Update est_time for stage-b-test-small-1-gpu tests (#16835)

2026-01-10 20:03:43 -08:00
parent cf14feba4d
commit cc25f9df50
24 changed files with 24 additions and 24 deletions
--- a/test/registered/attention/test_torch_native_attention_backend.py
+++ b/test/registered/attention/test_torch_native_attention_backend.py
@@ -18,7 +18,7 @@ from sglang.test.test_utils import (
 )

 # Torch native attention backend integration test with MMLU eval
-register_cuda_ci(est_time=150, suite="stage-b-test-small-1-gpu")
+register_cuda_ci(est_time=169, suite="stage-b-test-small-1-gpu")
 register_amd_ci(est_time=150, suite="stage-b-test-small-1-gpu-amd")


--- a/test/registered/backends/test_torch_compile.py
+++ b/test/registered/backends/test_torch_compile.py
@@ -16,7 +16,7 @@ from sglang.test.test_utils import (
    popen_launch_server,
 )

-register_cuda_ci(est_time=190, suite="stage-b-test-small-1-gpu")
+register_cuda_ci(est_time=144, suite="stage-b-test-small-1-gpu")
 register_amd_ci(est_time=1100, suite="stage-b-test-small-1-gpu-amd")


--- a/test/registered/core/test_deterministic.py
+++ b/test/registered/core/test_deterministic.py
@@ -15,7 +15,7 @@ from sglang.test.test_deterministic_utils import (
    TestDeterministicBase,
 )

-register_cuda_ci(est_time=228, suite="stage-b-test-small-1-gpu")
+register_cuda_ci(est_time=278, suite="stage-b-test-small-1-gpu")


 class TestFlashinferDeterministic(TestDeterministicBase):
--- a/test/registered/core/test_gpt_oss_1gpu.py
+++ b/test/registered/core/test_gpt_oss_1gpu.py
@@ -3,7 +3,7 @@ import unittest
 from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 from sglang.test.gpt_oss_common import BaseTestGptOss

-register_cuda_ci(est_time=402, suite="stage-b-test-small-1-gpu")
+register_cuda_ci(est_time=519, suite="stage-b-test-small-1-gpu")
 register_amd_ci(est_time=750, suite="stage-b-test-small-1-gpu-amd-mi35x")


--- a/test/registered/cuda_graph/test_piecewise_cuda_graph_small_1_gpu.py
+++ b/test/registered/cuda_graph/test_piecewise_cuda_graph_small_1_gpu.py
@@ -21,7 +21,7 @@ from sglang.test.test_utils import (
 )

 # CI Registration - Small 1-GPU tests (24GB GPU sufficient)
-register_cuda_ci(est_time=460, suite="stage-b-test-small-1-gpu")
+register_cuda_ci(est_time=539, suite="stage-b-test-small-1-gpu")


 class TestPiecewiseCudaGraphCorrectness(CustomTestCase):
--- a/test/registered/dllm/test_llada2_mini.py
+++ b/test/registered/dllm/test_llada2_mini.py
@@ -1,6 +1,6 @@
 from sglang.test.ci.ci_register import register_cuda_ci

-register_cuda_ci(est_time=520, suite="stage-b-test-small-1-gpu")
+register_cuda_ci(est_time=181, suite="stage-b-test-small-1-gpu")

 import unittest
 from types import SimpleNamespace
--- a/test/registered/hicache/test_hicache_variants.py
+++ b/test/registered/hicache/test_hicache_variants.py
@@ -1,6 +1,6 @@
 from sglang.test.ci.ci_register import register_cuda_ci

-register_cuda_ci(est_time=368, suite="stage-b-test-small-1-gpu")
+register_cuda_ci(est_time=524, suite="stage-b-test-small-1-gpu")

 """
 Consolidated HiCache variant tests.
--- a/test/registered/lora/test_lora_update.py
+++ b/test/registered/lora/test_lora_update.py
@@ -34,7 +34,7 @@ from sglang.test.test_utils import (
    popen_launch_server,
 )

-register_cuda_ci(est_time=451, suite="stage-b-test-small-1-gpu")
+register_cuda_ci(est_time=487, suite="stage-b-test-small-1-gpu")

