[CI] Remove nightly registrations redundant with scheduled stage runs (#31371)

2026-07-15 14:12:07 -07:00
parent 5abec3fbf8
commit e76cc75cfa
51 changed files with 19 additions and 36 deletions
@@ -19,7 +19,6 @@ from sglang.jit_kernel.tests.deepseek_v4.common import (
 from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci

 register_cuda_ci(est_time=30, stage="base-b-kernel-unit", runner_config="1-gpu-large")
-register_cuda_ci(est_time=30, suite="nightly-kernel-1-gpu", nightly=True)
 register_amd_ci(est_time=30, suite="nightly-amd-kernel-1-gpu", nightly=True)

 Context = Union[LegacyContext, PagedContext]
@@ -19,7 +19,6 @@ from sglang.jit_kernel.tests.deepseek_v4.common import (
 from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci

 register_cuda_ci(est_time=30, stage="base-b-kernel-unit", runner_config="1-gpu-large")
-register_cuda_ci(est_time=30, suite="nightly-kernel-1-gpu", nightly=True)
 register_amd_ci(est_time=30, suite="nightly-amd-kernel-1-gpu", nightly=True)

 Context = Union[LegacyContext, PagedContext]
@@ -22,7 +22,6 @@ from sglang.kernels.ops.attention.dsv4.fp4_indexer import (
 from sglang.test.ci.ci_register import register_cuda_ci

 register_cuda_ci(est_time=60, stage="base-b-kernel-unit", runner_config="1-gpu-large")
-register_cuda_ci(est_time=60, suite="nightly-kernel-1-gpu", nightly=True)

 HEAD_DIM = 128
 FP4_DIM = HEAD_DIM // 2
@@ -15,7 +15,6 @@ from sglang.srt.layers.quantization.fp8_utils import (
 from sglang.test.ci.ci_register import register_cuda_ci

 register_cuda_ci(est_time=20, stage="base-b-kernel-unit", runner_config="1-gpu-large")
-register_cuda_ci(est_time=80, suite="nightly-kernel-1-gpu", nightly=True)

 DEVICE = "cuda"
 DTYPE = torch.bfloat16
@@ -14,7 +14,6 @@ from sglang.jit_kernel.diffusion.cutedsl.scale_residual_norm_scale_shift import
 from sglang.test.ci.ci_register import register_cuda_ci

 register_cuda_ci(est_time=28, stage="base-b-kernel-unit", runner_config="1-gpu-large")
-register_cuda_ci(est_time=120, suite="nightly-kernel-1-gpu", nightly=True)

 DEVICE = "cuda"
 SHAPE_MAP = {
@@ -10,7 +10,6 @@ from sglang.jit_kernel.diffusion.triton.group_norm_silu import triton_group_norm
 from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci

 register_cuda_ci(est_time=8, stage="base-b-kernel-unit", runner_config="1-gpu-large")
-register_cuda_ci(est_time=120, suite="nightly-kernel-1-gpu", nightly=True)
 register_amd_ci(est_time=15, suite="nightly-amd-kernel-1-gpu", nightly=True)

 DEVICE = "cuda"
@@ -9,6 +9,7 @@ from sglang.jit_kernel.utils import get_ci_test_range
 from sglang.test.ci.ci_register import register_cuda_ci

 register_cuda_ci(est_time=44, stage="base-b-kernel-unit", runner_config="1-gpu-large")
+# Nightly is not redundant here: it sets SGLANG_JIT_KERNEL_RUN_FULL_TESTS=1 to expand get_ci_test_range sweeps.
 register_cuda_ci(est_time=176, suite="nightly-kernel-1-gpu", nightly=True)

 DEVICE = "cuda"
@@ -13,6 +13,7 @@ from sglang.jit_kernel.utils import get_ci_test_range
 from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci

 register_cuda_ci(est_time=15, stage="base-b-kernel-unit", runner_config="1-gpu-large")
+# Nightly is not redundant here: it sets SGLANG_JIT_KERNEL_RUN_FULL_TESTS=1 to expand get_ci_test_range sweeps.
 register_cuda_ci(est_time=120, suite="nightly-kernel-1-gpu", nightly=True)
 register_amd_ci(est_time=30, suite="nightly-amd-kernel-1-gpu", nightly=True)

