Reshape batch fix
This commit is contained in:
@@ -40,7 +40,7 @@ public:
|
||||
dnnType *srcData = (dnnType*)reinterpret_cast<const dnnType*>(inputs[0]);
|
||||
dnnType *dstData = reinterpret_cast<dnnType*>(outputs[0]);
|
||||
|
||||
checkCuda( cudaMemcpy(dstData, srcData, c*h*w*sizeof(dnnType), cudaMemcpyDeviceToDevice));
|
||||
checkCuda( cudaMemcpyAsync(dstData, srcData, batchSize*c*h*w*sizeof(dnnType), cudaMemcpyDeviceToDevice, stream));
|
||||
return 0;
|
||||
}
|
||||
|
||||
|
||||
Reference in New Issue
Block a user