 PROMPTS = [
    "SGL is a",
--- a/test/registered/mla/test_flashmla.py
+++ b/test/registered/mla/test_flashmla.py
@@ -21,7 +21,7 @@ from sglang.test.test_utils import (
 )

 # FlashMLA attention backend tests with MTP speculative decoding
-register_cuda_ci(est_time=230, suite="stage-b-test-small-1-gpu")
+register_cuda_ci(est_time=284, suite="stage-b-test-small-1-gpu")


 class TestFlashMLAAttnBackend(unittest.TestCase):
--- a/test/registered/mla/test_mla_int8_deepseek_v3.py
+++ b/test/registered/mla/test_mla_int8_deepseek_v3.py
@@ -16,7 +16,7 @@ from sglang.test.test_utils import (
 )

 # DeepSeek-V3 INT8 quantization tests (channel and block INT8)
-register_cuda_ci(est_time=300, suite="stage-b-test-small-1-gpu")
+register_cuda_ci(est_time=341, suite="stage-b-test-small-1-gpu")


 class TestMLADeepseekV3ChannelInt8(CustomTestCase):
--- a/test/registered/models/test_encoder_embedding_models.py
+++ b/test/registered/models/test_encoder_embedding_models.py
@@ -1,7 +1,7 @@
 from sglang.test.ci.ci_register import register_cuda_ci

 # Encoder embedding model tests (CUDA only)
-register_cuda_ci(est_time=221, suite="stage-b-test-small-1-gpu")
+register_cuda_ci(est_time=270, suite="stage-b-test-small-1-gpu")

 # Copyright 2023-2024 SGLang Team
 # Licensed under the Apache License, Version 2.0 (the "License");
--- a/test/registered/models/test_vlm_models.py
+++ b/test/registered/models/test_vlm_models.py
@@ -1,7 +1,7 @@
 from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci

 # VLM (Vision Language Model) tests
-register_cuda_ci(est_time=270, suite="stage-b-test-small-1-gpu")
+register_cuda_ci(est_time=228, suite="stage-b-test-small-1-gpu")
 register_amd_ci(est_time=420, suite="stage-b-test-small-1-gpu-amd")

 import argparse
--- a/test/registered/moe/test_triton_fused_moe.py
+++ b/test/registered/moe/test_triton_fused_moe.py
@@ -1,6 +1,6 @@
 from sglang.test.ci.ci_register import register_cuda_ci

-register_cuda_ci(est_time=12, suite="stage-b-test-small-1-gpu")
+register_cuda_ci(est_time=89, suite="stage-b-test-small-1-gpu")

 import unittest

--- a/test/registered/openai_server/features/test_enable_thinking.py
+++ b/test/registered/openai_server/features/test_enable_thinking.py
@@ -21,7 +21,7 @@ from sglang.test.test_utils import (
    popen_launch_server,
 )

-register_cuda_ci(est_time=70, suite="stage-b-test-small-1-gpu")
+register_cuda_ci(est_time=103, suite="stage-b-test-small-1-gpu")
 register_amd_ci(est_time=200, suite="stage-b-test-small-1-gpu-amd")


--- a/test/registered/ops/test_repeat_interleave.py
+++ b/test/registered/ops/test_repeat_interleave.py
@@ -1,7 +1,7 @@
 from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci

 # Ops - Repeat Interleave tests (1-GPU)
-register_cuda_ci(est_time=60, suite="stage-b-test-small-1-gpu")
+register_cuda_ci(est_time=8, suite="stage-b-test-small-1-gpu")
 register_amd_ci(est_time=75, suite="stage-b-test-small-1-gpu-amd")

 import time
--- a/test/registered/quant/test_fp8_kernel.py
+++ b/test/registered/quant/test_fp8_kernel.py
@@ -9,7 +9,7 @@ from sglang.srt.layers.quantization.fp8_kernel import (
 from sglang.test.ci.ci_register import register_cuda_ci
 from sglang.test.test_utils import CustomTestCase

