[Hardware][Intel-Gaudi] Add Intel Gaudi (HPU) inference backend (#6143)

Signed-off-by: yuwenzho <yuwen.zhou@intel.com>
Signed-off-by: Chendi.Xue <chendi.xue@intel.com>
Signed-off-by: Bob Zhu <bob.zhu@intel.com>
Signed-off-by: zehao-intel <zehao.huang@intel.com>
Signed-off-by: Konrad Zawora <kzawora@habana.ai>
Co-authored-by: Kunshang Ji <kunshang.ji@intel.com>
Co-authored-by: Sanju C Sudhakaran <scsudhakaran@habana.ai>
Co-authored-by: Michal Adamczyk <madamczyk@habana.ai>
Co-authored-by: Marceli Fylcek <mfylcek@habana.ai>
Co-authored-by: Himangshu Lahkar <49579433+hlahkar@users.noreply.github.com>
Co-authored-by: Vivek Goel <vgoel@habana.ai>
Co-authored-by: yuwenzho <yuwen.zhou@intel.com>
Co-authored-by: Dominika Olszewska <dolszewska@habana.ai>
Co-authored-by: barak goldberg <149692267+bgoldberg-habana@users.noreply.github.com>
Co-authored-by: Michal Szutenberg <37601244+szutenberg@users.noreply.github.com>
Co-authored-by: Jan Kaniecki <jkaniecki@habana.ai>
Co-authored-by: Agata Dobrzyniewicz <160237065+adobrzyniewicz-habana@users.noreply.github.com>
Co-authored-by: Krzysztof Wisniewski <kwisniewski@habana.ai>
Co-authored-by: Dudi Lester <160421192+dudilester@users.noreply.github.com>
Co-authored-by: Ilia Taraban <tarabanil@gmail.com>
Co-authored-by: Chendi.Xue <chendi.xue@intel.com>
Co-authored-by: Michał Kuligowski <mkuligowski@habana.ai>
Co-authored-by: Jakub Maksymczuk <jmaksymczuk@habana.ai>
Co-authored-by: Tomasz Zielinski <85164140+tzielinski-habana@users.noreply.github.com>
Co-authored-by: Sun Choi <schoi@habana.ai>
Co-authored-by: Iryna Boiko <iboiko@habana.ai>
Co-authored-by: Bob Zhu <41610754+czhu15@users.noreply.github.com>
Co-authored-by: hlin99 <73271530+hlin99@users.noreply.github.com>
Co-authored-by: Zehao Huang <zehao.huang@intel.com>
Co-authored-by: Andrzej Kotłowski <Andrzej.Kotlowski@intel.com>
Co-authored-by: Yan Tomsinsky <73292515+Yantom1@users.noreply.github.com>
Co-authored-by: Nir David <ndavid@habana.ai>
Co-authored-by: Yu-Zhou <yu.zhou@intel.com>
Co-authored-by: Ruheena Suhani Shaik <rsshaik@habana.ai>
Co-authored-by: Karol Damaszke <kdamaszke@habana.ai>
Co-authored-by: Marcin Swiniarski <mswiniarski@habana.ai>
Co-authored-by: Woosuk Kwon <woosuk.kwon@berkeley.edu>
Co-authored-by: Jacek Czaja <jacek.czaja@intel.com>
Co-authored-by: Jacek Czaja <jczaja@habana.ai>
Co-authored-by: Yuan <yuan.zhou@outlook.com>
This commit is contained in:
Konrad Zawora
2024-11-06 01:09:10 -08:00
committed by GitHub
co-authored by Kunshang Ji Sanju C Sudhakaran Michal Adamczyk Marceli Fylcek Himangshu Lahkar Vivek Goel yuwenzho Dominika Olszewska barak goldberg Michal Szutenberg Jan Kaniecki Agata Dobrzyniewicz Krzysztof Wisniewski Dudi Lester Ilia Taraban Chendi.Xue Michał Kuligowski Jakub Maksymczuk Tomasz Zielinski Sun Choi Iryna Boiko Bob Zhu hlin99 Zehao Huang Andrzej Kotłowski Yan Tomsinsky Nir David Yu-Zhou Ruheena Suhani Shaik Karol Damaszke Marcin Swiniarski Woosuk Kwon Jacek Czaja Jacek Czaja Yuan
parent a5fda50a10
commit a02a50e6e5
