EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

[CI] Parallelize unit tests in CI (#1219)

Browse Source
This commit is contained in:
Mingyi
2024-08-25 21:54:02 -07:00
committed by GitHub
parent 632d506d0b
commit 97589a60a2
6 changed files with 99 additions and 75 deletions
Download Patch File Download Diff File Expand all files Collapse all files

30
.github/workflows/accuracy-test.yml vendored
View File

@@ -23,21 +23,21 @@ jobs:
runs-on: 1-gpu-runner runs-on: 1-gpu-runner
steps: steps:
- name: Checkout code - name: Checkout code
uses: actions/checkout@v3 uses: actions/checkout@v3
- name: Install dependencies - name: Install dependencies
run: | run: |
pip install --upgrade pip pip install --upgrade pip
pip install -e "python[all]" pip install -e "python[all]"
pip install flashinfer -i https://flashinfer.ai/whl/cu121/torch2.4/ --force-reinstall pip install flashinfer -i https://flashinfer.ai/whl/cu121/torch2.4/ --force-reinstall
git clone https://github.com/merrymercy/human-eval.git git clone https://github.com/merrymercy/human-eval.git
cd human-eval cd human-eval
pip install -e . pip install -e .
- name: Evaluate Accuracy - name: Evaluate Accuracy
timeout-minutes: 20 timeout-minutes: 20
run: | run: |
cd test/srt cd test/srt
python3 test_eval_accuracy_large.py python3 test_eval_accuracy_large.py

44
.github/workflows/e2e-test.yml vendored
View File

@@ -23,29 +23,29 @@ jobs:
runs-on: 1-gpu-runner runs-on: 1-gpu-runner
steps: steps:
- name: Checkout code - name: Checkout code
uses: actions/checkout@v3 uses: actions/checkout@v3
- name: Install dependencies - name: Install dependencies
run: | run: |
pip install --upgrade pip pip install --upgrade pip
pip install -e "python[all]" pip install -e "python[all]"
pip install flashinfer -i https://flashinfer.ai/whl/cu121/torch2.4/ --force-reinstall pip install flashinfer -i https://flashinfer.ai/whl/cu121/torch2.4/ --force-reinstall
- name: Benchmark Serving Throughput - name: Benchmark Serving Throughput
timeout-minutes: 10 timeout-minutes: 10
run: | run: |
cd test/srt cd test/srt
python3 -m unittest test_serving_throughput.TestServingThroughput.test_default python3 -m unittest test_serving_throughput.TestServingThroughput.test_default
- name: Benchmark Serving Throughput (w/o RadixAttention) - name: Benchmark Serving Throughput (w/o RadixAttention)
timeout-minutes: 10 timeout-minutes: 10
run: | run: |
cd test/srt cd test/srt
python3 -m unittest test_serving_throughput.TestServingThroughput.test_default_without_radix_cache python3 -m unittest test_serving_throughput.TestServingThroughput.test_default_without_radix_cache
- name: Benchmark Serving Throughput (w/o ChunkedPrefill) - name: Benchmark Serving Throughput (w/o ChunkedPrefill)
timeout-minutes: 10 timeout-minutes: 10
run: | run: |
cd test/srt cd test/srt
python3 -m unittest test_serving_throughput.TestServingThroughput.test_default_without_chunked_prefill python3 -m unittest test_serving_throughput.TestServingThroughput.test_default_without_chunked_prefill

34
.github/workflows/moe-test.yml vendored
View File

@@ -23,23 +23,23 @@ jobs:
runs-on: 2-gpu-runner runs-on: 2-gpu-runner
steps: steps:
- name: Checkout code - name: Checkout code
uses: actions/checkout@v3 uses: actions/checkout@v3
- name: Install dependencies - name: Install dependencies
run: | run: |
pip install --upgrade pip pip install --upgrade pip
pip install -e "python[all]" pip install -e "python[all]"
pip install flashinfer -i https://flashinfer.ai/whl/cu121/torch2.4/ --force-reinstall pip install flashinfer -i https://flashinfer.ai/whl/cu121/torch2.4/ --force-reinstall
- name: Benchmark MoE Serving Throughput - name: Benchmark MoE Serving Throughput
timeout-minutes: 10 timeout-minutes: 10
run: | run: |
cd test/srt cd test/srt
python3 -m unittest test_moe_serving_throughput.TestServingThroughput.test_default python3 -m unittest test_moe_serving_throughput.TestServingThroughput.test_default
- name: Benchmark MoE Serving Throughput (w/o RadixAttention) - name: Benchmark MoE Serving Throughput (w/o RadixAttention)
timeout-minutes: 10 timeout-minutes: 10
run: | run: |
cd test/srt cd test/srt
python3 -m unittest test_moe_serving_throughput.TestServingThroughput.test_default_without_radix_cache python3 -m unittest test_moe_serving_throughput.TestServingThroughput.test_default_without_radix_cache