-register_cuda_ci(est_time=10, suite="stage-b-test-small-1-gpu")
+register_cuda_ci(est_time=132, suite="stage-b-test-small-1-gpu")


 class TestFP8Base(CustomTestCase):
--- a/test/registered/scheduler/test_no_overlap_scheduler.py
+++ b/test/registered/scheduler/test_no_overlap_scheduler.py
@@ -9,7 +9,7 @@ import unittest
 from sglang.test.ci.ci_register import register_cuda_ci
 from sglang.test.test_utils import CustomTestCase, run_mmlu_test

-register_cuda_ci(est_time=217, suite="stage-b-test-small-1-gpu")
+register_cuda_ci(est_time=245, suite="stage-b-test-small-1-gpu")


 class TestOverlapSchedule(CustomTestCase):
--- a/test/registered/scheduler/test_retract_decode.py
+++ b/test/registered/scheduler/test_retract_decode.py
@@ -17,7 +17,7 @@ from sglang.test.test_utils import (
 )
 from sglang.utils import is_in_ci

-register_cuda_ci(est_time=259, suite="stage-b-test-small-1-gpu")
+register_cuda_ci(est_time=311, suite="stage-b-test-small-1-gpu")
 register_amd_ci(est_time=450, suite="stage-b-test-small-1-gpu-amd")


--- a/test/registered/spec/eagle/test_eagle_infer_a.py
+++ b/test/registered/spec/eagle/test_eagle_infer_a.py
@@ -22,7 +22,7 @@ from sglang.test.test_utils import (
    popen_launch_server,
 )

-register_cuda_ci(est_time=500, suite="stage-b-test-small-1-gpu")
+register_cuda_ci(est_time=561, suite="stage-b-test-small-1-gpu")

 torch_dtype = torch.float16
 prefill_tolerance = 5e-2
--- a/test/registered/spec/eagle/test_eagle_infer_beta.py
+++ b/test/registered/spec/eagle/test_eagle_infer_beta.py
@@ -16,7 +16,7 @@ from sglang.test.test_utils import (
    popen_launch_server,
 )

-register_cuda_ci(est_time=194, suite="stage-b-test-small-1-gpu")
+register_cuda_ci(est_time=283, suite="stage-b-test-small-1-gpu")


 class TestEagleServerBase(CustomTestCase, MatchedStopMixin):
--- a/test/registered/spec/test_ngram_speculative_decoding.py
+++ b/test/registered/spec/test_ngram_speculative_decoding.py
@@ -12,7 +12,7 @@ from sglang.test.test_utils import (
    popen_launch_server,
 )

-register_cuda_ci(est_time=117, suite="stage-b-test-small-1-gpu")
+register_cuda_ci(est_time=230, suite="stage-b-test-small-1-gpu")

 GSM_DATASET_PATH = None

--- a/test/registered/spec/test_standalone_speculative_decoding.py
+++ b/test/registered/spec/test_standalone_speculative_decoding.py
@@ -18,7 +18,7 @@ from sglang.test.test_utils import (
 )

 # Standalone speculative decoding tests (FA3, Triton, FlashInfer backends)
-register_cuda_ci(est_time=150, suite="stage-b-test-small-1-gpu")
+register_cuda_ci(est_time=308, suite="stage-b-test-small-1-gpu")

 GSM_DATASET_PATH = None

--- a/test/registered/vlm/test_vision_openai_server_a.py
+++ b/test/registered/vlm/test_vision_openai_server_a.py
@@ -1,6 +1,6 @@
 from sglang.test.ci.ci_register import register_cuda_ci

-register_cuda_ci(est_time=890, suite="stage-b-test-small-1-gpu")
+register_cuda_ci(est_time=957, suite="stage-b-test-small-1-gpu")

 """
 Usage:
--- a/test/registered/vlm/test_vlm_input_format.py
+++ b/test/registered/vlm/test_vlm_input_format.py
@@ -1,6 +1,6 @@
 from sglang.test.ci.ci_register import register_cuda_ci

-register_cuda_ci(est_time=430, suite="stage-b-test-small-1-gpu")
+register_cuda_ci(est_time=447, suite="stage-b-test-small-1-gpu")

 import json
 import unittest