31 changed files with 4279 additions and 20 deletions
+7 -4
View File
@@ -17,6 +17,7 @@ from vllm.config import (CacheConfig, ConfigFormat, DecodingConfig,
from vllm.executor.executor_base import ExecutorBase
from vllm.logger import init_logger
from vllm.model_executor.layers.quantization import QUANTIZATION_METHODS
from vllm.platforms import current_platform
from vllm.transformers_utils.config import (
maybe_register_config_serialize_by_value)
from vllm.transformers_utils.utils import check_gguf_file
@@ -37,6 +38,7 @@ DEVICE_OPTIONS = [
"openvino",
"tpu",
"xpu",
"hpu",
]
@@ -110,7 +112,9 @@ class EngineArgs:
pipeline_parallel_size: int = 1
tensor_parallel_size: int = 1
max_parallel_loading_workers: Optional[int] = None
block_size: int = 16
# NOTE(kzawora): default block size for Gaudi should be 128
# smaller sizes still work, but very inefficiently
block_size: int = 16 if not current_platform.is_hpu() else 128
enable_prefix_caching: bool = False
disable_sliding_window: bool = False
use_v2_block_manager: bool = True
@@ -397,7 +401,7 @@ class EngineArgs:
parser.add_argument('--block-size',
type=int,
default=EngineArgs.block_size,
choices=[8, 16, 32],
choices=[8, 16, 32, 64, 128],
help='Token block size for contiguous chunks of '
'tokens. This is ignored on neuron devices and '
'set to max-model-len')
@@ -1132,8 +1136,7 @@ class EngineArgs:
multi_step_stream_outputs=self.multi_step_stream_outputs,
send_delta_data=(envs.VLLM_USE_RAY_SPMD_WORKER
and parallel_config.use_ray),
policy=self.scheduling_policy,
)
policy=self.scheduling_policy)
lora_config = LoRAConfig(
max_lora_rank=self.max_lora_rank,
max_loras=self.max_loras,
+8
View File
@@ -627,6 +627,14 @@ class AsyncLLMEngine(EngineClient):
elif engine_config.device_config.device_type == "cpu":
from vllm.executor.cpu_executor import CPUExecutorAsync
executor_class = CPUExecutorAsync
elif engine_config.device_config.device_type == "hpu":
if distributed_executor_backend == "ray":
initialize_ray_cluster(engine_config.parallel_config)
from vllm.executor.ray_hpu_executor import RayHPUExecutorAsync
executor_class = RayHPUExecutorAsync
else:
from vllm.executor.hpu_executor import HPUExecutorAsync
executor_class = HPUExecutorAsync
elif engine_config.device_config.device_type == "openvino":
assert distributed_executor_backend is None, (
"Distributed execution is not supported with "
+8
View File
@@ -528,6 +528,14 @@ class LLMEngine:
elif engine_config.device_config.device_type == "cpu":
from vllm.executor.cpu_executor import CPUExecutor
executor_class = CPUExecutor
elif engine_config.device_config.device_type == "hpu":
if distributed_executor_backend == "ray":
initialize_ray_cluster(engine_config.parallel_config)
from vllm.executor.ray_hpu_executor import RayHPUExecutor
executor_class = RayHPUExecutor
else:
from vllm.executor.hpu_executor import HPUExecutor
executor_class = HPUExecutor
elif engine_config.device_config.device_type == "openvino":
from vllm.executor.openvino_executor import OpenVINOExecutor
executor_class = OpenVINOExecutor