Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Aug 3, 2018

• Merge branch 'master' of https://github.com/Tencent/ncnn

  nihui

  committed
  9843b9e

  Copy full SHA for 9843b9e

• fix int8_scales from multiple blobs, fix #512

  nihui

  committed
  2bc5049

  Copy full SHA for 2bc5049

Commits on Aug 2, 2018

• Update README.md

  nihui

  authored
  af806a2

  Copy full SHA for af806a2

Commits on Aug 1, 2018

• fix pd using flag condition

  nihui

  committed
  da35291

  Copy full SHA for da35291

• sub-mat shall be allocator-aware

  nihui

  committed
  6b53670

  Copy full SHA for 6b53670

• armv7 int8 quantize/dequantize and conv1x1s1

  nihui

  committed
  e34aa77

  Copy full SHA for e34aa77

• light mode is the default, add mobilenetv2ssdlite example

  nihui

  committed
  55358f6

  Copy full SHA for 55358f6

Commits on Jul 31, 2018

• Create CONTRIBUTING.md

  nihui

  authored
  dbf1c40

  Copy full SHA for dbf1c40

Commits on Jul 30, 2018

• int8 inference on x86

  nihui

  committed
  4be27a0

  Copy full SHA for 4be27a0

Commits on Jul 29, 2018

• autotest never worked, delete it ;)

  nihui

  committed
  6eb6abf

  Copy full SHA for 6eb6abf

• core int8 inference, quantize and dequantize, net using flag, caffe2ncnn reads int8 scale table

  nihui

  committed
  a169cec

  Copy full SHA for a169cec

Commits on Jul 17, 2018

• high resolution timestamp on windows

  nihui

  authored
  b6b90c8

  Copy full SHA for b6b90c8

Commits on Jul 16, 2018

• install allocator.h

  nihui

  authored
  af49e2c

  Copy full SHA for af49e2c

Commits on Jul 15, 2018

• project-wide NOMINMAX on msvc

  nihui

  authored
  ae467fe

  Copy full SHA for ae467fe

• fix build on msvc

  nihui

  authored
  7e1f358

  Copy full SHA for 7e1f358

• implement ncnn blob/workspace allocator, fine-grained per-layer openmp threads control, fix #469

  nihui

  committed
  9706cd1

  Copy full SHA for 9706cd1

Commits on Jul 13, 2018

• sgemm outperform direct conv on large channel

  nihui

  authored
  5879cb4

  Copy full SHA for 5879cb4

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 (#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. (#456)

  

  kalcohol

  authored and

  nihui

  committed
  8491f2b

  Copy full SHA for 8491f2b

Commits on Jun 26, 2018

• Update README.md (#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 #429

  nihui

  committed
  ee98817

  Copy full SHA for ee98817

• optional image pixel api, fix #434

  nihui

  committed
  511baa6

  Copy full SHA for 511baa6

Pagination

Previous

Next