@@ -16,6 +16,7 @@ from sglang.jit_kernel.utils import get_ci_test_range
 from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci

 register_cuda_ci(est_time=10, stage="base-b-kernel-unit", runner_config="1-gpu-large")
+# Nightly is not redundant here: it sets SGLANG_JIT_KERNEL_RUN_FULL_TESTS=1 to expand get_ci_test_range sweeps.
 register_cuda_ci(est_time=60, suite="nightly-kernel-1-gpu", nightly=True)
 register_amd_ci(est_time=15, suite="nightly-amd-kernel-1-gpu", nightly=True)

@@ -30,6 +30,7 @@ from sglang.multimodal_gen.runtime.layers.attention.layer import (
 from sglang.test.ci.ci_register import register_cuda_ci

 register_cuda_ci(est_time=15, stage="base-b-kernel-unit", runner_config="1-gpu-large")
+# Nightly is not redundant here: it sets SGLANG_JIT_KERNEL_RUN_FULL_TESTS=1 to expand get_ci_test_range sweeps.
 register_cuda_ci(est_time=60, suite="nightly-kernel-1-gpu", nightly=True)

 DEVICE = "cuda"
@@ -13,6 +13,7 @@ from sglang.jit_kernel.utils import get_ci_test_range
 from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci

 register_cuda_ci(est_time=20, stage="base-b-kernel-unit", runner_config="1-gpu-large")
+# Nightly is not redundant here: it sets SGLANG_JIT_KERNEL_RUN_FULL_TESTS=1 to expand get_ci_test_range sweeps.
 register_cuda_ci(est_time=30, suite="nightly-kernel-1-gpu", nightly=True)
 register_amd_ci(est_time=20, stage="jit-kernel-unit", runner_config="amd")

@@ -7,7 +7,6 @@ from sglang.jit_kernel.add_constant import add_constant
 from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci

 register_cuda_ci(est_time=45, stage="base-b-kernel-unit", runner_config="1-gpu-large")
-register_cuda_ci(est_time=180, suite="nightly-kernel-1-gpu", nightly=True)
 register_amd_ci(est_time=8, stage="jit-kernel-unit", runner_config="amd")


@@ -8,7 +8,6 @@ from sglang.jit_kernel.awq_dequantize import awq_dequantize as jit_awq_dequantiz
 from sglang.test.ci.ci_register import register_cuda_ci

 register_cuda_ci(est_time=9, stage="base-b-kernel-unit", runner_config="1-gpu-large")
-register_cuda_ci(est_time=120, suite="nightly-kernel-1-gpu", nightly=True)

 try:
    from sgl_kernel import awq_dequantize as aot_awq_dequantize
@@ -12,7 +12,6 @@ from sglang.srt.layers.quantization.utils import pack_cols, quantize_weights
 from sglang.test.ci.ci_register import register_cuda_ci

 register_cuda_ci(est_time=10, stage="base-b-kernel-unit", runner_config="1-gpu-large")
-register_cuda_ci(est_time=120, suite="nightly-kernel-1-gpu", nightly=True)


 def _has_aot_awq_marlin_moe_repack() -> bool:
@@ -13,7 +13,6 @@ from sglang.test.ci.ci_register import register_cuda_ci
 from sglang.test.test_marlin_utils import get_weight_perm, marlin_weights

 register_cuda_ci(est_time=10, stage="base-b-kernel-unit", runner_config="1-gpu-large")
-register_cuda_ci(est_time=120, suite="nightly-kernel-1-gpu", nightly=True)


 def _has_aot_awq_marlin_repack() -> bool:
@@ -7,7 +7,6 @@ from sglang.jit_kernel.clamp_position import clamp_position_cuda
 from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci

 register_cuda_ci(est_time=12, stage="base-b-kernel-unit", runner_config="1-gpu-large")
-register_cuda_ci(est_time=120, suite="nightly-kernel-1-gpu", nightly=True)
 register_amd_ci(est_time=12, stage="jit-kernel-unit", runner_config="amd")


