ef6abf3dc7
Adds ALL files needed for Dockerfile build:
- qwen3_6_scripts/ (baseline patches + our optimizations)
- vllm/ (full vllm package)
- paged_attention_v2_pytorch.py (V2 with single-bmm optimization)
- Dockerfile + computility-run.yaml
Our optimizations vs baseline:
1. paged_attn.py: pre-gathered context KV (eliminates 194 gather calls),
Triton try/fallback, V2 heuristic, threshold 32K→64K
2. paged_attention_v2_pytorch.py: fills NotImplementedError,
single-bmm Phase 1 (195 launches → 3)
3. patch_enable_triton.py: HAS_TRITON=True with safety fallback
4. patch_triton_tuning.py: BLOCK=64, NUM_WARPS=4 for BI-V100
5. computility-run.yaml: gpu-memory-utilization 0.9→0.95,
max-num-batched-tokens 8192→16384
This repo can now be submitted to dev.modelhub.org.cn as-is.
38 lines
1.3 KiB
Python
38 lines
1.3 KiB
Python
import asyncio
|
|
from typing import List, Optional
|
|
|
|
import vllm.envs as envs
|
|
from vllm.executor.ray_gpu_executor import RayGPUExecutor, RayGPUExecutorAsync
|
|
from vllm.executor.xpu_executor import XPUExecutor
|
|
from vllm.logger import init_logger
|
|
from vllm.utils import get_vllm_instance_id, make_async
|
|
|
|
logger = init_logger(__name__)
|
|
|
|
|
|
class RayXPUExecutor(RayGPUExecutor, XPUExecutor):
|
|
|
|
def _get_env_vars_to_be_updated(self):
|
|
# Get the set of GPU IDs used on each node.
|
|
worker_node_and_gpu_ids = self._run_workers("get_node_and_gpu_ids",
|
|
use_dummy_driver=True)
|
|
|
|
VLLM_INSTANCE_ID = get_vllm_instance_id()
|
|
|
|
# Set environment variables for the driver and workers.
|
|
all_args_to_update_environment_variables = [({
|
|
"VLLM_INSTANCE_ID":
|
|
VLLM_INSTANCE_ID,
|
|
"VLLM_TRACE_FUNCTION":
|
|
str(envs.VLLM_TRACE_FUNCTION),
|
|
}, ) for (_, _) in worker_node_and_gpu_ids]
|
|
return all_args_to_update_environment_variables
|
|
|
|
|
|
class RayXPUExecutorAsync(RayXPUExecutor, RayGPUExecutorAsync):
|
|
|
|
def __init__(self, *args, **kwargs):
|
|
super().__init__(*args, **kwargs)
|
|
self.driver_exec_method = make_async(self.driver_worker.execute_method)
|
|
self.pp_locks: Optional[List[asyncio.Lock]] = None
|