Categorize `tests/kernels/` based on kernel type (#16799)

Signed-off-by: mgoin <mgoin64@gmail.com>

tests/kernels/test_block_int8.py → tests/kernels/quantization/test_block_int8.py

@@ -6,6 +6,7 @@ import itertools
 import pytest
 import torch
 
+from tests.kernels.utils_block import native_w8a8_block_matmul
 from vllm.config import VllmConfig, set_current_vllm_config
 from vllm.model_executor.layers.activation import SiluAndMul
 from vllm.model_executor.layers.fused_moe import fused_moe
@@ -13,8 +14,6 @@ from vllm.model_executor.layers.quantization.utils.int8_utils import (
     w8a8_block_int8_matmul)
 from vllm.platforms import current_platform
 
-from .utils_block import native_w8a8_block_matmul
-
 if current_platform.get_device_capability() < (7, 0):
     pytest.skip("INT8 Triton requires CUDA 7.0 or higher",
                 allow_module_level=True)

tests/kernels/test_cutlass_2of4_sparse.py → tests/kernels/quantization/test_cutlass_2of4_sparse.py

@@ -7,13 +7,12 @@ Run `pytest tests/kernels/test_semi_structured.py`.
 import pytest
 import torch
 
+from tests.kernels.utils import baseline_scaled_mm, to_fp8, to_int8
 from vllm import _custom_ops as ops
 from vllm.model_executor.layers.quantization.utils.w8a8_utils import (
     sparse_cutlass_supported)
 from vllm.platforms import current_platform
 
-from .utils import baseline_scaled_mm, to_fp8, to_int8
-
 CUDA_DEVICES = [
     f"cuda:{i}" for i in range(1 if torch.cuda.device_count() == 1 else 2)
 ]

tests/kernels/test_cutlass.py → tests/kernels/quantization/test_cutlass_scaled_mm.py

@@ -8,13 +8,11 @@ import random
 import pytest
 import torch
 
-from tests.kernels.utils import opcheck
+from tests.kernels.utils import baseline_scaled_mm, opcheck, to_fp8, to_int8
 from vllm import _custom_ops as ops
 from vllm.platforms import current_platform
 from vllm.utils import cdiv
 
-from .utils import baseline_scaled_mm, to_fp8, to_int8
-
 MNK_FACTORS = [
     (1, 256, 128),
     (1, 16384, 1024),