minke.yu/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 12 Packages Projects Releases Wiki Activity

ci: run Stage A CUDA tests as stage-a-test-small-1-gpu on 5090 (#20988)

Browse Source
This commit is contained in:
Lianmin Zheng
2026-03-20 02:55:16 -07:00
committed by GitHub
parent fa89d152c0
commit c1da420799
12 changed files with 78 additions and 100 deletions
Download Patch File Download Diff File Expand all files Collapse all files
test/README.md
+5 -3
View File
@@ -60,7 +60,7 @@ register_cuda_ci(est_time=200, suite="nightly-1-gpu", nightly=True)
# Multi-backend test
register_cuda_ci(est_time=80, suite="stage-b-test-small-1-gpu")
register_amd_ci(est_time=120, suite="stage-a-test-1")
register_amd_ci(est_time=120, suite="stage-a-test-small-1-gpu-amd")
# Temporarily disabled test
register_cuda_ci(est_time=80, suite="stage-b-test-small-1-gpu", disabled="flaky - see #12345")
@@ -72,6 +72,8 @@ When adding 1-GPU tests, choose the appropriate suite based on hardware compatib
| Suite | Runner | GPU | When to Use |
|-------|--------|-----|-------------|
| `stage-a-test-small-1-gpu` | `1-gpu-5090` | RTX 5090 (32GB, SM120) | Stage A per-commit smoke on 5090 (CUDA) |
| `stage-a-test-small-1-gpu-amd` | AMD CI runners | ROCm | Stage A per-commit smoke (AMD) |
| `stage-b-test-small-1-gpu` | `1-gpu-5090` | RTX 5090 (32GB, SM120) | 5090-compatible tests (preferred) |
| `stage-b-test-large-1-gpu` | `1-gpu-runner` | H100 (80GB, SM90) | Large models or 5090-incompatible tests |
@@ -98,13 +100,13 @@ If a test cannot run on 5090 due to any of the above, use `stage-b-test-large-1-
### Available Suites
**Per-Commit (CUDA)**:
- Stage A: `stage-a-test-1` (locked), `stage-a-test-2`, `stage-a-test-cpu`
- Stage A: `stage-a-test-small-1-gpu` (5090), `stage-a-test-2`, `stage-a-test-cpu`
- Stage B: `stage-b-test-small-1-gpu` (5090), `stage-b-test-large-1-gpu` (H100), `stage-b-test-large-2-gpu`
- Stage C (4-GPU): `stage-c-test-4-gpu-h100`, `stage-c-test-4-gpu-b200`, `stage-c-test-4-gpu-gb200`, `stage-c-test-deepep-4-gpu`
- Stage C (8-GPU): `stage-c-test-8-gpu-h20`, `stage-c-test-8-gpu-h200`, `stage-c-test-8-gpu-b200`, `stage-c-test-deepep-8-gpu-h200`
**Per-Commit (AMD)**:
- `stage-a-test-1`, `stage-b-test-small-1-gpu-amd`, `stage-b-test-large-2-gpu-amd`
- `stage-a-test-small-1-gpu-amd`, `stage-b-test-small-1-gpu-amd`, `stage-b-test-large-2-gpu-amd`
**Nightly**:
- `nightly-1-gpu`, `nightly-2-gpu`, `nightly-4-gpu`, `nightly-8-gpu`, etc.
test/registered/attention/test_wave_attention_kernels.py
+1 -1
View File
@@ -25,7 +25,7 @@ from sglang.srt.utils import get_device
from sglang.test.ci.ci_register import register_amd_ci
# Wave attention kernel unit tests (AMD only - requires wave_lang)
register_amd_ci(est_time=60, suite="stage-a-test-1-amd")
register_amd_ci(est_time=60, suite="stage-a-test-small-1-gpu-amd")
class TestWaveAttention(unittest.TestCase):
test/registered/debug_utils/test_cuda_coredump_smoke.py
+1 -1
View File
@@ -12,7 +12,7 @@ from sglang.test.ci.ci_register import register_cuda_ci
register_cuda_ci(
est_time=10,
suite="stage-a-test-1",
suite="stage-a-test-small-1-gpu",
disabled="Manual only: triggers intentional CUDA crash for coredump verification",
)
test/registered/quant/test_awq_dequant.py
+1 -1
View File
@@ -21,7 +21,7 @@ from sglang.srt.utils import get_device
from sglang.test.ci.ci_register import register_amd_ci
from sglang.test.test_utils import CustomTestCase
register_amd_ci(est_time=2, suite="stage-a-test-1-amd")
register_amd_ci(est_time=2, suite="stage-a-test-small-1-gpu-amd")
device = get_device()
test/registered/quant/test_fused_rms_fp8_group_quant.py
+1 -1
View File
@@ -7,7 +7,7 @@ import torch.nn.functional as F
from sglang.test.ci.ci_register import register_amd_ci
from sglang.test.test_utils import CustomTestCase
register_amd_ci(est_time=10, suite="stage-a-test-1-amd")
register_amd_ci(est_time=10, suite="stage-a-test-small-1-gpu-amd")
def _fp8_available() -> bool:
test/registered/test_srt_backend.py
+5 -3
View File
@@ -19,8 +19,8 @@ from sglang.test.test_programs import (
)
from sglang.test.test_utils import DEFAULT_MODEL_NAME_FOR_TEST, CustomTestCase
register_cuda_ci(est_time=80, suite="stage-a-test-1")
register_amd_ci(est_time=120, suite="stage-a-test-1-amd")
register_cuda_ci(est_time=80, suite="stage-a-test-small-1-gpu")
register_amd_ci(est_time=120, suite="stage-a-test-small-1-gpu-amd")
class TestSRTBackend(CustomTestCase):
@@ -29,7 +29,9 @@ class TestSRTBackend(CustomTestCase):
@classmethod
def setUpClass(cls):
cls.backend = sgl.Runtime(
model_path=DEFAULT_MODEL_NAME_FOR_TEST, cuda_graph_max_bs=4
model_path=DEFAULT_MODEL_NAME_FOR_TEST,
cuda_graph_max_bs=4,
mem_fraction_static=0.7,
)
sgl.set_default_backend(cls.backend)
test/run_suite.py
+3 -3
View File
@@ -24,7 +24,7 @@ HW_MAPPING = {
PER_COMMIT_SUITES = {
HWBackend.CPU: ["stage-a-cpu-only"],
HWBackend.AMD: [
"stage-a-test-1-amd",
"stage-a-test-small-1-gpu-amd",
"stage-b-test-small-1-gpu-amd",
"stage-b-test-small-1-gpu-amd-nondeterministic",
"stage-b-test-small-1-gpu-amd-mi35x",
@@ -35,7 +35,7 @@ PER_COMMIT_SUITES = {
"stage-c-test-large-8-gpu-amd-mi35x",
],
HWBackend.CUDA: [
"stage-a-test-1",
"stage-a-test-small-1-gpu",
"stage-b-test-small-1-gpu",
"stage-b-test-large-1-gpu",
"stage-b-test-large-2-gpu",
@@ -49,7 +49,7 @@ PER_COMMIT_SUITES = {
"stage-c-test-deepep-8-gpu-h200",
],
HWBackend.NPU: [
"stage-a-test-1",
"stage-a-test-small-1-gpu",
"stage-b-test-1-npu-a2",
"stage-b-test-2-npu-a2",
"stage-b-test-4-npu-a3",