Release v4.0.0 (#2294)

This commit is contained in:
Kihiro Bando
2025-05-13 15:55:29 -04:00
committed by GitHub
parent ad7b2f5e84
commit f115c3f854
299 changed files with 51495 additions and 4413 deletions
@@ -5,7 +5,7 @@ Users and developers may build either
in Visual Studio's graphical integrated development environment,
or on the command line with `cmake --build`.
# Software prerequisites
## Software prerequisites
1. Windows 10 or 11
@@ -22,7 +22,7 @@ or on the command line with `cmake --build`.
Visual Studio must be installed *before* the CUDA Toolkit.
Otherwise, Visual Studio's build system won't know about CUDA.
# Operating system settings
## Operating system settings
By default, Windows restricts the maximum file path length (`MAX_PATH`) to 260 characters.
CUTLASS has many files and directory paths that challenge this requirement.
@@ -48,7 +48,7 @@ before attempting to clone or build CUTLASS.
[This Microsoft help article](https://learn.microsoft.com/en-us/windows/win32/fileio/maximum-file-path-limitation?tabs=registry)
explains different ways to change the registry setting.
# Set up build environment
## Set up build environment
1. Run "git bash" to get a familiar command-line interface
@@ -62,7 +62,7 @@ explains different ways to change the registry setting.
Alternate approaches may rely on the CMake GUI and/or Windows' native command line.
# Building
## Building
A successful CMake run will create a `CUTLASS.sln` Visual Studio "solution" file in the build directory.
One can open this in Visual Studio and build the entire solution or any subset of projects as desired.
@@ -77,7 +77,7 @@ Unlike with CMake's Makefile or Ninja generators,
`CMAKE_BUILD_TYPE` has no effect on the Visual Studio generator,
because the Visual Studio generator creates all build configurations.
# Tips
## Tips
With Windows builds, one may find that CMake reruns unnecessarily.
For example, cancelling a build and starting it again may rerun CMake.
@@ -86,7 +86,7 @@ One work-around is to set the CMake option `CMAKE_SUPPRESS_REGENERATION=ON`.
However, this turns off CMake's ability to detect on its own when it needs to rerun.
As a result, one will need to know when to rerun CMake by hand.
## Copyright
### Copyright
Copyright (c) 2017 - 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
SPDX-License-Identifier: BSD-3-Clause
@@ -5,7 +5,7 @@ Clang as host compiler, and NVCC as device compiler.
This is NOT the same as building with
Clang as both host and device compiler ("CUDA Clang").
# Software prerequisites
## Software prerequisites
1. Clang (regularly tested with Clang 17;
occasionally tested with Clang 10 and greater)
@@ -29,9 +29,9 @@ A symptom of not installing all needed dependencies
is the following error when attempting to use clang:
`"/usr/bin/ld: cannot find -lstdc++: No such file or directory"`.
# Running CMake
## Running CMake
## Required CMake options
### Required CMake options
The Clang build requires specifying the following CMake options.
Replace `<path-to-clang++>` with the path to your `clang++` executable.
@@ -55,7 +55,7 @@ then one can set `CMAKE_CUDA_COMPILER` as follows.
* `CMAKE_CUDA_COMPILER=${PATH_TO_CUDA_TOOLKIT}/bin/nvcc`
## Copyright
### Copyright
Copyright (c) 2017 - 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
SPDX-License-Identifier: BSD-3-Clause
+10
View File
@@ -0,0 +1,10 @@
.. _cpp_build:
Build
=====
.. toctree::
:maxdepth: 1
Building on Windows with Visual Studio<building_in_windows_with_visual_studio.md>
Building with Clang as host compiler<building_with_clang_as_host_compiler.md>