Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Jul 9, 2018

• Update README.md

  nihui

  authored
  20c0794

  Copy full SHA for 20c0794

• Revert "optimize interleave section for load first, about 5%~10% speed gain"

  Show description for 4b8101e
  nihui

  committed
  4b8101e

  Copy full SHA for 4b8101e

Commits on Jul 7, 2018

• sgemm is always faster on common channel size

  nihui

  authored
  56a6674

  Copy full SHA for 56a6674

• optimize interleave section for load first, about 5%~10% speed gain

  nihui

  committed
  1e4eaee

  Copy full SHA for 1e4eaee

Commits on Jul 3, 2018

• Fixed two typos (Tencent#466)

  

  quxiaofeng

  authored and

  nihui

  committed
  d0cad77

  Copy full SHA for d0cad77

Commits on Jul 2, 2018

• single vldm is faster than two vld1 on armv7, and some pipeline optimize

  nihui

  committed
  6895cbf

  Copy full SHA for 6895cbf

• reorder kernel weight, pipeline friendly ;)

  nihui

  committed
  05d7562

  Copy full SHA for 05d7562

• add mobilenet-yolo

  nihui

  committed
  0bbdbf4

  Copy full SHA for 0bbdbf4

• fix yolo preprocess, comment about mobilenet-yolo

  nihui

  committed
  543d764

  Copy full SHA for 543d764

Commits on Jun 30, 2018

• -x

  nihui

  committed
  5c6ef31

  Copy full SHA for 5c6ef31

• add yolov2 example

  nihui

  committed
  eb089c0

  Copy full SHA for eb089c0

• fix debug build

  nihui

  committed
  a94e5ad

  Copy full SHA for a94e5ad

• convert BN ReLU6 Reorg YoloDetectionOutput Embed LSTM

  nihui

  committed
  0b6791e

  Copy full SHA for 0b6791e

• implement reorg yolodetectionoutput layer from caffe-yolov2

  nihui

  committed
  b8f4f02

  Copy full SHA for b8f4f02

Commits on Jun 27, 2018

• fix error C2059 and C2589 when using std::min and std::max. (Tencent#456)

  

  kalcohol

  authored and

  nihui

  committed
  8491f2b

  Copy full SHA for 8491f2b

Commits on Jun 26, 2018

• Update README.md (Tencent#452)

  

  BUG1989

  authored and

  nihui

  committed
  b3965e2

  Copy full SHA for b3965e2

Commits on Jun 25, 2018

• proper first row/col handling in resize family, fix Tencent#429

  nihui

  committed
  ee98817

  Copy full SHA for ee98817

• optional image pixel api, fix Tencent#434

  nihui

  committed
  511baa6

  Copy full SHA for 511baa6

• update qcom810 iphone5s benchmark result

  nihui

  authored
  74d1c14

  Copy full SHA for 74d1c14

Commits on Jun 24, 2018

• update qcom410 i.mx7 benchmark result

  nihui

  authored
  8275a08

  Copy full SHA for 8275a08

Commits on Jun 21, 2018

• more explicit alignment on armv7

  nihui

  committed
  2368d29

  Copy full SHA for 2368d29

• direct assembly port, enable convolution 1x1 sgemm on armv7

  nihui

  committed
  d172a34

  Copy full SHA for d172a34

Commits on Jun 20, 2018

• openmp++

  nihui

  committed
  b3e24ca

  Copy full SHA for b3e24ca

• sgemm convolution 1x1 wip, about 20%~75% faster on aarch64, while armv7 compiler is foolish qaq

  nihui

  committed
  0fdb8da

  Copy full SHA for 0fdb8da

• drop armv7 vaddvq_f32 hack

  nihui

  committed
  2b20bf9

  Copy full SHA for 2b20bf9

Commits on Jun 16, 2018

• switch to winograd5

  nihui

  committed
  72bb261

  Copy full SHA for 72bb261

Commits on Jun 15, 2018

• fix concat on height

  nihui

  committed
  a234e92

  Copy full SHA for a234e92

• convert caffe crop layer with three offset, fix Tencent#165

  nihui

  committed
  588487a

  Copy full SHA for 588487a

• crop on channel and crop by param

  nihui

  committed
  003873c

  Copy full SHA for 003873c

• Update README.md

  nihui

  authored
  184cea1

  Copy full SHA for 184cea1

Commits on Jun 11, 2018

• fix parsing inputs list in multiple lines

  nihui

  committed
  fd9ef57

  Copy full SHA for fd9ef57

Commits on Jun 8, 2018

• fix proposal roi_score_blob bug (Tencent#430)

  

  donch1989

  authored and

  nihui

  committed
  dc2a689

  Copy full SHA for dc2a689

Commits on Jun 7, 2018

• allocate after permute, reduce peak memory usage

  nihui

  committed
  99a343c

  Copy full SHA for 99a343c

• load sub-op in advance for group convolution

  nihui

  committed
  0ce0c11

  Copy full SHA for 0ce0c11

• arm neon assembly for winograd5

  nihui

  committed
  86f4264

  Copy full SHA for 86f4264

Pagination

Previous

Next