chore: update CI test est_time values (#38238)

Co-authored-by: sglang-bot <sglang-bot@users.noreply.github.com>

test/registered/quant/test_autoround_quantization.py

@@ -24,7 +24,7 @@ from sglang.test.test_utils import (
     popen_launch_server,
 )
 
-register_cuda_ci(est_time=120, stage="extra-a", runner_config="1-gpu-large")
+register_cuda_ci(est_time=220, stage="extra-a", runner_config="1-gpu-large")
 
 MMLU_NUM_EXAMPLES = 256
 MMLU_NUM_THREADS = 32

test/registered/quant/test_awq.py

@@ -13,7 +13,7 @@ from sglang.test.test_utils import (
     popen_launch_server,
 )
 
-register_cuda_ci(est_time=160, stage="base-b", runner_config="1-gpu-large")
+register_cuda_ci(est_time=220, stage="base-b", runner_config="1-gpu-large")
 register_amd_ci(est_time=200, suite="stage-b-test-1-gpu-large-amd")
 
 

test/registered/quant/test_block_int8.py

@@ -10,7 +10,7 @@ from sglang.srt.server_args import ServerArgs, set_global_server_args_for_schedu
 from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 from sglang.test.test_utils import CustomTestCase
 
-register_cuda_ci(est_time=44, stage="base-b", runner_config="1-gpu-small")
+register_cuda_ci(est_time=30, stage="base-b", runner_config="1-gpu-small")
 register_amd_ci(est_time=22, suite="stage-b-test-1-gpu-small-amd")
 
 

test/registered/quant/test_fp8_kernel.py

@@ -9,7 +9,7 @@ from sglang.kernels.ops.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, stage="base-b", runner_config="1-gpu-large")
+register_cuda_ci(est_time=9, stage="base-b", runner_config="1-gpu-large")
 
 from sglang.srt.utils import get_device, is_cuda, is_xpu
 

test/registered/quant/test_fp8_utils.py

@@ -19,7 +19,7 @@ from sglang.srt.runtime_context import get_platform
 from sglang.test.ci.ci_register import register_cuda_ci
 from sglang.test.test_utils import CustomTestCase
 
-register_cuda_ci(est_time=12, stage="base-b", runner_config="1-gpu-large")
+register_cuda_ci(est_time=9, stage="base-b", runner_config="1-gpu-large")
 
 
 class TestMxfp8MoeScaleLayout(CustomTestCase):

test/registered/quant/test_fp8kv_triton.py

@@ -12,7 +12,7 @@ from sglang.test.test_utils import (
     popen_launch_server,
 )
 
-register_cuda_ci(est_time=73, stage="extra-a", runner_config="1-gpu-large")
+register_cuda_ci(est_time=64, stage="extra-a", runner_config="1-gpu-large")
 register_amd_ci(est_time=94, suite="extra-a-test-1-gpu-large-amd")
 
 

test/registered/quant/test_gguf.py

@@ -6,7 +6,7 @@ import sglang as sgl
 from sglang.test.ci.ci_register import register_cuda_ci
 from sglang.test.test_utils import CustomTestCase
 
-register_cuda_ci(est_time=76, stage="base-b", runner_config="1-gpu-small")
+register_cuda_ci(est_time=142, stage="base-b", runner_config="1-gpu-small")
 
 
 class TestGGUF(CustomTestCase):

test/registered/quant/test_gptqmodel_dynamic.py

@@ -14,7 +14,7 @@ from sglang.test.test_utils import (
     popen_launch_server,
 )
 
-register_cuda_ci(est_time=100, stage="extra-a", runner_config="1-gpu-large")
+register_cuda_ci(est_time=101, stage="extra-a", runner_config="1-gpu-large")
 
 
 def check_quant_method(model_path: str, use_marlin_kernel: bool):

test/registered/quant/test_int8_kernel.py

@@ -11,7 +11,7 @@ from sglang.srt.server_args import ServerArgs, set_global_server_args_for_schedu
 from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 from sglang.test.test_utils import CustomTestCase
 
-register_cuda_ci(est_time=15, stage="base-b", runner_config="1-gpu-small")
+register_cuda_ci(est_time=11, stage="base-b", runner_config="1-gpu-small")
 register_amd_ci(est_time=15, suite="nightly-amd-kernel-1-gpu", nightly=True)
 