@@ -8,7 +8,6 @@ import triton
 from sglang.test.ci.ci_register import register_cuda_ci

 register_cuda_ci(est_time=17, stage="base-b-kernel-unit", runner_config="1-gpu-large")
-register_cuda_ci(est_time=120, suite="nightly-kernel-1-gpu", nightly=True)


 def torch_concat_mla_k(
@@ -48,6 +48,7 @@ register_cuda_ci(
    stage="base-b-kernel-unit",
    runner_config="8-gpu-h200",
 )
+# Nightly is not redundant here: it sets SGLANG_JIT_KERNEL_RUN_FULL_TESTS=1 to expand get_ci_test_range sweeps.
 register_cuda_ci(
    est_time=300,
    suite="nightly-kernel-8-gpu-h200",
@@ -30,7 +30,6 @@ except ImportError:
    TRITON_AVAILABLE = False

 register_cuda_ci(est_time=5, stage="base-b-kernel-unit", runner_config="1-gpu-large")
-register_cuda_ci(est_time=120, suite="nightly-kernel-1-gpu", nightly=True)


 def run_triton_kernel(A_log, dt_bias, q, k, v, a, b, initial_state, indices, scale):
@@ -32,7 +32,6 @@ from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci
 _is_hip = is_hip()

 register_cuda_ci(est_time=45, stage="base-b-kernel-unit", runner_config="1-gpu-large")
-register_cuda_ci(est_time=90, suite="nightly-kernel-1-gpu", nightly=True)
 register_amd_ci(est_time=45, suite="jit-kernel-unit-test-amd")

 HEAD_DIM = 128
@@ -11,6 +11,7 @@ from sglang.jit_kernel.utils import get_ci_test_range, get_jit_cuda_arch, is_hip
 from sglang.test.ci.ci_register import register_cuda_ci

 register_cuda_ci(est_time=37, stage="base-b-kernel-unit", runner_config="1-gpu-large")
+# Nightly is not redundant here: it sets SGLANG_JIT_KERNEL_RUN_FULL_TESTS=1 to expand get_ci_test_range sweeps.
 register_cuda_ci(est_time=148, suite="nightly-kernel-1-gpu", nightly=True)

 HIDDEN_DIMS = [1024, 4096, 5120, 6144, 7168]
@@ -16,7 +16,6 @@ from sglang.test.ci.ci_register import register_cuda_ci

 register_cuda_ci(est_time=120, stage="base-b-kernel-unit", runner_config="1-gpu-large")
 register_cuda_ci(est_time=120, stage="base-b-kernel-unit", runner_config="4-gpu-b200")
-register_cuda_ci(est_time=900, suite="nightly-kernel-1-gpu", nightly=True)

 # Skip this test on Hopper machine
 skip_condition = torch.cuda.get_device_capability() < (10, 0)
@@ -8,6 +8,7 @@ from sglang.jit_kernel.utils import get_ci_test_range
 from sglang.test.ci.ci_register import register_cuda_ci

 register_cuda_ci(est_time=10, stage="base-b-kernel-unit", runner_config="1-gpu-large")
+# Nightly is not redundant here: it sets SGLANG_JIT_KERNEL_RUN_FULL_TESTS=1 to expand get_ci_test_range sweeps.
 register_cuda_ci(est_time=120, suite="nightly-kernel-1-gpu", nightly=True)


@@ -7,7 +7,6 @@ from sglang.jit_kernel.fused_eh_norm import fused_eh_norm
 from sglang.test.ci.ci_register import register_cuda_ci

 register_cuda_ci(est_time=45, stage="base-b-kernel-unit", runner_config="1-gpu-large")
-register_cuda_ci(est_time=120, suite="nightly-kernel-1-gpu", nightly=True)

 pytestmark = pytest.mark.skipif(
    not torch.cuda.is_available() or torch.version.cuda is None,
@@ -17,7 +17,6 @@ import torch
 from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci

 register_cuda_ci(est_time=100, stage="base-b-kernel-unit", runner_config="1-gpu-large")
-register_cuda_ci(est_time=400, suite="nightly-kernel-1-gpu", nightly=True)
 register_amd_ci(est_time=100, suite="nightly-amd-kernel-1-gpu", nightly=True)

