Commits

Breadcrumbs

History for

onmaster

User selector

All users

Datepicker

All time

Commit History

Commits on May 11, 2019

• fuse sigmoid

  nihui

  committed
  3e003ff

  Copy full SHA for 3e003ff

Commits on May 10, 2019

• 1x1s1d1_lds_4_4_4 is non-optimal, delete it

  nihui

  committed
  5adfa29

  Copy full SHA for 5adfa29

• mat4 type in shared memory makes some driver unhappy ..

  nihui

  committed
  8ac300c

  Copy full SHA for 8ac300c

• lds optimize for conv3x3s1, conv1x1s1 and fc

  nihui

  committed
  f5ba97e

  Copy full SHA for f5ba97e

• set fixed local size

  nihui

  committed
  8322a14

  Copy full SHA for 8322a14

Commits on May 8, 2019

• move vulkan code to subdir, new layer interface create_pipeline and destroy_pipeline for post-loading works

  nihui

  committed
  7a8f68a

  Copy full SHA for 7a8f68a

Commits on Apr 29, 2019

• fuse deconv/innerproduct relu arm

  nihui

  committed
  c6e075c

  Copy full SHA for c6e075c

Commits on Apr 26, 2019

• fix build on msvc

  nihui

  committed
  be81ecf

  Copy full SHA for be81ecf

• gpu convolution/deconvolution/innerproduct fuse activation

  nihui

  committed
  528fe8e

  Copy full SHA for 528fe8e

• fuse relu leakyrelu clip into convolution/deconvolution/innerproduct

  nihui

  committed
  3f85caf

  Copy full SHA for 3f85caf

Commits on Apr 24, 2019

• ncnnoptimize tool

  nihui

  committed
  7984ffc

  Copy full SHA for 7984ffc

• get rid of the old workaround :)

  nihui

  committed
  b81e1f3

  Copy full SHA for b81e1f3

• mxnet2ncnn tool support symbol.softmax op (Tencent#938)

  Show description for e2e8e1b
  

  791136190

  authored and

  nihui

  committed
  e2e8e1b

  Copy full SHA for e2e8e1b

• add missing barrier for transfer dst, fix softmax pack4, fix Tencent#932

  nihui

  committed
  5d86014

  Copy full SHA for 5d86014

Commits on Apr 23, 2019

• bottom blob memory never alias, reuse blob memory more elegantly relying on refcount

  nihui

  committed
  4729ea3

  Copy full SHA for 4729ea3

Commits on Apr 21, 2019

• convolution padding same on gpu

  nihui

  committed
  274392e

  Copy full SHA for 274392e

• Optimization of the softmax layer (Tencent#914)

  Show description for 9ffe2b8
  

  gfjiangly

  authored and

  nihui

  committed
  9ffe2b8

  Copy full SHA for 9ffe2b8

• bind wait barrier count member to memory, fix Tencent#932

  nihui

  committed
  8724440

  Copy full SHA for 8724440

Commits on Apr 17, 2019

• fix lrn shader, fix Tencent#928

  nihui

  committed
  a0561e3

  Copy full SHA for a0561e3

Commits on Apr 16, 2019

• fix build with vulkan header version 65, fix Tencent#907

  nihui

  committed
  3d06c40

  Copy full SHA for 3d06c40

Commits on Apr 14, 2019

• compiled spirv nearly always claim uniform buffer 8bit / 16bit access capability

  nihui

  committed
  f92dcca

  Copy full SHA for f92dcca

Commits on Apr 12, 2019

• add compile shader module function, create pipeline from custom shader spv data

  nihui

  committed
  c180e87

  Copy full SHA for c180e87

Commits on Apr 10, 2019

• try posix_memalign first

  nihui

  committed
  de70714

  Copy full SHA for de70714

Commits on Apr 7, 2019

• fix fp16s fp16a deconvolution shader

  nihui

  committed
  9643916

  Copy full SHA for 9643916

Commits on Apr 6, 2019

• dynamic padding and crop offset size

  nihui

  committed
  c5ab0c8

  Copy full SHA for c5ab0c8

Commits on Apr 5, 2019

• interp share coeffs and xofs

  nihui

  committed
  c70b136

  Copy full SHA for c70b136

Commits on Apr 3, 2019

• interp bicubic shader, initialize mat member with zero

  nihui

  committed
  31db979

  Copy full SHA for 31db979

• add int8 winograd F(4,3) with neon assembly optimization (Tencent#891)

  Show description for 93a34a8
  

  BUG1989

  authored and

  nihui

  committed
  93a34a8

  Copy full SHA for 93a34a8

• use platform aligned malloc

  nihui

  committed
  9e9ae23

  Copy full SHA for 9e9ae23

Commits on Apr 2, 2019

• fix dst write out of range, fix Tencent#886

  nihui

  committed
  1634675

  Copy full SHA for 1634675

Commits on Mar 29, 2019

• fix NaN(var maybe minus due to accuracy sometimes) issue in InstanceNorm (Tencent#874)

  

  liurs1990

  authored and

  nihui

  committed
  1554438

  Copy full SHA for 1554438

• resize bicubic

  nihui

  committed
  dfffb29

  Copy full SHA for dfffb29

Commits on Mar 28, 2019

• reuse hresize result properly when enlarging, fix Tencent#863

  nihui

  committed
  c778265

  Copy full SHA for c778265

Commits on Mar 27, 2019

• move copy cut border function to operator

  nihui

  committed
  a4b74d2

  Copy full SHA for a4b74d2

• implement substract_mean_normalize with bias and scale op

  nihui

  committed
  5a905c7

  Copy full SHA for 5a905c7

Pagination

Previous

Next