 

test/registered/quant/test_is_layer_skipped.py

@@ -4,7 +4,7 @@ from sglang.srt.layers.quantization.utils import is_layer_skipped
 from sglang.test.ci.ci_register import register_cpu_ci
 from sglang.test.test_utils import CustomTestCase
 
-register_cpu_ci(est_time=5, suite="base-a-test-cpu")
+register_cpu_ci(est_time=10, suite="base-a-test-cpu")
 
 
 # Qwen3-Next FP8 actually publishes the equivalent of this in

test/registered/quant/test_kvfp4_quant_dequant.py

@@ -10,7 +10,7 @@ import torch
 from sglang.srt.layers.quantization.kvfp4_tensor import FP4MXBlock16KVQuantizeUtil
 from sglang.test.ci.ci_register import register_cuda_ci
 
-register_cuda_ci(est_time=30, stage="base-b", runner_config="1-gpu-large")
+register_cuda_ci(est_time=13, stage="base-b", runner_config="1-gpu-large")
 
 
 def calculate_accuracy_metrics(

test/registered/quant/test_marlin_moe.py

@@ -12,7 +12,7 @@ from sglang.test.ci.ci_register import register_cuda_ci
 from sglang.test.test_marlin_utils import awq_marlin_quantize, marlin_quantize
 from sglang.test.test_utils import CustomTestCase
 
-register_cuda_ci(est_time=108, stage="base-b", runner_config="1-gpu-small")
+register_cuda_ci(est_time=88, stage="base-b", runner_config="1-gpu-small")
 
 set_global_server_args_for_scheduler(ServerArgs(model_path="dummy"))
 

test/registered/quant/test_modelopt_fp8.py

@@ -15,7 +15,7 @@ from sglang.test.test_utils import (
 
 # Two classes run from this file: the default server plus the Rust-frontend
 # variant (when the embedded extension is built), each launches a server + eval.
-register_cuda_ci(est_time=124, stage="base-b", runner_config="1-gpu-large")
+register_cuda_ci(est_time=149, stage="base-b", runner_config="1-gpu-large")
 
 
 class TestModeloptFP8(CustomTestCase):

test/registered/quant/test_nvfp4_embedding.py

@@ -11,7 +11,7 @@ from sglang.srt.layers.quantization.modelopt_quant import (
 from sglang.test.ci.ci_register import register_cpu_ci
 from sglang.test.test_utils import CustomTestCase
 
-register_cpu_ci(est_time=5, suite="base-a-test-cpu")
+register_cpu_ci(est_time=10, suite="base-a-test-cpu")
 
 GROUP_SIZE = 16
 

test/registered/quant/test_nvfp4_gemm_sm120.py

@@ -12,7 +12,7 @@ from sglang.test.test_utils import (
     try_cached_model,
 )
 
-register_cuda_ci(est_time=109, stage="base-b", runner_config="1-gpu-small")
+register_cuda_ci(est_time=101, stage="base-b", runner_config="1-gpu-small")
 
 MODEL_PATH = "nvidia/Llama-3.1-8B-Instruct-NVFP4"
 

test/registered/quant/test_quant_config_parsing.py

@@ -5,7 +5,7 @@ from sglang.srt.configs.model_config import ModelConfig
 from sglang.test.ci.ci_register import register_cpu_ci
 from sglang.test.test_utils import CustomTestCase
 
-register_cpu_ci(est_time=15, suite="base-a-test-cpu")
+register_cpu_ci(est_time=11, suite="base-a-test-cpu")
 register_cpu_ci(est_time=8, suite="stage-b-test-cpu-intel")
 
 

test/registered/quant/test_triton_scaled_mm.py

@@ -9,7 +9,7 @@ from sglang.srt.utils.common import get_device
 from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 from sglang.test.test_utils import CustomTestCase
 
-register_cuda_ci(est_time=11, stage="base-b", runner_config="1-gpu-small")
+register_cuda_ci(est_time=10, stage="base-b", runner_config="1-gpu-small")
 register_amd_ci(est_time=12, suite="stage-b-test-1-gpu-small-amd")
 
 

test/registered/quant/test_w8a8_quantization.py

@@ -14,7 +14,7 @@ from sglang.test.test_utils import (
     popen_launch_server,
 )
 
-register_cuda_ci(est_time=232, stage="extra-a", runner_config="1-gpu-large")
+register_cuda_ci(est_time=204, stage="extra-a", runner_config="1-gpu-large")
 
 
 class BaseW8A8Test(CustomTestCase):