blob: 3e4ddfa9175a9a13f0fa682d3b16c7cf298eff72 (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
|
startup --output_user_root=${srcdir}/base
build --repository_cache=${srcdir}/repo
build --disk_cache=${srcdir}/cache
build --sandbox_base=${srcdir}/sandbox
build --repo_env=HERMETIC_PYTHON_VERSION=3.14
build --repo_env=HERMETIC_CUDA_VERSION=13.1.1
build --repo_env=HERMETIC_CUDNN_VERSION=9.18.1.3
#
# TODO(@daskol): This variables controls `@rules_ml_toolchain` in theory.
# build:cuda --repo_env=LOCAL_CUDA_PATH="/opt/cuda/targets/x86_64-linux"
# build:cuda --repo_env=LOCAL_CUDNN_PATH="/usr/include"
# build:cuda --repo_env=LOCAL_NCCL_PATH="/usr/include"
# build:cuda --repo_env=LOCAL_NVSHMEM_PATH="/usr/include"
#
# NOTE Adjust HERMETIC_CUDA_COMPUTE_CAPABILITIES for actual devices.
# https://github.com/google-ml-infra/rules_ml_toolchain/tree/main/gpu
build:cuda --repo_env=HERMETIC_CUDA_COMPUTE_CAPABILITIES="sm_80,sm_86,sm_87,sm_90,sm_100,sm_103"
build --repo_env=BAZEL_COMPILER="/usr/lib/llvm20/bin/clang-20"
build --repo_env=CC="/usr/lib/llvm20/bin/clang-20"
build --repo_env=CXX="/usr/lib/llvm20/bin/clang++"
build --action_env=CLANG_COMPILER_PATH="/usr/lib/llvm20/bin/clang-20"
build --action_env=CLANG_CUDA_COMPILER_PATH="/usr/lib/llvm20/bin/clang-20"
build --action_env=CUDAHOSTCXX="/usr/lib/llvm20/bin/clang++"
build --config=avx_posix
build --config=build_cuda_with_nvcc
build --config=clang
build --config=cuda13
build --config=cuda_clang_local
build --config=cuda_libraries_from_stubs
build --config=mkl_open_source_only
build --define=xnn_enable_avxvnniint8=false
build --jobs=12
build --local_resources=cpu=12
build --verbose_failures=true
|