[hardware][misc] introduce platform abstraction (vllm-project#6080)

tests/kernels/test_cutlass.py

@@ -8,13 +8,13 @@
 import torch
 
 from vllm import _custom_ops as ops
-from vllm.utils import get_device_capability_stateless
+from vllm.platforms import current_platform
 
 CUDA_DEVICES = [
     f"cuda:{i}" for i in range(1 if torch.cuda.device_count() == 1 else 2)
 ]
 
-capability = get_device_capability_stateless()
+capability = current_platform.get_device_capability()
 capability = capability[0] * 10 + capability[1]
 
 

tests/quantization/utils.py

@@ -1,15 +1,15 @@
 import torch
 
 from vllm.model_executor.layers.quantization import QUANTIZATION_METHODS
-from vllm.utils import get_device_capability_stateless
+from vllm.platforms import current_platform
 
 
 def is_quant_method_supported(quant_method: str) -> bool:
     # Currently, all quantization methods require Nvidia or AMD GPUs
     if not torch.cuda.is_available():
         return False
 
-    capability = get_device_capability_stateless()
+    capability = current_platform.get_device_capability()
     capability = capability[0] * 10 + capability[1]
     return (capability >=
             QUANTIZATION_METHODS[quant_method].get_min_capability())

vllm/attention/ops/blocksparse_attention/interface.py

@@ -2,13 +2,14 @@
 
 import torch
 
-from vllm.utils import get_device_capability_stateless, is_cpu, is_hip
+from vllm.platforms import current_platform
+from vllm.utils import is_cpu, is_hip
 
 from .utils import (dense_to_crow_col, get_head_sliding_step,
                     get_sparse_attn_mask)
 
 IS_COMPUTE_8_OR_ABOVE = (torch.cuda.is_available()
-                         and get_device_capability_stateless()[0] >= 8)
+                         and current_platform.get_device_capability()[0] >= 8)
 
 if IS_COMPUTE_8_OR_ABOVE:
     from .blocksparse_attention_kernel import blocksparse_flash_attn_varlen_fwd

vllm/attention/ops/prefix_prefill.py

@@ -5,7 +5,7 @@
 import triton
 import triton.language as tl
 
-from vllm.utils import get_device_capability_stateless
+from vllm.platforms import current_platform
 
 if triton.__version__ >= "2.1.0":
 
@@ -685,7 +685,7 @@ def context_attention_fwd(q,
                               alibi_slopes=None,
                               sliding_window=None):
 
-        cap = get_device_capability_stateless()
+        cap = current_platform.get_device_capability()
         BLOCK = 128 if cap[0] >= 8 else 64
         # shape constraints
         Lq, Lk, Lv = q.shape[-1], k.shape[-1], v.shape[-1]

vllm/lora/punica.py

@@ -5,14 +5,14 @@
 import torch
 
 from vllm import _custom_ops as ops
-from vllm.utils import get_device_capability_stateless
+from vllm.platforms import current_platform
 
 
 def _check_punica_support():
     if ops.is_custom_op_supported("_punica_C::dispatch_bgmv"):
         return
 
-    if get_device_capability_stateless() < (8, 0):
+    if current_platform.get_device_capability() < (8, 0):
         raise ImportError(
             "punica LoRA kernels require compute capability >= 8.0")
     else:

vllm/model_executor/layers/quantization/compressed_tensors/compressed_tensors.py

@@ -14,7 +14,7 @@
 from vllm.model_executor.layers.quantization.compressed_tensors.utils import (
     CompressionFormat, QuantizationArgs, QuantizationStrategy,
     find_first_name_or_class_match)
-from vllm.utils import get_device_capability_stateless
+from vllm.platforms import current_platform
 
 
 class CompressedTensorsConfig(QuantizationConfig):
@@ -85,7 +85,7 @@ def get_config_filenames(cls) -> List[str]:
         return []
 
     def _check_gptq_and_marlin_can_run(self):
-        capability = get_device_capability_stateless()
+        capability = current_platform.get_device_capability()
         capability = capability[0] * 10 + capability[1]
         if capability < 80:
             raise RuntimeError("The quantization config is not supported for ",

vllm/model_executor/layers/quantization/fp8.py

@@ -12,15 +12,16 @@
 from vllm.model_executor.layers.quantization.base_config import (
     QuantizationConfig, QuantizeMethodBase)
 from vllm.model_executor.utils import set_weight_attrs
-from vllm.utils import get_device_capability_stateless, print_warning_once
+from vllm.platforms import current_platform
+from vllm.utils import print_warning_once
 
 ACTIVATION_SCHEMES = ["static", "dynamic"]
 
 logger = init_logger(__name__)
 
 
 def cutlass_fp8_supported() -> bool:
-    capability = get_device_capability_stateless()
+    capability = current_platform.get_device_capability()
     capability = capability[0] * 10 + capability[1]
 
     return ops.cutlass_scaled_mm_supports_fp8(capability)

vllm/model_executor/layers/quantization/gptq_marlin.py

@@ -12,7 +12,7 @@
 from vllm.model_executor.layers.quantization.base_config import (
     QuantizationConfig)
 from vllm.model_executor.layers.vocab_parallel_embedding import ParallelLMHead
-from vllm.utils import get_device_capability_stateless
+from vllm.platforms import current_platform
 
 logger = init_logger(__name__)
 
@@ -173,7 +173,7 @@ def is_marlin_compatible(cls, quant_config: Dict[str, Any]):
             return False
 
