Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Dec 29, 2017

• build: bumped version to v0.12

  vpirogov

  committed
  ede1e33

  Copy full SHA for ede1e33

• build: bumped small libraries version to 2018.0.1.20181227

  vpirogov

  committed
  ee083b8

  Copy full SHA for ee083b8

• cpu: conv: 4fma: fix a threading bug

  

  Roman Dubtsov

  committed
  d19425b

  Copy full SHA for d19425b

• cpu: avx512_core: fix performance degradation for mb=1

  ankalinin

  committed
  f0d6ba0

  Copy full SHA for f0d6ba0

Commits on Dec 28, 2017

• benchdnn: do not use variadic arrays

  Show description for f20dc84
  

  Fomenko, Evarist M

  committed
  f20dc84

  Copy full SHA for f20dc84

• common: avoid implicit int-to-float conversion

  Show description for 0713991
  

  Fomenko, Evarist M

  committed
  0713991

  Copy full SHA for 0713991

• benchdnn: inputs: use oscale with exact representation

  

  Fomenko, Evarist M

  committed
  f686259

  Copy full SHA for f686259

• cpu: conv: add mkl gemmm for int8

  

  Fomenko, Evarist M

  committed
  5ec8f1d

  Copy full SHA for 5ec8f1d

• cpu: conv-via-gemm: split ws preparation functions

  

  Fomenko, Evarist M

  committed
  765c61c

  Copy full SHA for 765c61c

• api: add hwigo (grouped hwio) format

  Show description for e9fb128
  

  Fomenko, Evarist M

  committed
  e9fb128

  Copy full SHA for e9fb128

• common: nstl: add is_same type traits

  

  Fomenko, Evarist M

  committed
  d815ee2

  Copy full SHA for d815ee2

• common: math: well-defined saturate for f32

  

  Fomenko, Evarist M

  committed
  fec817c

  Copy full SHA for fec817c

• benchdnn: make ref code slightly faster

  

  Fomenko, Evarist M

  committed
  6dc8f17

  Copy full SHA for 6dc8f17

Commits on Dec 27, 2017

• tests: reduced amount of testing

  tprimak

  committed
  34df3eb

  Copy full SHA for 34df3eb

• cpu: avx512_core: 1x1 convolutions: don't use explicit broadcast for inference

  ankalinin

  committed
  958b10b

  Copy full SHA for 958b10b

Commits on Dec 26, 2017

• doc: Updated support matrix with the latest data, added clarification on compatible architectures

  vpirogov

  committed
  e0bfcaa

  Copy full SHA for e0bfcaa

Commits on Dec 22, 2017

• doc: corrected linking instructions and added explicit warning on mixing OpenMP runtimes.

  vpirogov

  committed
  e0a8ed4

  Copy full SHA for e0a8ed4

• doc: removed int8 functionality from the list of experimental features

  vpirogov

  committed
  76d940a

  Copy full SHA for 76d940a

• doc: updated the list of supported topologies

  vpirogov

  committed
  b146790

  Copy full SHA for b146790

Commits on Dec 21, 2017

• style: consistent use of trailing underscore in jit_1x1_conv_conf_t

  ankalinin

  committed
  73a4f28

  Copy full SHA for 73a4f28

• cpu: avx512_core: 1x1 fwd convolutions optimization

  ankalinin

  committed
  2137049

  Copy full SHA for 2137049

Commits on Dec 20, 2017

• common: replace rintf with nearbyintf

  

  Fomenko, Evarist M

  committed
  0d3c46f

  Copy full SHA for 0d3c46f

• cpu: reorder: add non-trivial scale mask support

  Show description for 598e821
  

  Fomenko, Evarist M

  committed
  598e821

  Copy full SHA for 598e821

• cpu: reorder: simplify reference reorder

  

  Fomenko, Evarist M

  committed
  d361be9

  Copy full SHA for d361be9

• cpu: reorder: per implementation attribute check

  

  Fomenko, Evarist M

  committed
  53b06f8

  Copy full SHA for 53b06f8

• src: cpu: use common memory page size

  igorsafo

  committed
  912031d

  Copy full SHA for 912031d

• src: cpu: gemm: use heap if not enough stack

  igorsafo

  committed
  a2d76fc

  Copy full SHA for a2d76fc

Commits on Dec 19, 2017

• cpu: avx512_core: 1x1 bwd_w convolutions optimization

  ankalinin

  committed
  6695424

  Copy full SHA for 6695424

Commits on Dec 14, 2017

• fix: changed bit to logical operand

  Show description for 82cf37f
  msotoflo

  committed
  82cf37f

  Copy full SHA for 82cf37f

Commits on Dec 13, 2017

• c++ api: bnorm: support attributes

  

  Fomenko, Evarist M

  committed
  d8783d3

  Copy full SHA for d8783d3

• fix: cpu: uninitialized variable error and missing 'break'

  Show description for acfabc4
  msotoflo

  committed
  acfabc4

  Copy full SHA for acfabc4

Commits on Dec 12, 2017

• cpu: gemm: added postops feature

  Show description for 55d03b3
  msotoflo

  committed
  55d03b3

  Copy full SHA for 55d03b3

• common: define acc data_type for s32 input/output

  

  Fomenko, Evarist M

  committed
  dac62bc

  Copy full SHA for dac62bc

• cpu: gemm: stop usign std::to_string

  Show description for 353d4d4
  

  Fomenko, Evarist M

  committed
  353d4d4

  Copy full SHA for 353d4d4

Commits on Dec 11, 2017

• cpu: pool: avx512_core: add s32 nhwc pooling

  

  Fomenko, Evarist M

  committed
  e7e1499

  Copy full SHA for e7e1499

Pagination

Previous

Next