Skip to content

Commit 3bc65f8

Browse files
juliagmt-googlejuliagmt-google
authored
Update benchmarks.yml
1 parent 7f0a09f commit 3bc65f8

File tree

1 file changed

+98
-97
lines changed

1 file changed

+98
-97
lines changed

‎.github/workflows/benchmarks.yml‎

Lines changed: 98 additions & 97 deletions
Original file line numberDiff line numberDiff line change
@@ -15,42 +15,42 @@ on:
1515
- 'yes'
1616
- 'no'
1717

18-
jobs:
19-
Tests:
20-
strategy:
21-
# Don't fail fast - want to see results for all builds even if one fails.
22-
fail-fast: false
23-
matrix:
24-
job_info: [
25-
{
26-
os: "linux-x86-n2-16",
27-
container: "us-central1-docker.pkg.dev/tensorflow-sigs/tensorflow/ml-build:latest",
28-
pretty_name: "XLA Linux x86 CPU",
29-
},
30-
{
31-
os: "linux-arm64-c4a-16",
32-
container: "us-central1-docker.pkg.dev/tensorflow-sigs/tensorflow/ml-build-arm64:latest",
33-
pretty_name: "XLA Linux ARM64 CPU",
34-
}
35-
]
36-
name: ${{ matrix.job_info.pretty_name }}
37-
runs-on: ${{ matrix.job_info.os }}
38-
container: ${{ matrix.job_info.container }}
39-
defaults:
40-
run:
41-
shell: bash
42-
timeout-minutes: 30
43-
steps:
44-
- name: Print machine specs
45-
run: |
46-
lscpu
47-
free -h # Memory information
48-
df -h # Disk space information
49-
uname -a # Kernel information
50-
- name: Wait For Connection
51-
uses: google-ml-infra/actions/ci_connection@main
52-
with:
53-
halt-dispatch-input: ${{ inputs.halt-for-connection }}
18+
#jobs:
19+
# Tests:
20+
# strategy:
21+
# # Don't fail fast - want to see results for all builds even if one fails.
22+
# fail-fast: false
23+
# matrix:
24+
# job_info: [
25+
# {
26+
# os: "linux-x86-n2-16",
27+
# container: "us-central1-docker.pkg.dev/tensorflow-sigs/tensorflow/ml-build:latest",
28+
# pretty_name: "XLA Linux x86 CPU",
29+
# },
30+
# {
31+
# os: "linux-arm64-c4a-16",
32+
# container: "us-central1-docker.pkg.dev/tensorflow-sigs/tensorflow/ml-build-arm64:latest",
33+
# pretty_name: "XLA Linux ARM64 CPU",
34+
# }
35+
# ]
36+
# name: ${{ matrix.job_info.pretty_name }}
37+
# runs-on: ${{ matrix.job_info.os }}
38+
# container: ${{ matrix.job_info.container }}
39+
# defaults:
40+
# run:
41+
# shell: bash
42+
# timeout-minutes: 30
43+
# steps:
44+
# - name: Print machine specs
45+
# run: |
46+
# lscpu
47+
# free -h # Memory information
48+
# df -h # Disk space information
49+
# uname -a # Kernel information
50+
# - name: Wait For Connection
51+
# uses: google-ml-infra/actions/ci_connection@main
52+
# with:
53+
# halt-dispatch-input: ${{ inputs.halt-for-connection }}
5454

5555
# - name: Check Python Version in Container
5656
# run: python3 --version
@@ -65,15 +65,15 @@ jobs:
6565
# echo "Python 3 found."
6666
# fi
6767

68-
- name: Checkout OpenXLA
69-
uses: actions/checkout@v3
70-
with:
71-
repository: openxla/xla
68+
# - name: Checkout OpenXLA
69+
# uses: actions/checkout@v3
70+
# with:
71+
# repository: openxla/xla
7272

73-
- name: Create results directory
74-
working-directory: xla
75-
run:
76-
mkdir results
73+
# - name: Create results directory
74+
# working-directory: xla
75+
# run:
76+
# mkdir results
7777

7878
# - name: Run setup.sh for E2E benchmarks flax_2b
7979
# working-directory: xla/backends/cpu/benchmarks/e2e/gemma2/flax_2b
@@ -107,66 +107,67 @@ jobs:
107107
# ./bazel-bin/xla/tools/run_hlo_module --input_format=hlo --platform=CPU "$file"
108108
# done
109109

110-
- name: Build CPU Benchmarks
111-
working-directory: xla
112-
run: bazel build -c opt --dynamic_mode=off //xla/backends/cpu/benchmarks:*
110+
# - name: Build CPU Benchmarks
111+
# working-directory: xla
112+
# run: bazel build -c opt --dynamic_mode=off //xla/backends/cpu/benchmarks:*
113113

