Add TPUExecutor

vllm/engine/llm_engine.py

@@ -178,6 +178,9 @@ class LLMEngine:
         if device_config.device_type == "neuron":
             from vllm.executor.neuron_executor import NeuronExecutor
             executor_class = NeuronExecutor
+        elif device_config.device_type == "tpu":
+            from vllm.executor.tpu_executor import TPUExecutor
+            executor_class = TPUExecutor
         elif parallel_config.worker_use_ray:
             initialize_ray_cluster(parallel_config)
             from vllm.executor.ray_gpu_executor import RayGPUExecutor

vllm/executor/tpu_executor.py

@@ -0,0 +1,131 @@
+import os
+from typing import Dict, List, Optional
+
+from vllm.config import (CacheConfig, DeviceConfig, LoRAConfig, ModelConfig,
+                         ParallelConfig, SchedulerConfig, VisionLanguageConfig)
+from vllm.executor.executor_base import ExecutorAsyncBase, ExecutorBase
+from vllm.executor.utils import check_block_size_valid
+from vllm.logger import init_logger
+from vllm.lora.request import LoRARequest
+from vllm.sequence import SamplerOutput, SequenceGroupMetadata
+from vllm.utils import make_async
+
+logger = init_logger(__name__)
+
+
+class TPUExecutor(ExecutorBase):
+
+    def __init__(
+        self,
+        model_config: ModelConfig,
+        cache_config: CacheConfig,
+        parallel_config: ParallelConfig,
+        scheduler_config: SchedulerConfig,
+        device_config: DeviceConfig,
+        lora_config: Optional[LoRAConfig],
+        vision_language_config: Optional[VisionLanguageConfig],
+    ) -> None:
+        self.model_config = model_config
+        self.cache_config = cache_config
+        self.parallel_config = parallel_config
+        self.scheduler_config = scheduler_config
+        self.device_config = device_config
+        assert lora_config is None, "LoRA is not supported for TPU backend."
+        self.vision_language_config = vision_language_config
+
+        # Instantiate the worker and load the model to the device.
+        self._init_worker()
+        # Profile the memory usage and initialize the cache.
+        self._init_cache()
+
+    def _init_worker(self):
+        os.environ["PJRT_DEVICE"] = "TPU"
+        from vllm.worker.tpu_worker import TPUWorker
+
+        assert self.parallel_config.world_size == 1, (
+            "TPUExecutor currently only supports a single TPU chip.")
+        self.driver_worker = TPUWorker(
+            self.model_config,
+            self.parallel_config,
+            self.scheduler_config,
+            self.device_config,
+        )
+        self.driver_worker.init_device()
+        self.driver_worker.load_model()
+
+    def _init_cache(self) -> None:
+        """Profiles the memory usage and initializes the KV cache.
+
+        The engine first profiles the existing memory usage.
+        Then, it allocates the remaining memory for KV blocks.
+
+        .. tip::
+            You may limit the usage of TPU HBM by adjusting the
+            `gpu_memory_utilization` parameter.
+        """
+        # Get the maximum number of blocks that can be allocated on TPU.
+        num_tpu_blocks = self.driver_worker.profile_num_available_blocks(
+                block_size=self.cache_config.block_size,
+                gpu_memory_utilization=self.cache_config.gpu_memory_utilization,
+                cache_dtype=self.cache_config.cache_dtype,
+            )
+        logger.info(f"# TPU blocks: {num_tpu_blocks}")
+
+        check_block_size_valid(num_tpu_blocks, self.cache_config.block_size,
+                               self.model_config.max_model_len)
+        self.cache_config.num_gpu_blocks = num_tpu_blocks
+        self.cache_config.num_cpu_blocks = 0
+
+        # Allocate the KV cache.
+        self.driver_worker.allocate_kv_cache(self.cache_config)
+        # Warm up the model.
+        self.driver_worker.warm_up_model()
+
+    def execute_model(
+        self,
+        seq_group_metadata_list: List[SequenceGroupMetadata],
+        blocks_to_swap_in: Dict[int, int],
+        blocks_to_swap_out: Dict[int, int],
+        blocks_to_copy: Dict[int, List[int]],
+    ) -> SamplerOutput:
+        output = self.driver_worker.execute_model(
+            seq_group_metadata_list=seq_group_metadata_list,
+            blocks_to_swap_in=blocks_to_swap_in,
+            blocks_to_swap_out=blocks_to_swap_out,
+            blocks_to_copy=blocks_to_copy,
+        )
+        return output
+
+    def add_lora(self, lora_request: LoRARequest) -> bool:
+        raise NotImplementedError("LoRA is not implemented for TPU backend.")
+
+    def remove_lora(self, lora_id: int) -> bool:
+        raise NotImplementedError("LoRA is not implemented for TPU backend.")
+
+    def list_loras(self) -> List[int]:
+        raise NotImplementedError("LoRA is not implemented for TPU backend.")
+
+    def check_health(self) -> None:
+        # TPUExecutor will always be healthy as long as it's running.
+        return
+
+
+class TPUExecutorAsync(TPUExecutor, ExecutorAsyncBase):
+
+    async def execute_model_async(
+        self,
+        seq_group_metadata_list: List[SequenceGroupMetadata],
+        blocks_to_swap_in: Dict[int, int],
+        blocks_to_swap_out: Dict[int, int],
+        blocks_to_copy: Dict[int, List[int]],
+    ) -> SamplerOutput:
+        output = await make_async(self.driver_worker.execute_model)(
+            seq_group_metadata_list=seq_group_metadata_list,
+            blocks_to_swap_in=blocks_to_swap_in,
+            blocks_to_swap_out=blocks_to_swap_out,
+            blocks_to_copy=blocks_to_copy)
+        return output
+
+    async def check_health_async(self) -> None:
+        # TPUExecutor will always be healthy as long as it's running.
+        return