-
Notifications
You must be signed in to change notification settings - Fork 243
157 lines (141 loc) · 6.19 KB
/
ci.gpu.yml
File metadata and controls
157 lines (141 loc) · 6.19 KB
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
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
name: CI (GPU)
on:
push:
branches:
- main
- "member-function-customization"
- "pull-request/[0-9]+"
concurrency:
group: ${{ github.workflow }}-on-${{ github.event_name }}-from-${{ github.ref_name }}
cancel-in-progress: true
jobs:
build-gpu:
if: github.repository == 'NVIDIA/stdexec'
name: GPU (${{ matrix.name }}, CUDA ${{ matrix.cuda }}, ${{ matrix.build }})
strategy:
fail-fast: false
matrix:
include:
- { name: "clang 20", cuda: "12.0", cxx: "clang++", build: "Release", tag: "llvm20-cuda12.0", gpu: "v100", sm: "70", driver: "latest", arch: "amd64" }
- { name: "clang 20", cuda: "12.9", cxx: "clang++", build: "Release", tag: "llvm20-cuda12.9", gpu: "v100", sm: "70", driver: "latest", arch: "amd64" }
- { name: "nvc++ 25.7", cuda: "12.9", cxx: "mpic++", build: "Release", tag: "nvhpc25.7", gpu: "v100", sm: "70", driver: "latest", arch: "amd64" }
- { name: "nvc++ 25.7", cuda: "12.9", cxx: "mpic++", build: "Debug", tag: "nvhpc25.7", gpu: "v100", sm: "70", driver: "latest", arch: "amd64" }
runs-on: linux-${{ matrix.arch }}-gpu-${{ matrix.gpu }}-${{ matrix.driver }}-1
container:
options: -u root
image: rapidsai/devcontainers:25.10-cpp-${{ matrix.tag }}
env:
DEVCONTAINER_UTILS_ENABLE_SCCACHE_DIST: "true"
NVIDIA_VISIBLE_DEVICES: "${{ env.NVIDIA_VISIBLE_DEVICES }}"
SCCACHE_BUCKET: "rapids-sccache-devs"
SCCACHE_DIST_CONNECT_TIMEOUT: "30"
SCCACHE_DIST_FALLBACK_TO_LOCAL_COMPILE: "true"
SCCACHE_DIST_KEEPALIVE_ENABLED: "true"
SCCACHE_DIST_KEEPALIVE_INTERVAL: "20"
SCCACHE_DIST_KEEPALIVE_TIMEOUT: "600"
SCCACHE_DIST_MAX_RETRIES: "4"
SCCACHE_DIST_REQUEST_TIMEOUT: "7140"
SCCACHE_DIST_TOKEN: "${{ secrets.STDEXEC_BUILD_CLUSTER_SECRET }}"
SCCACHE_DIST_URL: "https://${{ matrix.arch }}.linux.sccache.rapids.nvidia.com"
SCCACHE_ERROR_LOG: "/home/coder/stdexec/.cache/sccache.log"
SCCACHE_REGION: "us-east-2"
SCCACHE_S3_KEY_PREFIX: "nvidia-stdexec-dev"
SCCACHE_SERVER_PORT: "4225"
SCCACHE_SERVER_LOG: "sccache=debug"
permissions:
id-token: write # This is required for configure-aws-credentials
contents: read # This is required for actions/checkout
defaults:
run:
shell: su coder {0}
working-directory: /home/coder
steps:
- name: Run nvidia-smi to make sure GPU is working
run: nvidia-smi
- name: Checkout stdexec
uses: actions/checkout@v4
with:
path: stdexec
persist-credentials: false
- name: Setup environment
run: |
echo "ARTIFACT_PREFIX=${{runner.os}}-cuda${{matrix.cuda}}-${{matrix.tag}}-${{matrix.arch}}" >> "${GITHUB_ENV}"