         # If the capability of the device is too low, cannot convert.
-        major, minor = get_device_capability_stateless()
+        major, minor = current_platform.get_device_capability()
         device_capability = major * 10 + minor
         if device_capability < cls.get_min_capability():
             return False

vllm/model_executor/layers/quantization/utils/marlin_utils.py

@@ -12,9 +12,9 @@
     marlin_perm, marlin_scale_perm, marlin_scale_perm_single)
 from vllm.model_executor.layers.quantization.utils.quant_utils import (
     get_pack_factor, quantize_weights, sort_weights)
-from vllm.utils import get_device_capability_stateless
+from vllm.platforms import current_platform
 
-__cuda_arch = get_device_capability_stateless()
+__cuda_arch = current_platform.get_device_capability()
 
 MARLIN_TILE = 16
 

vllm/model_executor/model_loader/loader.py

@@ -35,7 +35,8 @@
 from vllm.model_executor.models.interfaces import (supports_lora,
                                                    supports_vision)
 from vllm.model_executor.utils import set_weight_attrs
-from vllm.utils import get_device_capability_stateless, is_tpu
+from vllm.platforms import current_platform
+from vllm.utils import is_tpu
 
 logger = init_logger(__name__)
 
@@ -46,7 +47,7 @@ def _get_quantization_config(
     """Get the quantization config."""
     if model_config.quantization is not None:
         quant_config = get_quant_config(model_config, load_config)
-        capability = get_device_capability_stateless()
+        capability = current_platform.get_device_capability()
         capability = capability[0] * 10 + capability[1]
         if capability < quant_config.get_min_capability():
             raise ValueError(

vllm/platforms/__init__.py

@@ -0,0 +1,18 @@
+from typing import Optional
+
+import torch
+
+from .interface import Platform, PlatformEnum
+
+current_platform: Optional[Platform]
+
+if torch.version.cuda is not None:
+    from .cuda import CudaPlatform
+    current_platform = CudaPlatform()
+elif torch.version.hip is not None:
+    from .rocm import RocmPlatform
+    current_platform = RocmPlatform()
+else:
+    current_platform = None
+
+__all__ = ['Platform', 'PlatformEnum', 'current_platform']

vllm/platforms/cuda.py

@@ -0,0 +1,34 @@
+"""Code inside this file can safely assume cuda platform, e.g. importing
+pynvml. However, it should not initialize cuda context.
+"""
+
+from functools import lru_cache, wraps
+from typing import Tuple
+
+import pynvml
+
+from .interface import Platform, PlatformEnum
+
+
+def with_nvml_context(fn):
+
+    @wraps(fn)
+    def wrapper(*args, **kwargs):
+        pynvml.nvmlInit()
+        try:
+            return fn(*args, **kwargs)
+        finally:
+            pynvml.nvmlShutdown()
+
+    return wrapper
+
+
+class CudaPlatform(Platform):
+    _enum = PlatformEnum.CUDA
+
+    @staticmethod
+    @lru_cache(maxsize=8)
+    @with_nvml_context
+    def get_device_capability(device_id: int = 0) -> Tuple[int, int]:
+        handle = pynvml.nvmlDeviceGetHandleByIndex(device_id)
+        return pynvml.nvmlDeviceGetCudaComputeCapability(handle)

vllm/platforms/interface.py

@@ -0,0 +1,21 @@
+import enum
+from typing import Tuple
+
+
+class PlatformEnum(enum.Enum):
+    CUDA = enum.auto()
+    ROCM = enum.auto()
+
+
+class Platform:
+    _enum: PlatformEnum
+
+    def is_cuda(self) -> bool:
+        return self._enum == PlatformEnum.CUDA
+
+    def is_rocm(self) -> bool:
+        return self._enum == PlatformEnum.ROCM
+
+    @staticmethod
+    def get_device_capability(device_id: int = 0) -> Tuple[int, int]:
+        raise NotImplementedError

vllm/platforms/rocm.py

@@ -0,0 +1,15 @@
+from functools import lru_cache
+from typing import Tuple
+
+import torch
+
+from .interface import Platform, PlatformEnum
+
+
+class RocmPlatform(Platform):
+    _enum = PlatformEnum.ROCM
+
+    @staticmethod
+    @lru_cache(maxsize=8)
+    def get_device_capability(device_id: int = 0) -> Tuple[int, int]:
+        return torch.cuda.get_device_capability(device_id)

vllm/utils.py

@@ -866,13 +866,6 @@ def is_full_nvlink(device_ids: List[int]) -> bool:
     return True
 
 
-@lru_cache(maxsize=8)
-@with_nvml_context
-def get_device_capability_stateless(device_id: int = 0) -> Tuple[int, int]:
-    handle = pynvml.nvmlDeviceGetHandleByIndex(device_id)
-    return pynvml.nvmlDeviceGetCudaComputeCapability(handle)
-
-
 #From: https://stackoverflow.com/a/4104188/2749989
 def run_once(f):
 

vllm/worker/worker.py

@@ -15,8 +15,8 @@
 from vllm.lora.request import LoRARequest
 from vllm.model_executor import set_random_seed
 from vllm.model_executor.model_loader.tensorizer import TensorizerConfig
+from vllm.platforms import current_platform
 from vllm.sequence import ExecuteModelRequest
-from vllm.utils import get_device_capability_stateless
 from vllm.worker.cache_engine import CacheEngine
 from vllm.worker.embedding_model_runner import EmbeddingModelRunner
 from vllm.worker.model_runner import GPUModelRunnerBase, ModelRunner
@@ -333,7 +333,7 @@ def init_worker_distributed_environment(
 def _check_if_gpu_supports_dtype(torch_dtype: torch.dtype):
     # Check if the GPU supports the dtype.
     if torch_dtype == torch.bfloat16:
-        compute_capability = get_device_capability_stateless()
+        compute_capability = current_platform.get_device_capability()
         if compute_capability[0] < 8:
             gpu_name = torch.cuda.get_device_name()
             raise ValueError(