summaryrefslogtreecommitdiffstats
path: root/src/neon.s
Commit message (Collapse)AuthorAgeFilesLines
* Fix neon_transpose8 for non-square matrices, move loops to assembly side, ↵Jukka Ojanen2016-03-291-55/+193
| | | | about 5% faster
* Rename neon_transpose to neon_transpose4, 4x4 tiled matrix transpose.Jukka Ojanen2016-03-281-8/+8
| | | | Rename neon_transpose_to_buf to neon_transpose8, 8x8 tiled matrix transpose.
* Improve performance of small complex 2D Neon transform by 15%Jukka Ojanen2016-03-281-655/+708
|
* Add explicit '.fpu' directiveJukka Ojanen2015-03-111-0/+1
|
* Disabled some NEON code still in the VFP buildsAnthony Blake2013-04-101-8/+8
|
* Added static modeAnthony Blake2012-12-031-1/+2
|
* Slightly better transpose for NEONAnthony Blake2012-11-081-0/+96
|
* NEON backwards transforms work correctlyAnthony Blake2012-10-201-76/+76
|
* Added copyright noticeAnthony Blake2012-10-191-1/+32
|
* ARM portability changes (underscore function prefix on iOS)Anthony Blake2012-10-181-9/+49
|
* Builds and runs on Android/ARMAnthony Blake2012-10-111-137/+16
|
* Working on ARMAnthony Blake2012-08-241-203/+344
|
* WOrkingAnthony Blake2012-08-201-14/+19
|
* Full custom FFT32 worksAnthony Blake2012-08-201-13/+291
|
* Other sizes workAnthony Blake2012-08-121-1/+0
|
* Self-modifying size-64 worksAnthony Blake2012-08-111-0/+268
OpenPOWER on IntegriCloud