echo "ARTIFACT_SUFFIX=${{github.run_id}}-${{github.run_attempt}}-$RANDOM" >> "${GITHUB_ENV}"
- id: sccache-preprocessor-cache
name: Setup sccache preprocessor cache
uses: actions/cache@v4
with:
path: /home/coder/.cache/sccache/preprocessor
restore-keys: sccache-preprocessor-cache-${{env.ARTIFACT_PREFIX}}
key: sccache-preprocessor-cache-${{env.ARTIFACT_PREFIX}}-${{env.ARTIFACT_SUFFIX}}
- id: sccache-dist-toolchains-cache
name: Setup sccache-dist client toolchains cache
uses: actions/cache@v4
with:
path: /home/coder/.cache/sccache-dist-client
restore-keys: sccache-toolchains-cache-${{env.ARTIFACT_PREFIX}}
key: sccache-toolchains-cache-${{env.ARTIFACT_PREFIX}}-${{env.ARTIFACT_SUFFIX}}
- if: github.repository_owner == 'NVIDIA'
name: Get AWS credentials for sccache bucket
uses: aws-actions/configure-aws-credentials@v4
with:
aws-region: us-east-2
role-duration-seconds: 28800 # 8 hours
role-to-assume: arn:aws:iam::279114543810:role/gha-oidc-NVIDIA
- name: Build and test GPU schedulers
run: |
set -e;
if test -n "${NVHPC_ROOT:+x}"; then
HPCX_INIT="$(find -L "$NVHPC_ROOT"/comm_libs/ -path '*/latest/hpcx-init.sh')";
if test -n "${HPCX_INIT:+x}"; then
. "$HPCX_INIT";
hpcx_load;
fi
fi
set -x;
mkdir -p /home/coder/stdexec/.cache
devcontainer-utils-install-sccache --repo rapidsai/sccache --version rapids;
devcontainer-utils-init-sccache-dist \
--enable-sccache-dist - <<< " \
--auth-type 'token' \
--auth-token '${SCCACHE_DIST_TOKEN}' \
";
sccache --dist-status
cat /home/coder/stdexec/.cache/*.log
# Copy source folder into ~/stdexec
cp -r "${GITHUB_WORKSPACE}"/stdexec ~/;
chown -R coder:coder ~/stdexec;
cd ~/stdexec;
# Configure
cmake -S . -B build -GNinja \
-DSTDEXEC_ENABLE_CUDA=ON \
-DSTDEXEC_ENABLE_IO_URING=OFF \
-DCMAKE_BUILD_TYPE=${{ matrix.build }} \
-DCMAKE_CXX_COMPILER=${{ matrix.cxx }} \
-DCMAKE_CUDA_COMPILER=${{ matrix.cxx }} \
-DCMAKE_CUDA_ARCHITECTURES=${{ matrix.sm }} \
;
# Compile
cmake --build build -v;
# Print sccache stats
sccache -s;
# Tests
ctest --test-dir build --verbose --output-on-failure --timeout 60;
# Examples
./build/examples/nvexec/maxwell_cpu_st --iterations=1000 --N=512 --run-cpp --run-inline-scheduler;
./build/examples/nvexec/maxwell_cpu_mt --iterations=1000 --N=512 --run-std --run-stdpar --run-thread-pool-scheduler;
./build/examples/nvexec/maxwell_gpu_s --iterations=1000 --N=512 --run-cuda --run-stdpar --run-stream-scheduler;
- if: ${{ !cancelled() }}
name: Upload sccache logs
uses: actions/upload-artifact@v4
with:
name: sccache-client-logs-${{env.ARTIFACT_PREFIX}}-${{env.ARTIFACT_SUFFIX}}
path: /home/coder/stdexec/.cache/*.log
compression-level: 9
ci-gpu:
runs-on: ubuntu-latest
name: CI (GPU)
needs:
- build-gpu
steps:
- run: echo "CI (GPU) success"