Cronologia Commit

Autore SHA1 Messaggio Data
  gaotingquan 976eeb533f when pir trt, it's need that enabling use gpu in convert_trt 8 mesi fa
  EnflameGCU 0599a222ed [GCU] Update gcu config (#3605) 8 mesi fa
  gaotingquan 19863acc7f bugfix 8 mesi fa
  gaotingquan db1bbfe948 support to set more trt config by model 8 mesi fa
  gaotingquan d3dc02cb7a support to set trt config by model 8 mesi fa
  gaotingquan 774737c06b support paddle_fp32 and paddle_fp16 mode 8 mesi fa
  zhang-prog eb3e60685b add cache func (#3509) 8 mesi fa
  gaotingquan 5abc0bfdca set trt config for PP-YOLOE_seg-S 8 mesi fa
  gaotingquan bf87804450 support trt_dynamic_shape_input_data for pir trt 8 mesi fa
  gaotingquan 0742a15ef8 Manually delete the predictor to trigger its destructor, ensuring that the shape_range_info file would be saved 8 mesi fa
  Lin Manhui bea5ad68f9 [Fix] Allow bypassing model white list checks when performing internal tests (#3499) 8 mesi fa
  gaotingquan ded6b3ab2b bugfix: call enable_use_gpu() before calling enable_tensorrt_engine() 8 mesi fa
  Lin Manhui 8412aa82cb Circumvent paddle inference bug (#3490) 8 mesi fa
  zhang-prog 3c9400e6dd Enhance benchmark (#3475) 8 mesi fa
  zhang-prog 5158085f23 [Feat] upgrade benchmark (#3416) 8 mesi fa
  gaotingquan 9a4a934d47 support set trt config by model 8 mesi fa
  a31413510 7b5288bd6f predict 8 mesi fa
  gaotingquan 68f3ff5b6f fix redundantly create predictor 9 mesi fa
  gaotingquan f48f8e4fdf call enable_gpu() on dcu device 9 mesi fa
  gaotingquan a5b997876d use dcu device type 9 mesi fa
  gaotingquan fcbf6b11d2 dcu also use gpu as the flag 9 mesi fa
  gaotingquan 8e9b8ddca2 change workspace_size to 30 in trt infer to be compatible with more models 10 mesi fa
  gaotingquan adb12d98b6 update trt 10 mesi fa
  gaotingquan e0fcc1ff5f support PIR TRT 10 mesi fa
  gaotingquan f1ba03b2e9 update Paddle Predictor 10 mesi fa
  Tingquan Gao 5fb83f71f3 deprecate NEW_PREDICTOR & USE_NEW_INFERENCE flags (#2932) 9 mesi fa