Skip to content

Fix illegal memory access in overlap mode & Use more fused triton kernels for building meta data #1958

Fix illegal memory access in overlap mode & Use more fused triton kernels for building meta data

Fix illegal memory access in overlap mode & Use more fused triton kernels for building meta data #1958

Workflow file for this run

name: PR Test
on:
push:
branches: [ main ]
paths:
- "python/sglang/**"
- "test/**"
pull_request:
branches: [ main ]
paths:
- "python/sglang/**"
- "test/**"
workflow_dispatch:
concurrency:
group: pr-test-${{ github.ref }}
cancel-in-progress: true
jobs:
unit-test-frontend:
if: github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request'
runs-on: 1-gpu-runner
steps:
- name: Checkout code
uses: actions/checkout@v3
- name: Install dependencies
run: |
bash scripts/ci_install_dependency.sh
- name: Run test
timeout-minutes: 10
run: |
cd test/lang
python3 run_suite.py --suite minimal
unit-test-backend-part-1:
if: github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request'
runs-on: 1-gpu-runner
steps:
- name: Checkout code
uses: actions/checkout@v3
- name: Install dependencies
run: |
bash scripts/ci_install_dependency.sh
- name: Run test
timeout-minutes: 25
run: |
cd test/srt
python3 run_suite.py --suite minimal --range-begin 0 --range-end 5
unit-test-backend-part-2:
if: github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request'
runs-on: 1-gpu-runner
steps:
- name: Checkout code
uses: actions/checkout@v3
- name: Install dependencies
run: |
bash scripts/ci_install_dependency.sh
- name: Run test
timeout-minutes: 25
run: |
cd test/srt
python3 run_suite.py --suite minimal --range-begin 5 --range-end 14
unit-test-backend-part-3:
if: github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request'
runs-on: 1-gpu-runner
steps:
- name: Checkout code
uses: actions/checkout@v3
- name: Install dependencies
run: |
bash scripts/ci_install_dependency.sh
- name: Run test
timeout-minutes: 25
run: |
cd test/srt
python3 run_suite.py --suite minimal --range-begin 14 --range-end 21
unit-test-backend-part-4:
if: github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request'
runs-on: 1-gpu-runner
steps:
- name: Checkout code
uses: actions/checkout@v3
- name: Install dependencies
run: |
bash scripts/ci_install_dependency.sh
- name: Run test
timeout-minutes: 25
run: |
cd test/srt
python3 run_suite.py --suite minimal --range-begin 21
performance-test-1-gpu-part-1:
if: github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request'
runs-on: 1-gpu-runner
steps:
- name: Checkout code
uses: actions/checkout@v3
- name: Install dependencies
run: |
bash scripts/ci_install_dependency.sh
- name: Benchmark single latency
timeout-minutes: 10
run: |
cd test/srt
python3 -m unittest test_bench_latency.TestBenchLatency.test_default
- name: Benchmark online latency
timeout-minutes: 10
run: |
cd test/srt
python3 -m unittest test_bench_serving.TestBenchServing.test_online_latency_default
- name: Benchmark offline throughput
timeout-minutes: 10
run: |
cd test/srt
python3 -m unittest test_bench_serving.TestBenchServing.test_offline_throughput_default
- name: Benchmark offline throughput (Non-streaming, small batch size)
timeout-minutes: 10
run: |
cd test/srt
python3 -m unittest test_bench_serving.TestBenchServing.test_offline_throughput_non_stream_small_batch_size
performance-test-1-gpu-part-2:
if: github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request'
runs-on: 1-gpu-runner
steps:
- name: Checkout code
uses: actions/checkout@v3
- name: Install dependencies
run: |
bash scripts/ci_install_dependency.sh
- name: Benchmark offline throughput (w/o RadixAttention)
timeout-minutes: 10
run: |
cd test/srt
python3 -m unittest test_bench_serving.TestBenchServing.test_offline_throughput_without_radix_cache
- name: Benchmark offline throughput (w/ Triton)
timeout-minutes: 10
run: |
cd test/srt
python3 -m unittest test_bench_serving.TestBenchServing.test_offline_throughput_with_triton_attention_backend
- name: Benchmark offline throughput (w/ FP8)
timeout-minutes: 10
run: |
cd test/srt
python3 -m unittest test_bench_serving.TestBenchServing.test_offline_throughput_default_fp8
performance-test-2-gpu:
if: github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request'
runs-on: 2-gpu-runner
steps:
- name: Checkout code
uses: actions/checkout@v3
- name: Install dependencies
run: |
bash scripts/ci_install_dependency.sh
- name: Benchmark offline throughput (TP=2)
timeout-minutes: 10
run: |
cd test/srt
python3 -m unittest test_bench_serving.TestBenchServing.test_moe_offline_throughput_default
- name: Benchmark offline throughput (w/o RadixAttention) (TP=2)
timeout-minutes: 10
run: |
cd test/srt
python3 -m unittest test_bench_serving.TestBenchServing.test_moe_offline_throughput_without_radix_cache
- name: Benchmark single latency (TP=2)
timeout-minutes: 10
run: |
cd test/srt
python3 -m unittest test_bench_latency.TestBenchLatency.test_moe_default
accuracy-test-1-gpu:
if: github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request'
runs-on: 1-gpu-runner
steps:
- name: Checkout code
uses: actions/checkout@v3
- name: Install dependencies
run: |
bash scripts/ci_install_dependency.sh
git clone https://github.com/merrymercy/human-eval.git
cd human-eval
pip install -e .
- name: Evaluate accuracy
timeout-minutes: 20
run: |
cd test/srt
python3 test_eval_accuracy_large.py
accuracy-test-2-gpu:
if: github.repository == 'sgl-project/sglang' || github.event_name == 'pull_request'
runs-on: 2-gpu-runner
steps:
- name: Checkout code
uses: actions/checkout@v3
- name: Install dependencies
run: |
bash scripts/ci_install_dependency.sh
git clone https://github.com/merrymercy/human-eval.git
cd human-eval
pip install -e .
- name: Evaluate accuracy (TP=2)
timeout-minutes: 20
run: |
cd test/srt
python3 test_moe_eval_accuracy_large.py
- name: Evaluate MLA accuracy (TP=2)
timeout-minutes: 10
run: |
cd test/srt
python3 test_mla.py
python3 test_mla_fp8.py
python3 test_dp_attention.py
- name: Evaluate data parallelism accuracy (DP=2)
timeout-minutes: 10
run: |
cd test/srt
python3 test_data_parallelism.py
finish:
needs: [
unit-test-frontend, unit-test-backend-part-1, unit-test-backend-part-2, unit-test-backend-part-3, unit-test-backend-part-4,
performance-test-1-gpu-part-1, performance-test-1-gpu-part-2, performance-test-2-gpu,
accuracy-test-1-gpu, accuracy-test-2-gpu
]
runs-on: ubuntu-latest
steps:
- name: Finish
run: echo "This is an empty step to ensure that all jobs are completed."