fix gpu ops bug.

mindspore/ccsrc/backend/kernel_compiler/gpu/arrays/split_gpu_kernel.cc

@@ -27,5 +27,8 @@ MS_REG_GPU_KERNEL_ONE(Split,
 MS_REG_GPU_KERNEL_ONE(
   Split, KernelAttr().AddAllSameAttr(true).AddInputAttr(kNumberTypeFloat16).AddOutputAttr(kNumberTypeFloat16),
   SplitGpuFwdKernel, half)
+MS_REG_GPU_KERNEL_ONE(
+  Split, KernelAttr().AddAllSameAttr(true).AddInputAttr(kNumberTypeUInt32).AddOutputAttr(kNumberTypeUInt32),
+  SplitGpuFwdKernel, uint32_t)
 }  // namespace kernel
 }  // namespace mindspore

mindspore/ccsrc/backend/kernel_compiler/gpu/cuda_impl/split_impl.cu

@@ -48,3 +48,6 @@ template void SplitKernel(const size_t size, const int axis_step, const int all_
 template void SplitKernel(const size_t size, const int axis_step, const int all_size_before_axis,
                           const int all_size_axis, const half* input, half** outputs,
                           cudaStream_t cuda_stream);
+template void SplitKernel(const size_t size, const int axis_step, const int all_size_before_axis,
+                          const int all_size_axis, const uint32_t* input, uint32_t** outputs,
+                          cudaStream_t cuda_stream);

mindspore/ccsrc/backend/kernel_compiler/gpu/nn/sgd_gpu_kernel.h

@@ -42,12 +42,18 @@ class SGDGpuKernel : public GpuKernel {
     T *accum = GetDeviceAddress<T>(inputs, 3);
     T *momentum = GetDeviceAddress<T>(inputs, 4);
     T *stat = GetDeviceAddress<T>(inputs, 5);
+    T *output_param = GetDeviceAddress<T>(outputs, 0);
 
     SGD(size_, dampening_, weight_decay_, nesterov_, lr, momentum, grad, param, accum, stat,
         reinterpret_cast<cudaStream_t>(stream));
+    CHECK_CUDA_RET_WITH_EXCEPT(kernel_node_,
+                               cudaMemcpyAsync(output_param, param, sizeof(T) * size_, cudaMemcpyDeviceToDevice,
+                                               reinterpret_cast<cudaStream_t>(stream)),
+                               "SGD cudaMemcpyAsync params to outputs failed");
     return true;
   }
   bool Init(const CNodePtr &kernel_node) override {
+    kernel_node_ = kernel_node;
     dampening_ = GetAttr<float>(kernel_node, "dampening");
     weight_decay_ = GetAttr<float>(kernel_node, "weight_decay");
     nesterov_ = GetAttr<bool>(kernel_node, "nesterov");