leavelet/cutlass
1
0
Fork 0
Code Issues Pull Requests Actions 26 Packages Projects Releases Wiki Activity

CUTLASS 2.5

Browse Source
This commit is contained in:
Andrew Kerr
2021-02-26 09:58:26 -05:00
parent ccb697bac7
commit 0e13748649
771 changed files with 15474 additions and 1715 deletions
Download Patch File Download Diff File Expand all files Collapse all files
examples/09_turing_tensorop_conv2dfprop/CMakeLists.txt
+1 -1
View File
@@ -1,4 +1,4 @@
# Copyright (c) 2017-2020, NVIDIA CORPORATION. All rights reserved.
# Copyright (c) 2017-2021, NVIDIA CORPORATION. All rights reserved.
#
# Redistribution and use in source and binary forms, with or without modification, are permitted
# provided that the following conditions are met:
examples/09_turing_tensorop_conv2dfprop/turing_tensorop_conv2dfprop.cu
+7 -1
View File
@@ -1,5 +1,5 @@
/***************************************************************************************************
* Copyright (c) 2017-2020, NVIDIA CORPORATION. All rights reserved.
* Copyright (c) 2017-2021, NVIDIA CORPORATION. All rights reserved.
*
* Redistribution and use in source and binary forms, with or without modification, are permitted
* provided that the following conditions are met:
@@ -485,6 +485,7 @@ Result profile_convolution(Options const &options) {
// Split K dimension into 1 partitions
int split_k_slices = 1;
// Construct Conv2dProblemSize with user defined output size
cutlass::conv::Conv2dProblemSize problem_size(
options.input_size,
options.filter_size,
@@ -495,6 +496,8 @@ Result profile_convolution(Options const &options) {
mode,
split_k_slices);
// Construct ImplicitGemm::Argument structure with conv2d
// problem size, data pointers, and epilogue values
typename ImplicitGemm::Arguments arguments{
problem_size,
tensor_a.device_ref(),
@@ -515,6 +518,9 @@ Result profile_convolution(Options const &options) {
// Allocate workspace memory
cutlass::device_memory::allocation<uint8_t> workspace(workspace_size);
result.status = implicit_gemm_op.can_implement(arguments);
CUTLASS_CHECK(result.status);
result.status = implicit_gemm_op.initialize(arguments, workspace.get());
CUTLASS_CHECK(result.status);