Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Nov 22, 2019

• build: win: fix broken mingw build

  Show description for 1c2e8e9
  

  Fomenko, Evarist M

  committed
  1c2e8e9

  Copy full SHA for 1c2e8e9

• api: c++: add missing const qualifiers

  Show description for 97f2637
  

  Fomenko, Evarist M

  committed
  97f2637

  Copy full SHA for 97f2637

• build: install: use library version in the right place on mac

  Show description for f7743ec
  

  Fomenko, Evarist M

  committed
  f7743ec

  Copy full SHA for f7743ec

• doc: build: mark TBBROOT as an optional argument when building with TBB

  

  Fomenko, Evarist M

  committed
  433e086

  Copy full SHA for 433e086

• build: lnx: enable using system-installed tbb

  Show description for bb4b6f4
  

  Fomenko, Evarist M

  committed
  bb4b6f4

  Copy full SHA for bb4b6f4

• build: check for TBB version on DNNL side

  Show description for 2abbaac
  

  Fomenko, Evarist M

  committed
  2abbaac

  Copy full SHA for 2abbaac

• build: retrieve TBB location from properties

  Show description for 7a13102
  

  Fomenko, Evarist M

  committed
  7a13102

  Copy full SHA for 7a13102

• tests: benchdnn: inputs: conv: regression test for int8_conv fixup

  

  Alok Bakshi

  committed
  5ecf72d

  Copy full SHA for 5ecf72d

• src: cpu: conv_int8: bcast leeway calculation fixup

  

  Alok Bakshi

  committed
  a0f4a19

  Copy full SHA for a0f4a19

• api: add giodhw/acbdef format

  Show description for bd6ff1e
  

  Fomenko, Evarist M

  committed
  bd6ff1e

  Copy full SHA for bd6ff1e

• benchdnn: rnn: make lstm testing more robust

  

  Fomenko, Evarist M

  committed
  dc5588a

  Copy full SHA for dc5588a

• benchdnn: rnn: use named gates

  

  Fomenko, Evarist M

  committed
  6ed385a

  Copy full SHA for 6ed385a

Commits on Nov 21, 2019

• cpu: conv: Remove unneeded skip_kh_loop in bf16 fwd compute_loop

  rjoursler

  committed
  95b88d0

  Copy full SHA for 95b88d0

• cpu: conv: Fix segfault in bf16 fwd_d due to push/pop imbalance

  rjoursler

  committed
  971cdf6

  Copy full SHA for 971cdf6

• doc: rnn: denote the output GRU gate as o instead of c

  

  Fomenko, Evarist M

  committed
  1884722

  Copy full SHA for 1884722

• cpu: gemm: add readable avx2-igemm compute kernel

  

  aaraujom

  authored and

  tprimak

  committed
  35b39a8

  Copy full SHA for 35b39a8

• cpu: gemm: add avx2 igemm optimizations

  Show description for 274be82
  

  aaraujom

  authored and

  tprimak

  committed
  274be82

  Copy full SHA for 274be82

• conv: int8 1d/2d/2d_dw conv support on AVX2

  

  Alok Bakshi

  committed
  c830ef3

  Copy full SHA for c830ef3

Commits on Nov 20, 2019

• build: fix windows gpu build

  nivas-x86

  committed
  4e484f1

  Copy full SHA for 4e484f1

• src: ocl: rnn: add scratchpad support

  skazakov1

  committed
  6108203

  Copy full SHA for 6108203

• fixup: benchdnn: ip: add shapes from maskrcnn

  Show description for 377121c
  kwiersch

  committed
  377121c

  Copy full SHA for 377121c

Commits on Nov 19, 2019

• examples: consistently use float data type

  

  Fomenko, Evarist M

  committed
  23710ae

  Copy full SHA for 23710ae

• src: cpu: 1x1: clean-up: assume pad_r, pad_t are zero

  nivas-x86

  committed
  67d6700

  Copy full SHA for 67d6700

• benchdnn: ip: add shapes from maskrcnn

  kwiersch

  committed
  a95d9c8

  Copy full SHA for a95d9c8

• doc: primitives: matmul: add GPU support

  h-sadia

  committed
  4d1db25

  Copy full SHA for 4d1db25

• test: benchdnn: add GPU test support

  h-sadia

  committed
  681c38e

  Copy full SHA for 681c38e

• test: gtest: matmul: extend tests for GPU

  h-sadia

  committed
  9897e70

  Copy full SHA for 9897e70

• src: ocl: add matmul primitive

  h-sadia

  committed
  279c829

  Copy full SHA for 279c829

• github: added opensource ci

  sakulish

  committed
  b945c43

  Copy full SHA for b945c43

• api: use macro for defining stream::flags operators

  

  Roman Dubtsov

  committed
  417a198

  Copy full SHA for 417a198

• api: clean-up notations for C API vars in dnnl.hpp

  

  Roman Dubtsov

  committed
  45f2712

  Copy full SHA for 45f2712

• api: properly mark private members in the dnnl.hpp

  Show description for 6f61e82
  

  Roman Dubtsov

  committed
  6f61e82

  Copy full SHA for 6f61e82

• api: update doxygen

  

  Roman Dubtsov

  committed
  2f4b807

  Copy full SHA for 2f4b807

• api: c++: make dnnl::engine behave as a regular handle

  Show description for 98320d8
  

  Roman Dubtsov

  committed
  98320d8

  Copy full SHA for 98320d8

Commits on Nov 18, 2019

• cpu: conv: implement threading by input width for bf16 bwd_d

  rjoursler

  committed
  dfcc95c

  Copy full SHA for dfcc95c

Pagination

Previous

Next