Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Sep 1, 2021

• feat(convolution): add nhwc int8 imma conv and deconv

  Show description for 3179884
  

  megvii-mge

  authored and

  hxdmegvii

  committed
  3179884

  Copy full SHA for 3179884

Commits on Aug 12, 2021

• fix(cutlass): update some interfaces for generating ConvolutionKey

  Show description for 3f3c28e
  

  megvii-mge

  authored and

  hxdmegvii

  committed
  3f3c28e

  Copy full SHA for 3f3c28e

Commits on Jul 19, 2021

• feat(convolution): add remove shared load mode for nhwc and ncxhwx

  Show description for 70ee938
  

  megvii-mge

  authored and

  hxdmegvii

  committed
  70ee938

  Copy full SHA for 70ee938

• feat(convolution): add single stage for imma

  Show description for baee135
  

  megvii-mge

  authored and

  hxdmegvii

  committed
  baee135

  Copy full SHA for baee135

• feat(convolution): add nhwc int4 conv

  Show description for fc45047
  

  megvii-mge

  authored and

  hxdmegvii

  committed
  fc45047

  Copy full SHA for fc45047

Commits on Jun 23, 2021

• fix(cutlass): fix default convolution config for dp4a simt

  Show description for 0397146
  megvii-mge

  committed
  0397146

  Copy full SHA for 0397146

• fix(cutlass/epilogue): fix compile

  Show description for 823b998
  megvii-mge

  committed
  823b998

  Copy full SHA for 823b998

• perf(cutlass/convolution): refactor conv2d tile iterator

  Show description for 2bd7e76
  megvii-mge

  committed
  2bd7e76

  Copy full SHA for 2bd7e76

• feat(cutlass/convolution): add convolution impl with int8 input and int4 output

  Show description for 2fe1197
  megvii-mge

  committed
  2fe1197

  Copy full SHA for 2fe1197

• feat(cutlass): add pack_pad for global_load

  Show description for eafd7f8
  megvii-mge

  committed
  eafd7f8

  Copy full SHA for eafd7f8

• fix(cutlass): merge two thread map template specializations

  Show description for 5b65a7a
  megvii-mge

  committed
  5b65a7a

  Copy full SHA for 5b65a7a

• feat(cutlass): support src_zero_point in uint4 x int4 convolution

  Show description for 6ac6e4a
  megvii-mge

  committed
  6ac6e4a

  Copy full SHA for 6ac6e4a

• feat(cutlass): add uint4 x int4 convolution

  Show description for 9f91e8c
  megvii-mge

  committed
  9f91e8c

  Copy full SHA for 9f91e8c

• feat(cutlass): add int4 nchw64 convolution

  Show description for 48f3c69
  megvii-mge

  committed
  48f3c69

  Copy full SHA for 48f3c69

• fix(cutlass): fix epilogue clamp

  Show description for 29142ac
  megvii-mge

  committed
  29142ac

  Copy full SHA for 29142ac

• fix(cutlass): fix rounding in host convolution reference

  Show description for c9e8a41
  megvii-mge

  committed
  c9e8a41

  Copy full SHA for c9e8a41

• feat(cutlass): add impl for numeric array conversion between int4b_t/uint4b_t and float

  Show description for 61bc238
  megvii-mge

  committed
  61bc238

  Copy full SHA for 61bc238

Commits on Mar 21, 2021

• Project import generated by Copybara.

  Show description for 61ff64e
  megvii-mge

  committed
  61ff64e

  Copy full SHA for 61ff64e

Commits on Sep 25, 2020

• Merge pull request NVIDIA#135 from NVIDIA/cutlass_2.3_final

  Show description for c2b80ad
  d-k-b

  authored
  c2b80ad

  Copy full SHA for c2b80ad

• CUTLASS 2.3.0 final.

  kerrmudgeon

  committed
  37a8f9e

  Copy full SHA for 37a8f9e

Commits on Sep 23, 2020

• CUTLASS 2.3 initial commit (NVIDIA#134)

  Show description for c53f333
  kerrmudgeon

  authored
  c53f333

  Copy full SHA for c53f333

Commits on Jul 13, 2020

• Typoes (NVIDIA#107)

  Show description for 4dac749
  hwu36

  authored
  4dac749

  Copy full SHA for 4dac749

Commits on Jun 17, 2020

• Added examples to enable the unity build (NVIDIA#102)

  Show description for fd7e058
  kerrmudgeon

  authored
  fd7e058

  Copy full SHA for fd7e058

Commits on Jun 15, 2020

• Updated mma_sm80.h to avoid perf penalty due to reinterpret_cast<>. (NVIDIA#100)

  Show description for 1ab1027
  kerrmudgeon

  authored
  1ab1027

  Copy full SHA for 1ab1027

Commits on Jun 8, 2020

• CUTLASS 2.2 (NVIDIA#96)

  Show description for 86931fe
  kerrmudgeon

  authored
  86931fe

  Copy full SHA for 86931fe

Commits on Apr 8, 2020

• update tools/library/CMakeLists to require python 3.6 according to NVIDIA#70 (NVIDIA#82)

  Show description for e33d90b
  thakkarV

  authored
  e33d90b

  Copy full SHA for e33d90b

Commits on Apr 7, 2020

• CUTLASS 2.1 (NVIDIA#83)

  Show description for 96dab34
  kerrmudgeon

  authored
  96dab34

  Copy full SHA for 96dab34

Commits on Nov 22, 2019

• Need Python 3.6 to use enum.auto() (NVIDIA#70)

  kerrmudgeon

  authored
  7c0cd26

  Copy full SHA for 7c0cd26

Commits on Nov 20, 2019

• Removed redundant conjugation operations from matrix_traits. (NVIDIA#65)

  kerrmudgeon

  authored
  45ecbc8

  Copy full SHA for 45ecbc8

• Improved formatting, clarity, and content of several documents. (NVIDIA#64)

  Show description for 8aca98f
  kerrmudgeon

  authored
  8aca98f

  Copy full SHA for 8aca98f

• Clang GPU compilation requires explicit CUDACC version flags (NVIDIA#63)

  

  d-k-b

  authored and

  kerrmudgeon

  committed
  f4d9c8f

  Copy full SHA for f4d9c8f

• CUTLASS 2.0 (NVIDIA#62)

  Show description for fb335f6
  kerrmudgeon

  authored
  fb335f6

  Copy full SHA for fb335f6

Commits on Jul 10, 2019

• Performance enhancement for Volta Tensor Cores TN layout (NVIDIA#53)

  Show description for b5cab17
  kerrmudgeon

  authored
  b5cab17

  Copy full SHA for b5cab17

Commits on May 13, 2019

• Merge pull request NVIDIA#47 from Artem-B/cutlass-1.3-clang

  Show description for eb41735
  TimmyLiu

  authored
  eb41735

  Copy full SHA for eb41735

Commits on May 10, 2019

• Addressed code review comments.

  Artem-B

  committed
  fb8b3a9

  Copy full SHA for fb8b3a9

Pagination

Previous

Next