New updates for 2.11 (#775)
* New updates. * Minor profiler updates Co-authored-by: Aniket Shivam <ashivam@nvidia.com>
This commit is contained in:
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
@@ -30,12 +30,8 @@
|
||||
**************************************************************************************************/
|
||||
|
||||
/*! \file
|
||||
|
||||
\brief A file contains the epilogue visitor with CTA row-wise broadcast
|
||||
|
||||
The epilogue rearranges the result of a matrix product through shared memory to match canonical
|
||||
tensor layouts in global memory. Epilogues support conversion and reduction operations.
|
||||
|
||||
\brief A generic wrapper around an epilogue visitor operation
|
||||
*/
|
||||
|
||||
#pragma once
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
@@ -30,8 +30,8 @@
|
||||
**************************************************************************************************/
|
||||
|
||||
/*! \file
|
||||
|
||||
\brief A file contains the binary ops
|
||||
|
||||
\brief Binary operations to be used within the epilogue visitor model.
|
||||
*/
|
||||
|
||||
#pragma once
|
||||
@@ -44,7 +44,7 @@ namespace cutlass {
|
||||
/////////////////////////////////////////////////////////////////////////////////////////////////
|
||||
|
||||
|
||||
/// Scalar multiplication
|
||||
/// Elementwise addition of two arrays
|
||||
template <typename T, int N>
|
||||
struct VectorAdd {
|
||||
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
@@ -30,8 +30,8 @@
|
||||
**************************************************************************************************/
|
||||
|
||||
/*! \file
|
||||
|
||||
\brief A file contains the unary ops
|
||||
|
||||
\brief Unary operations to be used within the epilogue visitor model.
|
||||
*/
|
||||
|
||||
#pragma once
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
@@ -30,8 +30,8 @@
|
||||
**************************************************************************************************/
|
||||
|
||||
/*! \file
|
||||
|
||||
\brief A file contains the epilogue visitor Op with accumulator
|
||||
|
||||
\brief Epilogue visitor operation that simply returns the accumulator
|
||||
*/
|
||||
|
||||
#pragma once
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
@@ -30,8 +30,8 @@
|
||||
**************************************************************************************************/
|
||||
|
||||
/*! \file
|
||||
|
||||
\brief A file contains the epilogue visitor Op with Binary op
|
||||
|
||||
\brief Epilogue visitor operator performing a binary operation between two visitor nodes
|
||||
*/
|
||||
|
||||
#pragma once
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
@@ -30,8 +30,8 @@
|
||||
**************************************************************************************************/
|
||||
|
||||
/*! \file
|
||||
|
||||
\brief A file contains the epilogue visitor Op with broadcasting vector to all columns
|
||||
|
||||
\brief Epilogue visitor operation that broadcasts a vector to all columns
|
||||
*/
|
||||
|
||||
#pragma once
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
@@ -30,8 +30,8 @@
|
||||
**************************************************************************************************/
|
||||
|
||||
/*! \file
|
||||
|
||||
\brief A file contains the epilogue visitor Op with reduction over columns in CTA
|
||||
|
||||
\brief Epilogue visitor operation that performs a column-wise reduction within a threadblock
|
||||
*/
|
||||
|
||||
#pragma once
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
@@ -30,8 +30,8 @@
|
||||
**************************************************************************************************/
|
||||
|
||||
/*! \file
|
||||
|
||||
\brief A file contains the epilogue visitor Op with Linear Combination
|
||||
|
||||
\brief Epilogue visitor operation that performs a linear combination of two visitor nodes
|
||||
*/
|
||||
|
||||
#pragma once
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
@@ -30,8 +30,8 @@
|
||||
**************************************************************************************************/
|
||||
|
||||
/*! \file
|
||||
|
||||
\brief A file contains the epilogue visitor Op with broadcasting vector to all rows
|
||||
|
||||
\brief Epilogue visitor operation that broadcasts a vector to all rows
|
||||
*/
|
||||
|
||||
#pragma once
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
@@ -30,8 +30,8 @@
|
||||
**************************************************************************************************/
|
||||
|
||||
/*! \file
|
||||
|
||||
\brief A file contains the epilogue visitor Op with reduction over rows in CTA
|
||||
|
||||
\brief Epilogue visitor operation that performs a column-wise reduction within a threadblock
|
||||
*/
|
||||
|
||||
#pragma once
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
@@ -30,8 +30,8 @@
|
||||
**************************************************************************************************/
|
||||
|
||||
/*! \file
|
||||
|
||||
\brief A file contains the epilogue visitor Op with Unary operation
|
||||
|
||||
\brief Epilogue visitor operator performing a unary operation atop a visitor node
|
||||
*/
|
||||
|
||||
#pragma once
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
@@ -30,12 +30,11 @@
|
||||
**************************************************************************************************/
|
||||
|
||||
/*! \file
|
||||
\brief A file contains all functioning classes needed by GemmLayernorm.
|
||||
\brief Epilogue visitor type used for partial computation of a layernorm operation
|
||||
|
||||
GemmLayernorm example = GEMM0 with partial reduction fused in epilogue (EpilogueVisitorLayerNorm)
|
||||
+ lightweight full reduction kernel (ApplyFinalReduction)
|
||||
+ GEMM1 with elemenwise operations fused in mainloop (GemmLayernormMainloopFusion)
|
||||
|
||||
+ GEMM1 with elementwise operations fused in mainloop (GemmLayernormMainloopFusion)
|
||||
*/
|
||||
|
||||
#pragma once
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
|
||||
@@ -1,5 +1,5 @@
|
||||
/***************************************************************************************************
|
||||
* Copyright (c) 2017 - 2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* Copyright (c) 2017 - 2023 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
|
||||
* SPDX-License-Identifier: BSD-3-Clause
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
|
||||
Reference in New Issue
Block a user