A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 15, 04:45 UTC
[Refactor] Rename commnication utils (#20091)
Signed-off-by: yewentao256 <zhyanwentao@126.com>
Wentao Ye committed
Jun 26, 2025 at 18:19 UTC
562308816ceabd8414f49ff2aa291480f69fa1a5
5 files changed
+4
-4
tests/kernels/moe/test_deepep_deepgemm_moe.py
+1
-1
@@ -22,7 +22,7 @@ from vllm.model_executor.layers.quantization.utils.fp8_utils import (
22
per_token_group_quant_fp8)
23
from vllm.platforms import current_platform
24
25
-from .deepep_utils import ProcessGroupInfo, parallel_launch
25
+from .utils import ProcessGroupInfo, parallel_launch
26
27
has_deep_ep = importlib.util.find_spec("deep_ep") is not None
28
tests/kernels/moe/test_deepep_moe.py
+1
-1
@@ -23,7 +23,7 @@ from vllm.model_executor.layers.quantization.utils.fp8_utils import (
23
per_token_group_quant_fp8)
24
from vllm.platforms import current_platform
25
26
-from .deepep_utils import ProcessGroupInfo, parallel_launch
26
+from .utils import ProcessGroupInfo, parallel_launch
27
28
has_deep_ep = importlib.util.find_spec("deep_ep") is not None
29
tests/kernels/moe/test_pplx_cutlass_moe.py
+1
-1
@@ -15,7 +15,7 @@ from vllm.model_executor.layers.fused_moe.modular_kernel import (
15
FusedMoEModularKernel)
16
from vllm.platforms import current_platform
17
18
-from .deepep_utils import ProcessGroupInfo, parallel_launch
18
+from .utils import ProcessGroupInfo, parallel_launch
19
20
try:
21
from pplx_kernels import AllToAll
tests/kernels/moe/test_pplx_moe.py
+1
-1
@@ -29,7 +29,7 @@ from vllm.model_executor.layers.fused_moe.modular_kernel import (
29
FusedMoEModularKernel)
30
from vllm.platforms import current_platform
31
32
-from .deepep_utils import ProcessGroupInfo, parallel_launch
32
+from .utils import ProcessGroupInfo, parallel_launch
33
34
requires_pplx = pytest.mark.skipif(
35
not has_pplx,