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

[CI] Update B200 est_times to prevent timeouts on slower machine (#22609)

Browse Source 
Co-authored-by: Alison Shao <alison.shao@MacBook-Pro-D2W773R9CD.local>
This commit is contained in:
Alison Shao
2026-04-11 21:40:34 -07:00
committed by GitHub
co-authored by Alison Shao
parent f21d23a211
commit d6c9d9116b
9 changed files with 9 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files
test/registered/4-gpu-models/test_gpt_oss_4gpu.py
+1 -1
View File
@@ -4,7 +4,7 @@ from sglang.test.ci.ci_register import register_cuda_ci
from sglang.test.gpt_oss_common import BaseTestGptOss from sglang.test.gpt_oss_common import BaseTestGptOss
register_cuda_ci(est_time=328, suite="stage-c-test-4-gpu-h100") register_cuda_ci(est_time=328, suite="stage-c-test-4-gpu-h100")
register_cuda_ci(est_time=312, suite="stage-c-test-4-gpu-b200") register_cuda_ci(est_time=740, suite="stage-c-test-4-gpu-b200")
class TestGptOss4Gpu(BaseTestGptOss): class TestGptOss4Gpu(BaseTestGptOss):
test/registered/4-gpu-models/test_nvidia_nemotron_3_super_nvfp4.py
+1 -1
View File
@@ -11,7 +11,7 @@ from sglang.test.test_utils import (
popen_launch_server, popen_launch_server,
) )
register_cuda_ci(est_time=294, suite="stage-c-test-4-gpu-b200") register_cuda_ci(est_time=710, suite="stage-c-test-4-gpu-b200")
NEMOTRON_3_SUPER_NVFP4_MODEL = "nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4" NEMOTRON_3_SUPER_NVFP4_MODEL = "nvidia/NVIDIA-Nemotron-3-Super-120B-A12B-NVFP4"
test/registered/attention/test_flash_attention_4.py
+1 -1
View File
@@ -13,7 +13,7 @@ from sglang.test.test_utils import (
) )
# FlashAttention4 integration test (requires SM 100+ / Blackwell B200) # FlashAttention4 integration test (requires SM 100+ / Blackwell B200)
register_cuda_ci(est_time=259, suite="stage-b-test-4-gpu-b200") register_cuda_ci(est_time=332, suite="stage-b-test-4-gpu-b200")
@unittest.skipIf(get_device_sm() < 100, "Test requires CUDA SM 100 or higher") @unittest.skipIf(get_device_sm() < 100, "Test requires CUDA SM 100 or higher")
test/registered/moe/test_cutedsl_moe.py
+1 -1
View File
@@ -16,7 +16,7 @@ except ImportError:
CuteDslMoEWrapper = None CuteDslMoEWrapper = None
convert_sf_to_mma_layout = None convert_sf_to_mma_layout = None
register_cuda_ci(est_time=13, suite="stage-c-test-4-gpu-b200") register_cuda_ci(est_time=590, suite="stage-c-test-4-gpu-b200")
SKIP_TEST = torch.cuda.get_device_capability() < (10, 0) SKIP_TEST = torch.cuda.get_device_capability() < (10, 0)
SKIP_REASON = "Nvfp4 Requires compute capability of 10 or above." SKIP_REASON = "Nvfp4 Requires compute capability of 10 or above."
test/registered/quant/test_deepseek_v3_fp4_4gpu.py
+1 -1
View File
@@ -14,7 +14,7 @@ from sglang.test.test_utils import (
write_github_step_summary, write_github_step_summary,
) )
register_cuda_ci(est_time=1146, suite="stage-c-test-4-gpu-b200") register_cuda_ci(est_time=1380, suite="stage-c-test-4-gpu-b200")
FULL_DEEPSEEK_V3_FP4_MODEL_PATH = "nvidia/DeepSeek-V3-0324-FP4" FULL_DEEPSEEK_V3_FP4_MODEL_PATH = "nvidia/DeepSeek-V3-0324-FP4"
SERVER_LAUNCH_TIMEOUT = 1200 SERVER_LAUNCH_TIMEOUT = 1200
test/registered/quant/test_fp8_blockwise_gemm.py
+1 -1
View File
@@ -12,7 +12,7 @@ from sglang.test.test_utils import (
try_cached_model, try_cached_model,
) )
register_cuda_ci(est_time=302, suite="stage-c-test-4-gpu-b200") register_cuda_ci(est_time=630, suite="stage-c-test-4-gpu-b200")
MODEL_PATH = "Qwen/Qwen3-4B-Instruct-2507-FP8" MODEL_PATH = "Qwen/Qwen3-4B-Instruct-2507-FP8"
MXFP8_MODEL_PATH = "zianglih/Qwen3-4B-Instruct-2507-MXFP8" MXFP8_MODEL_PATH = "zianglih/Qwen3-4B-Instruct-2507-MXFP8"
test/registered/quant/test_nvfp4_gemm.py
+1 -1
View File
@@ -12,7 +12,7 @@ from sglang.test.test_utils import (
try_cached_model, try_cached_model,
) )
register_cuda_ci(est_time=322, suite="stage-c-test-4-gpu-b200") register_cuda_ci(est_time=550, suite="stage-c-test-4-gpu-b200")
MODEL_PATH = "nvidia/Llama-3.1-8B-Instruct-NVFP4" MODEL_PATH = "nvidia/Llama-3.1-8B-Instruct-NVFP4"
test/registered/spec/eagle/test_deepseek_v3_fp4_mtp_small.py
+1 -1
View File
@@ -16,7 +16,7 @@ from sglang.test.test_utils import (
write_github_step_summary, write_github_step_summary,
) )
register_cuda_ci(est_time=416, suite="stage-b-test-4-gpu-b200") register_cuda_ci(est_time=510, suite="stage-b-test-4-gpu-b200")
FULL_DEEPSEEK_V3_FP4_MODEL_PATH = "nvidia/DeepSeek-V3-0324-FP4" FULL_DEEPSEEK_V3_FP4_MODEL_PATH = "nvidia/DeepSeek-V3-0324-FP4"
SERVER_LAUNCH_TIMEOUT = 1200 SERVER_LAUNCH_TIMEOUT = 1200
test/registered/spec/eagle/test_eagle_infer_beta_dp_attention.py
+1 -1
View File
@@ -17,7 +17,7 @@ from sglang.test.test_utils import (
) )
# EAGLE with DP attention on B200 (tp=2, dp=2, requires 4 B200 GPUs) # EAGLE with DP attention on B200 (tp=2, dp=2, requires 4 B200 GPUs)
register_cuda_ci(est_time=68, suite="stage-c-test-4-gpu-b200") register_cuda_ci(est_time=136, suite="stage-c-test-4-gpu-b200")
def test_gsm8k(base_url: str, model: str): def test_gsm8k(base_url: str, model: str):