Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Mar 9, 2021

• Use std::array for fixed-sized arrays in hardcoded models

  Show description for 52e061d
  

  Maratyszcza

  authored and

  xnnpack-bot

  committed
  52e061d

  Copy full SHA for 52e061d

Commits on Mar 8, 2021

• Round KC up for assembly microkernels.

  Show description for b7941cb
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  b7941cb

  Copy full SHA for b7941cb

• Enable QS8 IGEMM for Cortex A55

  Show description for b75840f
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  b75840f

  Copy full SHA for b75840f

• QS8 IGEMM for Cortex A55

  Show description for 89e12f8
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  89e12f8

  Copy full SHA for 89e12f8

• Benchmark randomized QU8 MobileNet v1 model

  Show description for 12a23bb
  

  Maratyszcza

  authored and

  xnnpack-bot

  committed
  12a23bb

  Copy full SHA for 12a23bb

Commits on Mar 5, 2021

• Increase minimum value to avoid f16_vrdivc producing inf

  Show description for fb8d1f1
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  fb8d1f1

  Copy full SHA for fb8d1f1

• Remove 12x8 QS8 GEMM and IGEMM Neon dotproduct microkernels.

  Show description for 62b4ff7
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  62b4ff7

  Copy full SHA for 62b4ff7

• Merge pull request #1327 from simonmaurer:macos_arm64_bazel_final

  Show description for 863d595
  xnnpack-bot

  committed
  863d595

  Copy full SHA for 863d595

Commits on Mar 4, 2021

• Update emscripten config settings to conform to the official emscripten toolchain.

  Show description for f69ac5d
  xnnpack-bot

  committed
  f69ac5d

  Copy full SHA for f69ac5d

Commits on Mar 3, 2021

• updated pthreadpool to track latest master branch, using commit b96413b10dd8edaa4f6c0a60c6cf5ef55eebeef78164d5d69294c8173457f0ec

  simonmaurer

  committed
  4d2d3e5

  Copy full SHA for 4d2d3e5

• regrouped macos_arm64 config with other macos configs

  simonmaurer

  committed
  ae33ab8

  Copy full SHA for ae33ab8

• fix: remove uncommented flag for macos_arm64

  simonmaurer

  committed
  62f0c6b

  Copy full SHA for 62f0c6b

• added aarch64 deps for macos_arm64

  simonmaurer

  committed
  3b403c2

  Copy full SHA for 3b403c2

• added config macos_arm64 for build files BUILD.bazel and cpuinfo.BUILD

  simonmaurer

  committed
  2de19cd

  Copy full SHA for 2de19cd

• .bazelrc: added config macos_arm64

  simonmaurer

  committed
  b5b3c31

  Copy full SHA for b5b3c31

• QS8 enable 4x8c4__neondot for ARM32

  Show description for fb0ab0b
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  fb0ab0b

  Copy full SHA for fb0ab0b

Commits on Mar 2, 2021

• QS8 C8 Neon microkernels with MUL and MLA versions.

  Show description for da78da1
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  da78da1

  Copy full SHA for da78da1

• QS8 Neon dot product intrinsics GEMM and IGEMM microkernels reduced remainder code.

  Show description for 618d85d
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  618d85d

  Copy full SHA for 618d85d

• Re-label branch targets in c4-neondot assembly QS8 GEMM microkernels.

  Show description for d76a37b
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  d76a37b

  Copy full SHA for d76a37b

Commits on Mar 1, 2021

• QS8 1x16c4 ld32 GEMM microkernel using NEON dot product

  Show description for 4a4be4e
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  4a4be4e

  Copy full SHA for 4a4be4e

• QS8 Cortex A55 GEMM microkernel bump kc to be a multiple of channels.

  Show description for 7aa4bfd
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  7aa4bfd

  Copy full SHA for 7aa4bfd

• Quantized GEMM/IGEMM microkernels bump kc to be a multiple of channels.

  Show description for 6d8ca7d
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  6d8ca7d

  Copy full SHA for 6d8ca7d

Commits on Feb 26, 2021

• QS8 Neon IGEMM microkernels with 8 bit MUL using DUP

  Show description for 02121ca
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  02121ca

  Copy full SHA for 02121ca

• QS8 LD64 C4 dot product GEMM microkernel reduced remainder handling

  Show description for 8f6a1ed
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  8f6a1ed

  Copy full SHA for 8f6a1ed

• QS8 C16 GEMM microkernel source renamed from mull to mlal

  Show description for fd1dee7
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  fd1dee7

  Copy full SHA for fd1dee7

• Fix mismatch in block layout in mixed-layout Depth-To-Space operator

  Show description for 77e93a2
  

  Maratyszcza

  authored and

  xnnpack-bot

  committed
  77e93a2

  Copy full SHA for 77e93a2

• QS8 LD32 GEMM microkernel for big cores with dotproduct

  Show description for a5e242c
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  a5e242c

  Copy full SHA for a5e242c

Commits on Feb 25, 2021

• C8 MLA Neon GEMM/IGEMM microkernels count k down from kc.

  Show description for 01c341b
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  01c341b

  Copy full SHA for 01c341b

Commits on Feb 23, 2021

• Enable Quantized C2 microkernel for Neon

  Show description for a414daa
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  a414daa

  Copy full SHA for a414daa

• QS8 Neon IGEMM C16 microkernel with two 8 bit multiplies and vpadal to accumulate.

  Show description for 36f95cf
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  36f95cf

  Copy full SHA for 36f95cf

Commits on Feb 22, 2021

• Fix bug in NHWC Convolution with depthwise kernels

  Show description for 55d53a4
  

  Maratyszcza

  authored and

  xnnpack-bot

  committed
  55d53a4

  Copy full SHA for 55d53a4

Commits on Feb 19, 2021

• QS8 Neon GEMM C16 microkernel with two 8 bit multiplies and vpadal to accumulate.

  Show description for 71c4d1a
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  71c4d1a

  Copy full SHA for 71c4d1a

• Add A55 microkernel to QS8 E2E benchmark.

  Show description for a6cba1d
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  a6cba1d

  Copy full SHA for a6cba1d

Commits on Feb 16, 2021

• Remove scalar C4 QS8 and QU8 gemm microkernels.

  Show description for 6d138db
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  6d138db

  Copy full SHA for 6d138db

• QS8 C8 Neon remove remainder handling code and rewind the A pointers by kc

  Show description for a0fe11d
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  a0fe11d

  Copy full SHA for a0fe11d

Pagination

Previous

Next