Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Apr 26, 2024

• Fix GEMM config for xnn_qd8_f16_qc8w_igemm_minmax_ukernel_4x16c4__neondotfp16arith

  Show description for 410da19
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  410da19

  Copy full SHA for 410da19

• Rdsum microkernels are accumlating

  Show description for 57e0562
  

  alankelly

  authored and

  xnnpack-bot

  committed
  57e0562

  Copy full SHA for 57e0562

• Add SSE rdsum microkernels

  Show description for 26b30aa
  

  alankelly

  authored and

  xnnpack-bot

  committed
  26b30aa

  Copy full SHA for 26b30aa

• Fix GEMM config for xnn_qd8_f16_qc8w_gemm_minmax_ukernel_2x8c2s4__neonfp16arith

  Show description for 39054b1
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  39054b1

  Copy full SHA for 39054b1

Commits on Apr 25, 2024

• Fix GEMM config for xnn_qd8_f16_qc8w_igemm_minmax_ukernel_4x16c8__neoni8mm

  Show description for e0a0d2f
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  e0a0d2f

  Copy full SHA for e0a0d2f

• Fix GEMM cnofig for xnn_qd8_f16_qc8w_igemm_minmax_ukernel_4x16c4__neondotfp16arith

  Show description for dd70d58
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  dd70d58

  Copy full SHA for dd70d58

• Enable F16-F32ACC-RSUM AVX512SKX microkernel for sum of FP16 values

  Show description for 1563cbb
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  1563cbb

  Copy full SHA for 1563cbb

• Fix GEMM cnofig for xnn_qd8_f32_qc8w_gemm_minmax_ukernel_2x8c2s4__neon_mlal

  Show description for 767b701
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  767b701

  Copy full SHA for 767b701

• Introduce TransposeConv with dynamic range quantization Subgraph API

  Show description for 507353a
  

  ablavatski

  authored and

  xnnpack-bot

  committed
  507353a

  Copy full SHA for 507353a

• Introduce QD8 TranposeConv operator

  Show description for d8e2747
  

  ablavatski

  authored and

  xnnpack-bot

  committed
  d8e2747

  Copy full SHA for d8e2747

• RDSum microkernels are no longer minmax and have their own tester.

  Show description for c21626e
  

  alankelly

  authored and

  xnnpack-bot

  committed
  c21626e

  Copy full SHA for c21626e

• Clean-up rdsum benches

  Show description for 6bccb6c
  

  alankelly

  authored and

  xnnpack-bot

  committed
  6bccb6c

  Copy full SHA for 6bccb6c

• Internal config change

  Show description for 64f880b
  

  alankelly

  authored and

  xnnpack-bot

  committed
  64f880b

  Copy full SHA for 64f880b

Commits on Apr 24, 2024

• Merge pull request #6326 from ejparkqc:master

  Show description for 78f8763
  xnnpack-bot

  committed
  78f8763

  Copy full SHA for 78f8763

• Merge pull request #6267 from mcr229:fc_qcint32_bias

  Show description for 8637b70
  xnnpack-bot

  committed
  8637b70

  Copy full SHA for 8637b70

• F16-F32ACC-RSUM AVX512SKX microkernel for sum of FP16 values

  Show description for cae3022
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  cae3022

  Copy full SHA for cae3022

Commits on Apr 23, 2024

• Improve GEMM unittest performance

  Show description for 0ce75bc
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  0ce75bc

  Copy full SHA for 0ce75bc

• Change one of the flags for hexagon-sim to use v68

  ejparkqc

  committed
  bc07173

  Copy full SHA for bc07173

• F16-RMAX scalar optimized keep max sign complement

  Show description for 114859a
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  114859a

  Copy full SHA for 114859a

• Removing build script using Hexagon v66

  ejparkqc

  committed
  d30d13f

  Copy full SHA for d30d13f

• Change Hexagon minimum supported version to v68 (from v66)

  ejparkqc

  committed
  71871aa

  Copy full SHA for 71871aa

• Add f32 rsum discontig benchmarks

  Show description for e7889d3
  

  alankelly

  authored and

  xnnpack-bot

  committed
  e7889d3

  Copy full SHA for e7889d3

• Add f32 rsum discontig neon microkernels

  Show description for 8d82c68
  

  alankelly

  authored and

  xnnpack-bot

  committed
  8d82c68

  Copy full SHA for 8d82c68

• Add rsum discontiguous ukernels.

  Show description for ecf49b2
  

  alankelly

  authored and

  xnnpack-bot

  committed
  ecf49b2

  Copy full SHA for ecf49b2

• F16-RMAX - enable rmax F16C microkernel for F16C instead of AVX2

  Show description for 171473f
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  171473f

  Copy full SHA for 171473f

• Fix math.h spelling - compliment changed to complement

  Show description for a23a174
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  a23a174

  Copy full SHA for a23a174

Commits on Apr 22, 2024

• F16-RMAX - enable rmax scalar for all platforms

  Show description for 4a19a43
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  4a19a43

  Copy full SHA for 4a19a43

• F16-RMINMAX - move math_min_f16 and math_max_f16 to math.h

  Show description for 81e1854
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  81e1854

  Copy full SHA for 81e1854

• F16-RMAX benchmark include f16c_u32 for AVX

  Show description for 6617e54
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  6617e54

  Copy full SHA for 6617e54

• Mean op can handle arbitrary reduction axis in the contiguous axes.

  Show description for a39a16d
  

  alankelly

  authored and

  xnnpack-bot

  committed
  a39a16d

  Copy full SHA for a39a16d

• Fix missing `#include`s in the `XNNPACK/src` subdirectory.

  Show description for 2efa5b9
  

  gonnet

  authored and

  xnnpack-bot

  committed
  2efa5b9

  Copy full SHA for 2efa5b9

• Remove printf that were used during debugging reduce and resize

  Show description for ba59006
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  ba59006

  Copy full SHA for ba59006

Commits on Apr 21, 2024

• GEMM unittest step thru NC by NextPrime

  Show description for 3f3175e
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  3f3175e

  Copy full SHA for 3f3175e

Commits on Apr 20, 2024

• Step thru k-block test range using prime numbers

  Show description for 273714b
  

  fbarchard

  authored and

  xnnpack-bot

  committed
  273714b

  Copy full SHA for 273714b

Commits on Apr 19, 2024

• Add iterative `vsqrt` microkernels for `x86_64`, which computes `x*rsqrt(x)`, i.e.

  Show description for 8567945
  

  gonnet

  authored and

  xnnpack-bot

  committed
  8567945

  Copy full SHA for 8567945

Pagination

Previous

Next