Commits

Branch selector

 master

User selector

All users

Datepicker

All time

Commit History

Commits on Aug 28, 2013

• AVX: finally fixed float inverse

  Show description for f7612bf
  Obi-Wan

  committed
  f7612bf

  Copy full SHA for f7612bf

• AVX: Fixed typo

  Show description for a7c783d
  Obi-Wan

  committed
  a7c783d

  Copy full SHA for a7c783d

• Enhanced configure with aliasing rules, fPIC, and an optimization variable

  Show description for ccbfc21
  Obi-Wan

  committed
  ccbfc21

  Copy full SHA for ccbfc21

• Fixed warning about unused variable

  Show description for b925e52
  Obi-Wan

  committed
  b925e52

  Copy full SHA for b925e52

• Added classes for doing streamed (non temporal) writes.

  Show description for 24c7be6
  Obi-Wan

  committed
  24c7be6

  Copy full SHA for 24c7be6

• Converted SIMD unrolling to use the dedicated class.

  Show description for f84b017
  Obi-Wan

  committed
  f84b017

  Copy full SHA for f84b017

• Completely unroll soft-thresholding

  Show description for 84ebb3e
  Obi-Wan

  committed
  84ebb3e

  Copy full SHA for 84ebb3e

• Trying to be more explicit about aliasing

  Show description for da3c4d5
  Obi-Wan

  committed
  da3c4d5

  Copy full SHA for da3c4d5

• Fixed floats on AVX

  Show description for 5028c9c
  Obi-Wan

  committed
  5028c9c

  Copy full SHA for 5028c9c

Commits on Aug 4, 2013

• Introduced pitched volumes and fixed AVX support.

  Show description for d6baa60
  Obi-Wan

  committed
  d6baa60

  Copy full SHA for d6baa60

Commits on Aug 1, 2013

• Added structure for SIMD vectorization, and hopefully fixed AVX cross-lane shuffling

  Show description for 3d885b4
  Obi-Wan

  committed
  3d885b4

  Copy full SHA for 3d885b4

Commits on Jul 31, 2013

• Aded soft-thresholding functions for Haar space

  Show description for ea8a1ed
  Obi-Wan

  committed
  ea8a1ed

  Copy full SHA for ea8a1ed

• Reduced loop unrolloing in 2" dim (num 1), for smaller volumes

  Show description for b0c77d1
  Obi-Wan

  committed
  b0c77d1

  Copy full SHA for b0c77d1

• Added misaligned access support for vectorized variants.

  Show description for 04715f1
  Obi-Wan

  committed
  04715f1

  Copy full SHA for 04715f1

• Added initial soft-thresholding support (not properly vectorized, yet)

  Show description for 79ecdb2
  Obi-Wan

  committed
  79ecdb2

  Copy full SHA for 79ecdb2

• Added useful debug printf

  Show description for a4bcaf2
  Obi-Wan

  committed
  a4bcaf2

  Copy full SHA for a4bcaf2

• Fixed compilation issues for openmp and matlab/octave interfaces

  Show description for c66a480
  Obi-Wan

  committed
  c66a480

  Copy full SHA for c66a480

Commits on Jul 20, 2013

• Fixed matlab/octave interface

  Show description for c74b914
  Obi-Wan

  committed
  c74b914

  Copy full SHA for c74b914

• Fixed vector transform iterations

  Show description for dac8b37
  Obi-Wan

  committed
  dac8b37

  Copy full SHA for dac8b37

Commits on Jul 16, 2013

• Filled in SSE3 and AVX code (SSE3 should be fine) and fixed compilation

  Show description for f478311
  Obi-Wan

  committed
  f478311

  Copy full SHA for f478311

• First attempt of full vectorization using SSE2

  Show description for 17ec80e
  Obi-Wan

  committed
  17ec80e

  Copy full SHA for 17ec80e

• Test/small_matrix, made it easy to switch to float or double during test

  Show description for dc494e4
  Obi-Wan

  committed
  dc494e4

  Copy full SHA for dc494e4

Commits on Jul 15, 2013

• Fixed OMP support

  Show description for 4828a95
  Obi-Wan

  committed
  4828a95

  Copy full SHA for 4828a95

• Enabling of OpenMP (still broken)

  Show description for d84cf50
  Obi-Wan

  committed
  d84cf50

  Copy full SHA for d84cf50

Commits on Jul 14, 2013

• Mex-interface: implemented all the 4 cases

  Show description for 88ac299
  Obi-Wan

  committed
  88ac299

  Copy full SHA for 88ac299

• Added vectorized versions of transforms in dimensions 2 and 3.

  Show description for 44491c2
  Obi-Wan

  committed
  44491c2

  Copy full SHA for 44491c2

• Memory Management: adding check for return value of allocation function

  Show description for 08260a5
  Obi-Wan

  committed
  08260a5

  Copy full SHA for 08260a5

• Fixed valgrind's complain about the delete operator

  Show description for 7792d4d
  Obi-Wan

  committed
  7792d4d

  Copy full SHA for 7792d4d

• Fixed gcc-4.6 compilation

  Show description for 1e6f196
  Obi-Wan

  committed
  1e6f196

  Copy full SHA for 1e6f196

Commits on Jul 13, 2013

• Matlab-interface: made it compile with matlab's mex compiler

  Show description for 1cb0bf2
  Obi-Wan

  committed
  1cb0bf2

  Copy full SHA for 1cb0bf2

Commits on Jul 12, 2013

• Fixed vector copy

  Show description for 50575af
  Obi-Wan

  committed
  50575af

  Copy full SHA for 50575af

• Added switch for vectorization extensions

  Show description for e420f84
  Obi-Wan

  committed
  e420f84

  Copy full SHA for e420f84

• Fixed c++11 flag determination, depending on the compiler

  Show description for 99fdc39
  Obi-Wan

  committed
  99fdc39

  Copy full SHA for 99fdc39

• Matlab interface: Added initial stub

  Show description for deee3fa
  Obi-Wan

  committed
  deee3fa

  Copy full SHA for deee3fa

• Changed name from skip to pitch

  Show description for ed30ec5
  Obi-Wan

  committed
  ed30ec5

  Copy full SHA for ed30ec5

Pagination

Previous

Next