49
.github/workflows/unit-test.yml vendored
View File

@@ -18,28 +18,39 @@ concurrency:
cancel-in-progress: true cancel-in-progress: true
jobs: jobs:
unit-test: unit-test-jobs:
if: github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request' if: github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request'
runs-on: 1-gpu-runner runs-on: 1-gpu-runner
strategy:
matrix:
test_type: ['backend-0', 'backend-1', 'frontend']
steps: steps:
- name: Checkout code - name: Checkout code
uses: actions/checkout@v3 uses: actions/checkout@v3
- name: Install dependencies - name: Install dependencies
run: | run: |
pip install --upgrade pip pip install --upgrade pip
pip install -e "python[dev]" pip install -e "python[dev]"
pip install flashinfer -i https://flashinfer.ai/whl/cu121/torch2.4/ --force-reinstall pip install flashinfer -i https://flashinfer.ai/whl/cu121/torch2.4/ --force-reinstall
- name: Test Backend Runtime - name: Run test
timeout-minutes: 20 timeout-minutes: 20
run: | run: |
cd test/srt if [ "${{ matrix.test_type }}" = "frontend" ]; then
python3 run_suite.py --suite minimal cd test/lang
python3 run_suite.py --suite minimal
elif [ "${{ matrix.test_type }}" = "backend-0" ]; then
cd test/srt
python3 run_suite.py --suite minimal --range-begin 0 --range-end 8
elif [ "${{ matrix.test_type }}" = "backend-1" ]; then
cd test/srt
python3 run_suite.py --suite minimal --range-begin 8
fi
- name: Test Frontend Language unit-test:
timeout-minutes: 10 needs: unit-test-jobs
run: | runs-on: ubuntu-latest
cd test/lang steps:
python3 run_suite.py --suite minimal - name: Merge step
run: echo "This is an empty merge step"

2
python/sglang/test/test_utils.py
View File

@@ -465,7 +465,7 @@ def run_unittest_files(files: List[str], timeout_per_file: float):
def run_one_file(filename): def run_one_file(filename):
filename = os.path.join(os.getcwd(), filename) filename = os.path.join(os.getcwd(), filename)
print(f"\n\nRun {filename}\n\n", flush=True) print(f"\n\nRun:\npython3 {filename}\n\n", flush=True)
process = subprocess.Popen( process = subprocess.Popen(
["python3", filename], stdout=None, stderr=None, env=os.environ ["python3", filename], stdout=None, stderr=None, env=os.environ
) )

15
test/srt/run_suite.py
View File

@@ -1,6 +1,5 @@
import argparse import argparse
import glob import glob
import multiprocessing as mp
from sglang.test.test_utils import run_unittest_files from sglang.test.test_utils import run_unittest_files
@@ -49,6 +48,18 @@ if __name__ == "__main__":
choices=list(suites.keys()) + ["all"], choices=list(suites.keys()) + ["all"],
help="The suite to run", help="The suite to run",
) )
arg_parser.add_argument(
"--range-begin",
type=int,
default=0,
help="The begin index of the range of the files to run.",
)
arg_parser.add_argument(
"--range-end",
type=int,
default=None,
help="The end index of the range of the files to run.",
)
args = arg_parser.parse_args() args = arg_parser.parse_args()
if args.suite == "all": if args.suite == "all":
@@ -56,5 +67,7 @@ if __name__ == "__main__":
else: else:
files = suites[args.suite] files = suites[args.suite]
files = files[args.range_begin : args.range_end]
exit_code = run_unittest_files(files, args.timeout_per_file) exit_code = run_unittest_files(files, args.timeout_per_file)
exit(exit_code) exit(exit_code)