 gaotingquan
|
ded6b3ab2b
bugfix: call enable_use_gpu() before calling enable_tensorrt_engine()
|
8 months ago |
 Lin Manhui
|
8412aa82cb
Circumvent paddle inference bug (#3490)
|
8 months ago |
 zhang-prog
|
3c9400e6dd
Enhance benchmark (#3475)
|
8 months ago |
|
zhang-prog
|
5158085f23
[Feat] upgrade benchmark (#3416)
|
8 months ago |
|
gaotingquan
|
9a4a934d47
support set trt config by model
|
9 months ago |
 a31413510
|
7b5288bd6f
predict
|
9 months ago |
|
gaotingquan
|
68f3ff5b6f
fix redundantly create predictor
|
9 months ago |
|
gaotingquan
|
f48f8e4fdf
call enable_gpu() on dcu device
|
9 months ago |
|
gaotingquan
|
a5b997876d
use dcu device type
|
9 months ago |
|
gaotingquan
|
fcbf6b11d2
dcu also use gpu as the flag
|
9 months ago |
|
gaotingquan
|
8e9b8ddca2
change workspace_size to 30 in trt infer to be compatible with more models
|
10 months ago |
|
gaotingquan
|
adb12d98b6
update trt
|
10 months ago |
|
gaotingquan
|
e0fcc1ff5f
support PIR TRT
|
10 months ago |
|
gaotingquan
|
f1ba03b2e9
update Paddle Predictor
|
10 months ago |
|
Tingquan Gao
|
5fb83f71f3
deprecate NEW_PREDICTOR & USE_NEW_INFERENCE flags (#2932)
|
9 months ago |
