CUTLASS 3.2.1 (#1113)
* Updates for 3.2.1 release. * Minor fix in gemm op profiler for raster order. * Add scheduler mapping for raster order in the kernels.
This commit is contained in:
@@ -157,14 +157,6 @@ protected:
|
||||
CUTLASS_TRACE_HOST(" cudaFuncSetAttribute() returned error " << cudaGetErrorString(cudart_result));
|
||||
return Status::kErrorInternal;
|
||||
}
|
||||
|
||||
cudart_result = cudaFuncSetAttribute(
|
||||
Kernel2<GemmKernel>,
|
||||
cudaFuncAttributePreferredSharedMemoryCarveout, 100); // 100% shared memory
|
||||
if (cudart_result != cudaSuccess) {
|
||||
CUTLASS_TRACE_HOST(" cudaFuncSetAttribute() returned error " << cudaGetErrorString(cudart_result));
|
||||
return Status::kErrorInternal;
|
||||
}
|
||||
}
|
||||
|
||||
// Update SM occupancy member
|
||||
@@ -228,12 +220,6 @@ public:
|
||||
{
|
||||
CUTLASS_TRACE_HOST("GemmUniversalBase::can_implement()");
|
||||
|
||||
// Initialize static kernel and device properties, if necessary.
|
||||
Status result = init_device_props();
|
||||
if (result != Status::kSuccess) {
|
||||
return result;
|
||||
}
|
||||
|
||||
dim3 grid = get_grid_shape(args);
|
||||
|
||||
if (!(grid.y <= std::numeric_limits<uint16_t>::max() &&
|
||||
|
||||
Reference in New Issue
Block a user