apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-10-06 00:57:33 +08:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

[Other] Optimize paddle backend (#1265)

Browse Source 
* Optimize paddle backend

* optimize paddle backend

* add version support
This commit is contained in:
Jason
2023-02-08 19:12:03 +08:00
committed by GitHub
parent 60ba4b06c1
commit a4b0565b9a
10 changed files with 265 additions and 174 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
python/fastdeploy/__init__.py
View File

@@ -39,3 +39,5 @@ from . import text
from . import encryption
from .download import download, download_and_decompress, download_model, get_model_list
from . import serving
from .code_version import version, git_version
__version__ = version

58
python/fastdeploy/runtime.py
View File

@@ -364,7 +364,10 @@ class RuntimeOption:
def set_paddle_mkldnn(self, use_mkldnn=True):
"""Enable/Disable MKLDNN while using Paddle Inference backend, mkldnn is enabled by default.
"""
return self._option.set_paddle_mkldnn(use_mkldnn)
logging.warning(
"`RuntimeOption.set_paddle_mkldnn` will be derepcated in v1.2.0, please use `RuntimeOption.paddle_infer_option.enable_mkldnn = True` instead."
)
self._option.paddle_infer_option.enable_mkldnn = True
def set_openvino_device(self, name="CPU"):
"""Set device name for OpenVINO, default 'CPU', can also be 'AUTO', 'GPU', 'GPU.1'....
@@ -400,17 +403,26 @@ class RuntimeOption:
def enable_paddle_log_info(self):
"""Enable print out the debug log information while using Paddle Inference backend, the log information is disabled by default.
"""
return self._option.enable_paddle_log_info()
logging.warning(
"RuntimeOption.enable_paddle_log_info` will be deprecated in v1.2.0, please use `RuntimeOption.paddle_infer_option.enable_log_info = True` instead."
)
self._option.paddle_infer_option.enable_log_info = True
def disable_paddle_log_info(self):
"""Disable print out the debug log information while using Paddle Inference backend, the log information is disabled by default.
"""
return self._option.disable_paddle_log_info()
logging.warning(
"RuntimeOption.disable_paddle_log_info` will be deprecated in v1.2.0, please use `RuntimeOption.paddle_infer_option.enable_log_info = False` instead."
)
self._option.paddle_infer_option.enable_log_info = False
def set_paddle_mkldnn_cache_size(self, cache_size):
"""Set size of shape cache while using Paddle Inference backend with MKLDNN enabled, default will cache all the dynamic shape.
"""
return self._option.set_paddle_mkldnn_cache_size(cache_size)
logging.warning(
"RuntimeOption.set_paddle_mkldnn_cache_size` will be deprecated in v1.2.0, please use `RuntimeOption.paddle_infer_option.mkldnn_cache_size = {}` instead.".
format(cache_size))
self._option.paddle_infer_option.mkldnn_cache_size = cache_size
def enable_lite_fp16(self):
"""Enable half precision inference while using Paddle Lite backend on ARM CPU, fp16 is disabled by default.
@@ -498,6 +510,16 @@ class RuntimeOption:
def enable_paddle_to_trt(self):
"""While using TensorRT backend, enable_paddle_to_trt() will change to use Paddle Inference backend, and use its integrated TensorRT instead.
"""
logging.warning(
"`RuntimeOption.enable_paddle_to_trt` will be deprecated in v1.2.l0, if you want to run tensorrt with Paddle Inference backend, please use the following method, "
)
logging.warning(" ==============================================")
logging.warning(" import fastdeploy as fd")
logging.warning(" option = fd.RuntimeOption()")
logging.warning(" option.use_gpu(0)")
logging.warning(" option.use_paddle_infer_backend()")
logging.warning(" option.paddle_infer_option.enabel_trt = True")
logging.warning(" ==============================================")
return self._option.enable_paddle_to_trt()
def set_trt_max_workspace_size(self, trt_max_workspace_size):
@@ -519,22 +541,34 @@ class RuntimeOption:
def enable_paddle_trt_collect_shape(self):
"""Enable collect subgraph shape information while using Paddle Inference with TensorRT
"""
return self._option.enable_paddle_trt_collect_shape()
logging.warning(
"`RuntimeOption.enable_paddle_trt_collect_shape` will be deprecated in v1.2.0, please use `RuntimeOption.paddle_infer_option.collect_trt_shape = True` instead."
)
self._option.paddle_infer_option.collect_trt_shape = True
def disable_paddle_trt_collect_shape(self):
"""Disable collect subgraph shape information while using Paddle Inference with TensorRT
"""
return self._option.disable_paddle_trt_collect_shape()
logging.warning(
"`RuntimeOption.disable_paddle_trt_collect_shape` will be deprecated in v1.2.0, please use `RuntimeOption.paddle_infer_option.collect_trt_shape = False` instead."
)
self._option.paddle_infer_option.collect_trt_shape = False
def delete_paddle_backend_pass(self, pass_name):
"""Delete pass by name in paddle backend
"""
return self._option.delete_paddle_backend_pass(pass_name)
logging.warning(
"`RuntimeOption.delete_paddle_backend_pass` will be deprecated in v1.2.0, please use `RuntimeOption.paddle_infer_option.delete_pass` instead."
)
self._option.paddle_infer_option.delete_pass(pass_name)
def disable_paddle_trt_ops(self, ops):
"""Disable some ops in paddle trt backend
"""
return self._option.disable_paddle_trt_ops(ops)
logging.warning(
"`RuntimeOption.disable_paddle_trt_ops` will be deprecated in v1.2.0, please use `RuntimeOption.paddle_infer_option.disable_trt_ops()` instead."
)
self._option.disable_trt_ops(ops)
def use_ipu(self,
device_num=1,
@@ -593,6 +627,14 @@ class RuntimeOption:
"""
return self._option.trt_option
@property
def paddle_infer_option(self):
"""Get PaddleBackendOption object to configure Paddle Inference backend
:return PaddleBackendOption
"""
return self._option.paddle_infer_option
def enable_profiling(self, inclue_h2d_d2h=False, repeat=100, warmup=50):
"""Set the profile mode as 'true'.
:param inclue_h2d_d2h Whether to include time of H2D_D2H for time of runtime.