 # =============================================================================
@@ -49,7 +49,6 @@ except ImportError:
    _is_fp8_fnuz = False

 register_cuda_ci(est_time=24, stage="base-b-kernel-unit", runner_config="1-gpu-large")
-register_cuda_ci(est_time=120, suite="nightly-kernel-1-gpu", nightly=True)
 register_amd_ci(est_time=24, suite="nightly-amd-kernel-1-gpu", nightly=True)

 PAGE_SIZE = 64
@@ -27,7 +27,6 @@ except ImportError:
    KERNELS_AVAILABLE = False

 register_cuda_ci(est_time=6, stage="base-b-kernel-unit", runner_config="1-gpu-large")
-register_cuda_ci(est_time=120, suite="nightly-kernel-1-gpu", nightly=True)
 register_amd_ci(est_time=10, suite="nightly-amd-kernel-1-gpu", nightly=True)


@@ -24,7 +24,6 @@ from sglang.test.test_marlin_utils import (
 )

 register_cuda_ci(est_time=13, stage="base-b-kernel-unit", runner_config="1-gpu-large")
-register_cuda_ci(est_time=120, suite="nightly-kernel-1-gpu", nightly=True)

 MNK_FACTORS = [
    (1, 1, 1),
@@ -14,7 +14,6 @@ from sglang.test.ci.ci_register import register_cuda_ci
 from sglang.test.test_marlin_utils import get_weight_perm, marlin_weights

 register_cuda_ci(est_time=16, stage="base-b-kernel-unit", runner_config="1-gpu-large")
-register_cuda_ci(est_time=120, suite="nightly-kernel-1-gpu", nightly=True)

 MARLIN_K_CHUNKS = [128]
 MARLIN_N_CHUNKS = [64, 256]
@@ -17,7 +17,6 @@ from sglang.jit_kernel.hadamard import (
 from sglang.test.ci.ci_register import register_cuda_ci

 register_cuda_ci(est_time=128, stage="base-b-kernel-unit", runner_config="1-gpu-large")
-register_cuda_ci(est_time=512, suite="nightly-kernel-1-gpu", nightly=True)

 # Exact M×N Hadamard matrices (±1 entries) copied from
 # python/sglang/jit_kernel/csrc/fast-hadamard-transform/code_gen.py.
@@ -15,7 +15,6 @@ from sglang.srt.utils import is_cuda, is_hip, is_npu, is_xpu
 from sglang.test.ci.ci_register import register_cuda_ci

 register_cuda_ci(est_time=10, stage="base-b-kernel-unit", runner_config="1-gpu-large")
-register_cuda_ci(est_time=120, suite="nightly-kernel-1-gpu", nightly=True)

 pytestmark = pytest.mark.skipif(
    not torch.cuda.is_available()
@@ -13,7 +13,6 @@ from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci

 register_amd_ci(est_time=30, stage="stage-b", runner_config="1-gpu-small-amd")
 register_cuda_ci(est_time=10, stage="base-b-kernel-unit", runner_config="1-gpu-large")
-register_cuda_ci(est_time=120, suite="nightly-kernel-1-gpu", nightly=True)

 pytestmark = pytest.mark.skipif(
    not torch.cuda.is_available()
@@ -11,6 +11,7 @@ from sglang.jit_kernel.utils import get_ci_test_range
 from sglang.test.ci.ci_register import register_cuda_ci

 register_cuda_ci(est_time=28, stage="base-b-kernel-unit", runner_config="1-gpu-large")
+# Nightly is not redundant here: it sets SGLANG_JIT_KERNEL_RUN_FULL_TESTS=1 to expand get_ci_test_range sweeps.
 register_cuda_ci(est_time=120, suite="nightly-kernel-1-gpu", nightly=True)


