Commits

Breadcrumbs

History for

onmaster

User selector

All users

Datepicker

All time

Commit History

Commits on Aug 20, 2018

• quantized padding in convolution, use range sweets

  nihui

  committed
  6f1b0b0

  Copy full SHA for 6f1b0b0

• padding is elemsize aware, copy_make_border is now a padding wrapper

  nihui

  committed
  bf1c58b

  Copy full SHA for bf1c58b

• Mat range for zero-copy slicing reference

  nihui

  committed
  d321d65

  Copy full SHA for d321d65

Commits on Aug 16, 2018

• use max threads as cpu count on openmp capable runtime

  nihui

  committed
  f78e8fd

  Copy full SHA for f78e8fd

Commits on Aug 15, 2018

• fix bus error for int8 storage

  nihui

  committed
  2790f3c

  Copy full SHA for 2790f3c

Commits on Aug 13, 2018

• add armv7 int8 conv3x3s1,using vaddw to replace vadd and vmovl

  

  BUG1989

  authored and

  nihui

  committed
  1b0e334

  Copy full SHA for 1b0e334

Commits on Aug 12, 2018

• restore the old conv3x3s2 as reference, fast dilation convolution fails on striding

  nihui

  committed
  72411b7

  Copy full SHA for 72411b7

• pack weight and more unroll makes improvement, ~20% faster for conv3x3s2

  nihui

  committed
  1f20eb4

  Copy full SHA for 1f20eb4

Commits on Aug 10, 2018

• fix asm "invalid operand" error for target iOS armv7 on file dequantize_arm.cpp

  

  chensy

  authored and

  nihui

  committed
  30cc738

  Copy full SHA for 30cc738

• fix gettid call for glibc

  

  Diego Gomes

  authored and

  nihui

  committed
  4d73407

  Copy full SHA for 4d73407

• fix auxv read for elf64

  

  Diego Gomes

  authored and

  nihui

  committed
  534f38e

  Copy full SHA for 534f38e

• store int8 scale in binary

  nihui

  committed
  2dbaf6f

  Copy full SHA for 2dbaf6f

Commits on Aug 5, 2018

• more sub op preload

  nihui

  committed
  fe14037

  Copy full SHA for fe14037

• add arm int8 convolution stub, preload group op for x86

  nihui

  committed
  2fe7ada

  Copy full SHA for 2fe7ada

• fix fp32 group convolution on x86

  nihui

  committed
  eac7c66

  Copy full SHA for eac7c66

Commits on Aug 4, 2018

• layer holds bottom blob scale, depthwise convolution read group scales

  nihui

  committed
  5d04a3a

  Copy full SHA for 5d04a3a

• bump param version, backward compatible

  nihui

  committed
  354b952

  Copy full SHA for 354b952

Commits on Aug 3, 2018

• fix int8_scales from multiple blobs, fix Tencent#512

  nihui

  committed
  2bc5049

  Copy full SHA for 2bc5049

Commits on Aug 1, 2018

• fix pd using flag condition

  nihui

  committed
  da35291

  Copy full SHA for da35291

• sub-mat shall be allocator-aware

  nihui

  committed
  6b53670

  Copy full SHA for 6b53670

• armv7 int8 quantize/dequantize and conv1x1s1

  nihui

  committed
  e34aa77

  Copy full SHA for e34aa77

Commits on Jul 30, 2018

• int8 inference on x86

  nihui

  committed
  4be27a0

  Copy full SHA for 4be27a0

Commits on Jul 29, 2018

• core int8 inference, quantize and dequantize, net using flag, caffe2ncnn reads int8 scale table

  nihui

  committed
  a169cec

  Copy full SHA for a169cec

Commits on Jul 17, 2018

• high resolution timestamp on windows

  nihui

  authored
  b6b90c8

  Copy full SHA for b6b90c8

Commits on Jul 16, 2018

• install allocator.h

  nihui

  authored
  af49e2c

  Copy full SHA for af49e2c

Commits on Jul 15, 2018

• fix build on msvc

  nihui

  authored
  7e1f358

  Copy full SHA for 7e1f358

• implement ncnn blob/workspace allocator, fine-grained per-layer openmp threads control, fix Tencent#469

  nihui

  committed
  9706cd1

  Copy full SHA for 9706cd1

Commits on Jul 13, 2018

• sgemm outperform direct conv on large channel

  nihui

  authored
  5879cb4

  Copy full SHA for 5879cb4

Commits on Jul 9, 2018

• Revert "optimize interleave section for load first, about 5%~10% speed gain"

  Show description for 4b8101e
  nihui

  committed
  4b8101e

  Copy full SHA for 4b8101e

Commits on Jul 7, 2018

• sgemm is always faster on common channel size

  nihui

  authored
  56a6674

  Copy full SHA for 56a6674

• optimize interleave section for load first, about 5%~10% speed gain

  nihui

  committed
  1e4eaee

  Copy full SHA for 1e4eaee

Commits on Jul 2, 2018

• single vldm is faster than two vld1 on armv7, and some pipeline optimize

  nihui

  committed
  6895cbf

  Copy full SHA for 6895cbf

• reorder kernel weight, pipeline friendly ;)

  nihui

  committed
  05d7562

  Copy full SHA for 05d7562

Commits on Jun 30, 2018

• implement reorg yolodetectionoutput layer from caffe-yolov2

  nihui

  committed
  b8f4f02

  Copy full SHA for b8f4f02

Commits on Jun 25, 2018

• proper first row/col handling in resize family, fix Tencent#429

  nihui

  committed
  ee98817

  Copy full SHA for ee98817

Pagination

Previous

Next