biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Frontend][torch.compile] CompilationConfig Overhaul (#20283): name change compilation level to compilation mode, deprecation compilation level (#26355)

Browse Source 
Signed-off-by: morrison-turnansky <mturnans@redhat.com>
Signed-off-by: Morrison Turnansky <mturnans@redhat.com>
Co-authored-by: Luka Govedič <ProExpertProg@users.noreply.github.com>
This commit is contained in:
Morrison Turnansky
2025-10-14 22:51:16 -04:00
committed by GitHub
parent e66d787bce
commit 96b9aa5aa0
42 changed files with 270 additions and 248 deletions
Download Patch File Download Diff File Expand all files Collapse all files

22
tests/v1/cudagraph/test_cudagraph_dispatch.py
View File

@@ -11,7 +11,7 @@ from vllm.compilation.cuda_graph import CUDAGraphWrapper
from vllm.compilation.monitor import set_cudagraph_capturing_enabled
from vllm.config import (
CompilationConfig,
CompilationLevel,
CompilationMode,
CUDAGraphMode,
ParallelConfig,
SchedulerConfig,
@@ -42,7 +42,7 @@ def _create_vllm_config(
mock_config.parallel_config = ParallelConfig()
# Mimic the behavior of VllmConfig.__post_init__()
if compilation_config.level == CompilationLevel.PIECEWISE:
if compilation_config.mode == CompilationMode.VLLM_COMPILE:
compilation_config.set_splitting_ops_for_v1()
return mock_config
@@ -50,23 +50,23 @@ def _create_vllm_config(
class TestCudagraphDispatcher:
@pytest.mark.parametrize(
"case_id,cudagraph_mode_str,compilation_level",
"case_id,cudagraph_mode_str,compilation_mode",
[
# Test case 0: Full CG for mixed batches, no separate routine
(0, "FULL", CompilationLevel.NO_COMPILATION),
(0, "FULL", CompilationMode.NONE),
# Test case 1: Full CG for uniform batches, piecewise for mixed
(1, "FULL_AND_PIECEWISE", CompilationLevel.NO_COMPILATION),
(1, "FULL_AND_PIECEWISE", CompilationMode.NONE),
# Test case 2: Full CG for uniform batches, no CG for mixed
(2, "FULL_DECODE_ONLY", CompilationLevel.NO_COMPILATION),
# Test case 3: Piecewise for all
(3, "PIECEWISE", CompilationLevel.PIECEWISE),
(2, "FULL_DECODE_ONLY", CompilationMode.NONE),
# Test case 3: PIECEWISE for all
(3, "PIECEWISE", CompilationMode.VLLM_COMPILE),
],
)
def test_dispatcher(self, cudagraph_mode_str, compilation_level):
def test_dispatcher(self, cudagraph_mode_str, compilation_mode):
# Setup dispatcher
comp_config = CompilationConfig(
cudagraph_mode=cudagraph_mode_str,
level=compilation_level,
mode=compilation_mode,
cudagraph_capture_sizes=[1, 8],
)
@@ -242,7 +242,7 @@ class TestCudagraphIntegration:
def setup_method(self):
# only FULL mode for non-uniform batches
self.comp_config = CompilationConfig(
level=CompilationLevel.PIECEWISE,
mode=CompilationMode.VLLM_COMPILE,
cudagraph_mode="FULL",
cudagraph_capture_sizes=[10, 20],
)