dst_ptr, src_gpu_place, src_ptr, size, stream);
     } else {
-      if (ctx_place.GetType() == src_place.GetType()) {
+      if (ctx_place.GetDeviceId() == src_place.GetDeviceId()) {
         memory_utils::Copy(
             dst_gpu_place, dst_ptr, src_gpu_place, src_ptr, size, stream);
         phi::DeviceContextPool::Instance().Get(src.place())->Wait();
-      } else if (ctx_place.GetType() == dst_place.GetType()) {
+      } else if (ctx_place.GetDeviceId() == dst_place.GetDeviceId()) {
         phi::DeviceContextPool::Instance().Get(src.place())->Wait();
         memory_utils::Copy(
             dst_gpu_place, dst_ptr, src_gpu_place, src_ptr, size, stream);
+        phi::DeviceContextPool::Instance().Get(dst_place)->Wait();
       } else {
         PADDLE_THROW(errors::Unavailable(
             "Context place dose not match the source and destination place."));
