[CI][IR] Speed up vLLM IR test group (#55965)

Signed-off-by: Djordje Ramic <djoramic@amd.com>
This commit is contained in:
djramic
2026-09-08 21:28:06 -05:00
committed by GitHub
parent 0b066293f3
commit 21a117ad87
4 changed files with 11 additions and 0 deletions
+3
View File
@@ -1,5 +1,6 @@
# SPDX-License-Identifier: Apache-2.0
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
import pytest
import torch
from torch import Tensor
from torch.fx.experimental.proxy_tensor import make_fx
@@ -7,6 +8,8 @@ from torch.fx.experimental.proxy_tensor import make_fx
import vllm.ir.op
from vllm.ir.op import IrOp, IrOpInplaceOverload
pytestmark = pytest.mark.skip_global_cleanup
@vllm.ir.register_op(allow_inplace=True)
def _custom_mm2(x: Tensor, w: Tensor) -> Tensor:
+2
View File
@@ -13,6 +13,8 @@ from torch.fx.experimental.proxy_tensor import make_fx
import vllm.ir.op
from vllm.ir.op import RESERVED_PROVIDERS, IrOp, IrOpImpl
pytestmark = pytest.mark.skip_global_cleanup
class CustomError(Exception):
pass
+4
View File
@@ -8,9 +8,13 @@ Per-op correctness tests live alongside their op definitions
(e.g. tests/kernels/ir/test_layernorm.py).
"""
import pytest
import vllm.kernels # noqa: F401 — registers provider implementations
from vllm.ir.op import IrOp
pytestmark = pytest.mark.skip_global_cleanup
def test_all_ops_have_input_generator():
missing = [name for name, op in IrOp.registry.items() if not op.has_input_generator]
+2
View File
@@ -16,6 +16,8 @@ from tests.utils import set_random_seed
from vllm import ir
from vllm.platforms import current_platform
pytestmark = pytest.mark.skip_global_cleanup
rms_norm_native = ir.ops.rms_norm.impls["native"].impl_fn
IS_GPGPU_DEVICE = current_platform.is_cuda_alike() or current_platform.is_xpu()