[CI] Reorganize compile tests so new tests are automatically included in CI (#28625)

Signed-off-by: Yanan Cao <gmagogsfm@gmail.com>

tests/compile/README.md

@@ -0,0 +1,5 @@
+# compile test folder structure
+
+- `compile/test_*.py` : various unit tests meant for testing particular code path/features. Future tests are most likely added here. New test files added here will be included in CI automatically
+- `compile/fullgraph/` : full model tests, including all tests previously in compile/piecewise. These tests do not target particular features. New test files added here will be included in CI automatically
+- `compile/distributed/` : tests that require multiple GPUs. New test files added here will **NOT** be included in CI automatically as these tests generally need to be manually configured to run in runners with particular number/type of GPUs.

tests/compile/distributed/test_async_tp.py

@@ -27,13 +27,13 @@ from vllm.distributed.parallel_state import (
 from vllm.platforms import current_platform
 from vllm.utils.system_utils import update_environment_variables
 
-from ..models.registry import HF_EXAMPLE_MODELS
-from ..utils import (
+from ...models.registry import HF_EXAMPLE_MODELS
+from ...utils import (
     compare_two_settings,
     create_new_process_for_each_test,
     multi_gpu_test,
 )
-from .backend import TestBackend
+from ..backend import TestBackend
 
 FP8_DTYPE = current_platform.fp8_dtype()
 

tests/compile/distributed/test_fusion_all_reduce.py

@@ -33,8 +33,8 @@ from vllm.model_executor.layers.quantization.utils.w8a8_utils import (
 from vllm.platforms import current_platform
 from vllm.utils.system_utils import update_environment_variables
 
-from ..utils import has_module_attribute, multi_gpu_test
-from .backend import TestBackend
+from ...utils import has_module_attribute, multi_gpu_test
+from ..backend import TestBackend
 
 
 class TestAllReduceRMSNormModel(torch.nn.Module):

tests/compile/distributed/test_fusions_e2e.py

@@ -18,7 +18,7 @@ from vllm.platforms import current_platform
 from vllm.utils.flashinfer import has_flashinfer
 from vllm.utils.torch_utils import is_torch_equal_or_newer
 
-from ..utils import flat_product, multi_gpu_test
+from ...utils import flat_product, multi_gpu_test
 
 is_blackwell = lambda: current_platform.is_device_capability(100)
 """Are we running on Blackwell, a lot of tests depend on it"""

tests/compile/distributed/test_sequence_parallelism.py

@@ -32,8 +32,8 @@ from vllm.model_executor.layers.quantization.utils.w8a8_utils import Fp8LinearOp
 from vllm.platforms import current_platform
 from vllm.utils.system_utils import update_environment_variables
 
-from ..utils import multi_gpu_test
-from .backend import TestBackend
+from ...utils import multi_gpu_test
+from ..backend import TestBackend
 
 FP8_DTYPE = current_platform.fp8_dtype()
 prompts = [

tests/compile/fullgraph/test_basic_correctness.py

@@ -7,7 +7,7 @@ import pytest
 from vllm.config import CompilationMode
 from vllm.utils.torch_utils import cuda_device_count_stateless
 
-from ..utils import compare_all_settings
+from ...utils import compare_all_settings
 
 
 @dataclasses.dataclass

tests/compile/fullgraph/test_full_graph.py

@@ -15,7 +15,7 @@ from vllm.config import CompilationConfig, CompilationMode, CUDAGraphMode, PassC
 from vllm.platforms import current_platform
 from vllm.utils.torch_utils import is_torch_equal_or_newer
 
-from ..utils import create_new_process_for_each_test
+from ...utils import create_new_process_for_each_test
 
 
 def models_list(*, all: bool = True, keywords: list[str] | None = None):