@@ -12,7 +12,6 @@ from sglang.jit_kernel.moe_lora_align import moe_lora_align_block_size
 from sglang.test.ci.ci_register import register_cuda_ci

 register_cuda_ci(est_time=28, stage="base-b-kernel-unit", runner_config="1-gpu-large")
-register_cuda_ci(est_time=120, suite="nightly-kernel-1-gpu", nightly=True)


 def round_up(x, base):
@@ -21,7 +21,6 @@ from sglang.test.test_marlin_utils import (
 )

 register_cuda_ci(est_time=10, stage="base-b-kernel-unit", runner_config="1-gpu-large")
-register_cuda_ci(est_time=120, suite="nightly-kernel-1-gpu", nightly=True)


 def _has_aot_moe_wna16_marlin_gemm() -> bool:
@@ -11,7 +11,6 @@ from sglang.jit_kernel.mxfp8 import (
 from sglang.test.ci.ci_register import register_cuda_ci

 register_cuda_ci(est_time=5, stage="base-b-kernel-unit", runner_config="1-gpu-large")
-register_cuda_ci(est_time=120, suite="nightly-kernel-1-gpu", nightly=True)


 def align(val: int, alignment: int = 128) -> int:
@@ -10,6 +10,7 @@ from sglang.jit_kernel.utils import get_ci_test_range
 from sglang.test.ci.ci_register import register_cuda_ci

 register_cuda_ci(est_time=16, stage="base-b-kernel-unit", runner_config="1-gpu-large")
+# Nightly is not redundant here: it sets SGLANG_JIT_KERNEL_RUN_FULL_TESTS=1 to expand get_ci_test_range sweeps.
 register_cuda_ci(est_time=120, suite="nightly-kernel-1-gpu", nightly=True)

 try:
@@ -12,6 +12,7 @@ from sglang.srt.utils import is_hip
 from sglang.test.ci.ci_register import register_cuda_ci

 register_cuda_ci(est_time=16, stage="base-b-kernel-unit", runner_config="1-gpu-large")
+# Nightly is not redundant here: it sets SGLANG_JIT_KERNEL_RUN_FULL_TESTS=1 to expand get_ci_test_range sweeps.
 register_cuda_ci(est_time=120, suite="nightly-kernel-1-gpu", nightly=True)

 if not torch.cuda.is_available():
@@ -11,7 +11,6 @@ from sglang.jit_kernel.rope import rotary_embedding
 from sglang.test.ci.ci_register import register_cuda_ci

 register_cuda_ci(est_time=18, stage="base-b-kernel-unit", runner_config="1-gpu-large")
-register_cuda_ci(est_time=120, suite="nightly-kernel-1-gpu", nightly=True)


@triton.jit
@@ -9,6 +9,7 @@ from sglang.jit_kernel.utils import get_ci_test_range
 from sglang.test.ci.ci_register import register_cuda_ci

 register_cuda_ci(est_time=37, stage="base-b-kernel-unit", runner_config="1-gpu-large")
+# Nightly is not redundant here: it sets SGLANG_JIT_KERNEL_RUN_FULL_TESTS=1 to expand get_ci_test_range sweeps.
 register_cuda_ci(est_time=148, suite="nightly-kernel-1-gpu", nightly=True)


@@ -9,6 +9,7 @@ from sglang.jit_kernel.utils import get_ci_test_range
 from sglang.test.ci.ci_register import register_cuda_ci

 register_cuda_ci(est_time=15, stage="base-b-kernel-unit", runner_config="1-gpu-large")
+# Nightly is not redundant here: it sets SGLANG_JIT_KERNEL_RUN_FULL_TESTS=1 to expand get_ci_test_range sweeps.
 register_cuda_ci(est_time=120, suite="nightly-kernel-1-gpu", nightly=True)


@@ -10,7 +10,6 @@ import torch
 from sglang.test.ci.ci_register import register_cuda_ci

 register_cuda_ci(est_time=6, stage="base-b-kernel-unit", runner_config="1-gpu-large")
-register_cuda_ci(est_time=120, suite="nightly-kernel-1-gpu", nightly=True)


@pytest.mark.parametrize("batch_size", [1, 99, 989])
@@ -7,7 +7,6 @@ from sglang.jit_kernel.resolve_future_token_ids import resolve_future_token_ids_
 from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci

 register_cuda_ci(est_time=9, stage="base-b-kernel-unit", runner_config="1-gpu-large")
-register_cuda_ci(est_time=120, suite="nightly-kernel-1-gpu", nightly=True)
 register_amd_ci(est_time=9, stage="jit-kernel-unit", runner_config="amd")