114-
- name: Run CPU benchmarks
115-
working-directory: xla
116-
continue-on-error: true
117-
run: |
118-
find ./bazel-bin/xla/backends/cpu/benchmarks/ -maxdepth 1 -type f -executable -name "*_test" -print0 | while IFS= read -r -d $'\0' benchmark; do
119-
benchmark_name=$(basename "$benchmark" | sed 's/_test$//')
120-
echo "Running benchmark: $benchmark_name"
114+
# - name: Run CPU benchmarks
115+
# working-directory: xla
116+
# continue-on-error: true
117+
# run: |
118+
# find ./bazel-bin/xla/backends/cpu/benchmarks/ -maxdepth 1 -type f -executable -name "*_test" -print0 | while IFS= read -r -d $'\0' benchmark; do
119+
# benchmark_name=$(basename "$benchmark" | sed 's/_test$//')
120+
# echo "Running benchmark: $benchmark_name"
121121

122-
# Run the benchmark with default parameters.
123-
$benchmark --benchmark_filter=".*"
124-
$benchmark --benchmark_filter=".*" > "results/$benchmark_name.log" 2>&1
122+
# # Run the benchmark with default parameters.
123+
# $benchmark --benchmark_filter=".*"
124+
# $benchmark --benchmark_filter=".*" > "results/$benchmark_name.log" 2>&1
125125

126-
# Check the exit code of the benchmark
127-
if [ $? -ne 0 ]; then
128-
echo "Error: Benchmark '$benchmark_name' failed. Check the log file: results/$benchmark_name.log"
129-
else
130-
echo "Benchmark '$benchmark_name' completed successfully."
131-
fi
132-
done
133-
- name: Upload Results ${{ matrix.build_target }}
134-
uses: actions/upload-artifact@v4
135-
with:
136-
name: cpu-xla-benchmarks-${{ matrix.job_info.os }}
137-
path: xla/results
138-
# build-xla-gpu-and-test:
139-
# runs-on: "linux-x86-n2-16" #linux-x86-g2-48-l4-4gpu # Use a GPU-enabled runner
140-
# container:
141-
# image: "gcr.io/tensorflow-testing/nosla-cuda12.3-cudnn9.1-ubuntu20.04-manylinux2014-multipython:latest"
142-
# options: --gpus all --privileged # Might need privileged mode, use with caution
126+
# # Check the exit code of the benchmark
127+
# if [ $? -ne 0 ]; then
128+
# echo "Error: Benchmark '$benchmark_name' failed. Check the log file: results/$benchmark_name.log"
129+
# else
130+
# echo "Benchmark '$benchmark_name' completed successfully."
131+
# fi
132+
# done
133+
# - name: Upload Results ${{ matrix.build_target }}
134+
# uses: actions/upload-artifact@v4
135+
# with:
136+
# name: cpu-xla-benchmarks-${{ matrix.job_info.os }}
137+
# path: xla/results
138+
jobs:
139+
build-xla-gpu-and-test:
140+
runs-on: "linux-x86-n2-16" #linux-x86-g2-48-l4-4gpu # Use a GPU-enabled runner
141+
container:
142+
image: "gcr.io/tensorflow-testing/nosla-cuda12.3-cudnn9.1-ubuntu20.04-manylinux2014-multipython:latest"
143+
options: --gpus all --privileged # Might need privileged mode, use with caution
143144

144-
# steps:
145-
# - name: Checkout XLA
146-
# uses: actions/checkout@v3
147-
# with:
148-
# repository: openxla/xla # Replace with your fork if needed
149-
# - name: Checkout repository
150-
# uses: actions/checkout@v3
151-
# with:
152-
# repository: juliagmt-google/xla
153-
# ref: dev
145+
steps:
146+
- name: Checkout XLA
147+
uses: actions/checkout@v3
148+
with:
149+
repository: openxla/xla # Replace with your fork if needed
150+
- name: Checkout repository
151+
uses: actions/checkout@v3
152+
with:
153+
repository: juliagmt-google/xla
154+
ref: dev
154155

155-
# - name: Print machine specs
156-
# run: |
157-
# lscpu
158-
# free -h # Memory information
159-
# df -h # Disk space information
160-
# uname -a # Kernel information
156+
- name: Print machine specs
157+
run: |
158+
lscpu
159+
free -h # Memory information
160+
df -h # Disk space information
161+
uname -a # Kernel information
161162
162-
# - name: Create results directory
163-
# working-directory: xla
164-
# run: mkdir results
163+
- name: Create results directory
164+
working-directory: xla
165+
run: mkdir results
165166

166-
# - name: Wait For Connection
167-
# uses: google-ml-infra/actions/ci_connection@main
168-
# with:
169-
# halt-dispatch-input: ${{ inputs.halt-for-connection }}
167+
- name: Wait For Connection
168+
uses: google-ml-infra/actions/ci_connection@main
169+
with:
170+
halt-dispatch-input: ${{ inputs.halt-for-connection }}
170171
# - name: Set up Python 3.10 # Choose your desired Python version
171172
# uses: actions/setup-python@v4
172173
# with:

0 commit comments

Comments
 (0)