chenchenghao/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/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)