pytorch

mirror of https://github.com/pytorch/pytorch.git synced 2025-10-20 21:14:14 +08:00

Files

Ethan Wee 04bd7e6850 [ROCm] Remove use of warpsize on host-side compilation (#156979 )

Changes needed for ROCm7.0:
* `warpSize` is _not_ a compile-time constant on device-side compilation for ROCm anymore
* `warpSize` is _not_ defined on host-side compilation, hence `at::cuda::warp_size()` must be used to query warpsize at runtime
* Redefining `C10_WARP_SIZE` to be a compile-time constant, with a reasonable value for device-side compilation, but an unreasonable value of 1 for host-side compilation

Pull Request resolved: https://github.com/pytorch/pytorch/pull/156979
Approved by: https://github.com/jeffdaily

Co-authored-by: Jeff Daily <jeff.daily@amd.com>

2025-07-01 04:55:31 +00:00

BUILD.bazel

port //c10/macros to common build structure (#70852 )

2022-01-19 20:56:12 +00:00

build.bzl

Reapply D77381084 / #156964 : Rename torch::standalone to headeronly (#157251 )

2025-06-30 23:25:30 +00:00

cmake_configure_file.bzl

use cross-platform sed command in cmake_configure_file (#73589 )

2022-03-02 12:28:51 +00:00

cmake_macros.h.in

[ROCm] Disabling Kernel Asserts for ROCm by default - fix and clean up and refactoring (#114660 )