Commits

Branch selector

 master

User selector

qjia7

Datepicker

All time

Commit History

Commits on Sep 16, 2022

• webgpu: Fix wgsl failure from MatMulSplitKProgram (#6838)

  Show description for 52d30d6
  qjia7

  authored
  52d30d6

  Copy full SHA for 52d30d6

Commits on Sep 15, 2022

• webgpu: Expand DepthwiseConv2DVec4Program to support any stride (#6820)

  Show description for 7423ee7
  

  qjia7

  and

  Linchenn

  authored
  7423ee7

  Copy full SHA for 7423ee7

Commits on Aug 31, 2022

• webgpu: Change workPerThread to 1 for addn, concat, transpose (#6802)

  Show description for cdc8ce1
  qjia7

  authored
  cdc8ce1

  Copy full SHA for cdc8ce1

Commits on Aug 26, 2022

• webgpu: tighten the condition of Conv2DDerInputProgram (#6797)

  qjia7

  authored
  d931cc7

  Copy full SHA for d931cc7

• webgpu: Tune the condition of Conv2DDerInputProgram (#6790)

  Show description for 9581a31
  qjia7

  authored
  9581a31

  Copy full SHA for 9581a31

Commits on Aug 24, 2022

• webgpu: Tune the condition of shared binary (#6783)

  Show description for 720499c
  qjia7

  authored
  720499c

  Copy full SHA for 720499c

Commits on Aug 23, 2022

• webgpu: Add non-shared argminmax program (#6778)

  Show description for 25cd4f4
  

  qjia7

  and

  Linchenn

  authored
  25cd4f4

  Copy full SHA for 25cd4f4

Commits on Aug 22, 2022

• webgpu: Optimize AvgPool when filter size = input size (#6762)

  Show description for cf328d3
  qjia7

  authored
  cf328d3

  Copy full SHA for cf328d3

• Add cpu forward for Cast (#6760)

  Show description for 0557e78
  

  qjia7

  and

  Linchenn

  authored
  0557e78

  Copy full SHA for 0557e78

Commits on Aug 17, 2022

• webgpu: enlarge the splitted dimInner size (#6755)

  qjia7

  authored
  a366cc2

  Copy full SHA for a366cc2

Commits on Aug 12, 2022

• webgpu: Merge MatMulPackedProgram and MatMulPackedVec4Program (#6688)

  Show description for a1e5aea
  qjia7

  authored
  a1e5aea

  Copy full SHA for a1e5aea

Commits on Jul 21, 2022

• webgpu: Refactor matMul read/write method (#6623)

  Show description for 3468a34
  qjia7

  authored
  3468a34

  Copy full SHA for 3468a34

Commits on Jul 14, 2022

• webgpu: Fix unknown identifier: 'coords' (#6622)

  qjia7

  authored
  37130cf

  Copy full SHA for 37130cf

Commits on Jul 8, 2022

• webgpu: Optimize FromPixelsProgram (#6614)

  Show description for 7a3a58e
  qjia7

  authored
  7a3a58e

  Copy full SHA for 7a3a58e

Commits on Jul 1, 2022

• webgpu: Move bias activation to a common part (#6579)

  qjia7

  authored
  18127e9

  Copy full SHA for 18127e9

Commits on Jun 29, 2022

• webgpu: matmul split k and reduce (#6285)

  Show description for f4c5554
  qjia7

  authored
  f4c5554

  Copy full SHA for f4c5554

Commits on Jun 17, 2022

• webgpu: Extend DepthwiseConv2D3x3Program to support any filter size (#6521)

  Show description for 685ba0c
  qjia7

  authored
  685ba0c

  Copy full SHA for 685ba0c

Commits on Jun 16, 2022

• webgpu: Relax the condition of MatMulReduceProgram (#6539)

  Show description for b30fc73
  qjia7

  authored
  b30fc73

  Copy full SHA for b30fc73

Commits on Jun 14, 2022

• webgpu: Fix the incorrect result of pain binary (#6534)

  qjia7

  authored
  bceaf5c

  Copy full SHA for bceaf5c

Commits on Jun 13, 2022

• webgpu: Support depthwise conv2d with nchw format (#6084)

  qjia7

  authored
  bc1059b

  Copy full SHA for bc1059b

Commits on Jun 8, 2022

• webgpu: clean up conv2d (#6500)

  qjia7

  authored
  9d655f9

  Copy full SHA for 9d655f9

Commits on Jun 2, 2022

• webgpu: Support nchw conv2d vec4 (#6446)

  Show description for add62a3
  qjia7

  authored
  add62a3

  Copy full SHA for add62a3

• webgpu: Fix the breaking of atomicCompareExchangeWeak (#6484)

  Show description for c7fdb43
  qjia7

  authored
  c7fdb43

  Copy full SHA for c7fdb43

Commits on May 27, 2022

• webgpu: Enable NCHW fusedConv2d (#6460)

  qjia7

  authored
  a7cb36a

  Copy full SHA for a7cb36a

Commits on May 24, 2022

• webgpu: Optimize nchw conv2d (#6411)

  qjia7

  authored
  b53bd47

  Copy full SHA for b53bd47

Commits on May 20, 2022

• webgpu: extend Conv2DMMVec4Program to support inChannels=3 (#6394)

  Show description for 980fca3
  qjia7

  authored
  980fca3

  Copy full SHA for 980fca3

Commits on May 7, 2022

• webgpu: Remove deprecated conv2d algorithms (#6375)

  Show description for eb4e751
  qjia7

  authored
  eb4e751

  Copy full SHA for eb4e751

Commits on May 6, 2022

• webgpu: Fix bug in conv2d vec4 (#6374)

  Show description for e8e3fa2
  qjia7

  authored
  e8e3fa2

  Copy full SHA for e8e3fa2

Commits on Apr 29, 2022

• Change makeTensorFromTensorId to makeTensorFromTensorInfo (#6355)

  Show description for 55edb1a
  

  qjia7

  and

  lina128

  authored
  55edb1a

  Copy full SHA for 55edb1a

Commits on Apr 28, 2022

• webgpu: Update README (#6353)

  qjia7

  authored
  9d6f9bc

  Copy full SHA for 9d6f9bc

Commits on Apr 24, 2022

• webgpu: Fix incorrect comparison (#6344)

  qjia7

  authored
  2c3d5ba

  Copy full SHA for 2c3d5ba

Commits on Apr 21, 2022

• webgpu: support NCHW dataFormat for conv2d (#6293)

  Show description for fc590c1
  qjia7

  authored
  fc590c1

  Copy full SHA for fc590c1

Commits on Apr 2, 2022

• webgpu: Fix leakyRelu error for conv2d vec4 (#6292)

  qjia7

  authored
  b42baa7

  Copy full SHA for b42baa7

Commits on Apr 1, 2022

• webgpu: Add Leakyrelu support to fused ops (#6288)

  Show description for 20b04f7
  qjia7

  authored
  20b04f7

  Copy full SHA for 20b04f7

Commits on Mar 30, 2022

• webgpu: Fix DepthwiseConv2D3x3Program bug when dilation > 1 (#6264)

  Show description for 5779cbf
  qjia7

  authored
  5779cbf

  Copy full SHA for 5779cbf

Pagination

Previous

Next