@@ -9,6 +9,7 @@ from sglang.srt.utils import is_hip
 from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci

 register_cuda_ci(est_time=45, stage="base-b-kernel-unit", runner_config="1-gpu-large")
+# Nightly is not redundant here: it sets SGLANG_JIT_KERNEL_RUN_FULL_TESTS=1 to expand get_ci_test_range sweeps.
 register_cuda_ci(est_time=240, suite="nightly-kernel-1-gpu", nightly=True)
 register_amd_ci(est_time=45, suite="jit-kernel-unit-test-amd")

@@ -14,6 +14,7 @@ from sglang.jit_kernel.utils import get_ci_test_range
 from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci

 register_cuda_ci(est_time=30, stage="base-b-kernel-unit", runner_config="1-gpu-large")
+# Nightly is not redundant here: it sets SGLANG_JIT_KERNEL_RUN_FULL_TESTS=1 to expand get_ci_test_range sweeps.
 register_cuda_ci(est_time=120, suite="nightly-kernel-1-gpu", nightly=True)
 register_amd_ci(est_time=30, stage="jit-kernel-unit", runner_config="amd")

@@ -9,6 +9,7 @@ from sglang.srt.utils import is_hip
 from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci

 register_cuda_ci(est_time=64, stage="base-b-kernel-unit", runner_config="1-gpu-large")
+# Nightly is not redundant here: it sets SGLANG_JIT_KERNEL_RUN_FULL_TESTS=1 to expand get_ci_test_range sweeps.
 register_cuda_ci(est_time=256, suite="nightly-kernel-1-gpu", nightly=True)
 register_amd_ci(est_time=64, suite="jit-kernel-unit-test-amd")

@@ -10,7 +10,6 @@ from sglang.srt.utils import is_sm90_supported
 from sglang.test.ci.ci_register import register_cuda_ci

 register_cuda_ci(est_time=120, stage="base-b-kernel-unit", runner_config="1-gpu-large")
-register_cuda_ci(est_time=300, suite="nightly-kernel-1-gpu", nightly=True)


 DTYPE_FP8 = torch.float8_e4m3fn
@@ -9,6 +9,7 @@ from sglang.jit_kernel.utils import get_ci_test_range
 from sglang.test.ci.ci_register import register_amd_ci, register_cuda_ci

 register_cuda_ci(est_time=28, stage="base-b-kernel-unit", runner_config="1-gpu-large")
+# Nightly is not redundant here: it sets SGLANG_JIT_KERNEL_RUN_FULL_TESTS=1 to expand get_ci_test_range sweeps.
 register_cuda_ci(est_time=120, suite="nightly-kernel-1-gpu", nightly=True)
 register_amd_ci(est_time=55, stage="jit-kernel-unit", runner_config="amd")

@@ -35,6 +35,7 @@ from sglang.srt.distributed.device_communicators.triton_symm_mem_ag import (
 from sglang.test.ci.ci_register import register_cuda_ci

 register_cuda_ci(est_time=240, stage="base-b-kernel-unit", runner_config="8-gpu-h200")
+# Nightly is not redundant here: it sets SGLANG_JIT_KERNEL_RUN_FULL_TESTS=1 to expand get_ci_test_range sweeps.
 register_cuda_ci(est_time=240, suite="nightly-kernel-8-gpu-h200", nightly=True)

 # ---------------------------------------------------------------------------
@@ -17,6 +17,7 @@ from sglang.jit_kernel.utils import get_ci_test_range
 from sglang.test.ci.ci_register import register_cuda_ci

 register_cuda_ci(est_time=16, stage="base-b-kernel-unit", runner_config="1-gpu-large")
+# Nightly is not redundant here: it sets SGLANG_JIT_KERNEL_RUN_FULL_TESTS=1 to expand get_ci_test_range sweeps.
 register_cuda_ci(est_time=120, suite="nightly-kernel-1-gpu", nightly=True)

 CORRECTNESS_BATCH_SIZES = get_ci_test_range(
@@ -32,11 +32,6 @@ register_cuda_ci(
    stage="base-b-kernel-unit",
    runner_config="8-gpu-h200",
 )
-register_cuda_ci(
-    est_time=300,
-    suite="nightly-kernel-8-gpu-h200",
-    nightly=True,
-)


 # ---------------------------------------------------------------------------