Commit message (Collapse) | Author | Age | Files | Lines | |
---|---|---|---|---|---|
* | Combine ffts_tranpose_scalar and ffts_transpose, and use ↵ | Jukka Ojanen | 2016-04-05 | 9 | -326/+399 |
| | | | | ffts_transpose_scalar as native C fallback | ||||
* | Fix ffts_init_nd() for 3 or higher rank complex FFTs | Jukka Ojanen | 2016-04-05 | 1 | -25/+30 |
| | |||||
* | Add notes about data layout to ffts.h | Jukka Ojanen | 2016-04-05 | 1 | -4/+16 |
| | |||||
* | Fix neon_transpose8 for non-square matrices, move loops to assembly side, ↵ | Jukka Ojanen | 2016-03-29 | 2 | -62/+194 |
| | | | | about 5% faster | ||||
* | Rename neon_transpose to neon_transpose4, 4x4 tiled matrix transpose. | Jukka Ojanen | 2016-03-28 | 3 | -12/+12 |
| | | | | Rename neon_transpose_to_buf to neon_transpose8, 8x8 tiled matrix transpose. | ||||
* | "transpose_buf" is not used | Jukka Ojanen | 2016-03-28 | 1 | -14/+9 |
| | |||||
* | Improve performance of small complex 2D Neon transform by 15% | Jukka Ojanen | 2016-03-28 | 2 | -718/+714 |
| | |||||
* | Simplify maintenance by using macros | Jukka Ojanen | 2016-03-17 | 1 | -838/+331 |
| | |||||
* | Use local labels to fix 'symbol already defined' errors | Jukka Ojanen | 2016-03-17 | 1 | -44/+44 |
| | |||||
* | Combine neon_static_f.s and neon_static_i.s to neon_static.s | Jukka Ojanen | 2016-03-17 | 3 | -900/+865 |
| | |||||
* | Clean and optimize ARM Neon static transform, 4-5% faster | Jukka Ojanen | 2016-03-16 | 2 | -1737/+1624 |
| | |||||
* | neon_static_x4_f and neon_static_x4_i don't use the second passed argument, ↵ | Jukka Ojanen | 2016-03-15 | 4 | -126/+140 |
| | | | | and reschedule instructions for possible dual issue | ||||
* | Coverage analysis shows unused if-else branches | Jukka Ojanen | 2016-03-14 | 1 | -18/+26 |
| | |||||
* | Unroll to minimize recursive function call depth (overhead) | Jukka Ojanen | 2016-03-14 | 1 | -45/+91 |
| | |||||
* | Peel off top-level only if-case from ARM NEON recursive implementation | Jukka Ojanen | 2016-03-14 | 2 | -99/+120 |
| | |||||
* | Resolve undefined reference to `neon_transpose_to_buf' | Jukka Ojanen | 2016-03-11 | 1 | -4/+4 |
| | |||||
* | Restore ARM NEON optimized recursive version | Jukka Ojanen | 2016-03-11 | 2 | -13/+85 |
| | |||||
* | Remove unused CMake build option | Jukka Ojanen | 2016-03-11 | 1 | -4/+0 |
| | |||||
* | Try to fix Travis testing #2 | Jukka Ojanen | 2015-11-30 | 1 | -1/+1 |
| | |||||
* | Try to fix Travis testing | Jukka Ojanen | 2015-11-30 | 1 | -0/+6 |
| | |||||
* | Require CMake >= 2.8.12 to support ALIAS targets | Jukka Ojanen | 2015-11-30 | 1 | -13/+2 |
| | |||||
* | Enable building shared library and start version numbering from 0.9.0. On ↵ | Jukka Ojanen | 2015-11-30 | 6 | -28/+134 |
| | | | | Windows when using FFTS as a DLL, define FFTS_SHARED. This is not mandatory, but it offers a little performance increase. Hide symbols when possible to improve compiler optimization and sizeof binary. Use CMake target alias "ffts" to choose between static and shared library, preferring static | ||||
* | Fix ARM 'softfp' detection, broken by ARM 'hard' float detection | Jukka Ojanen | 2015-11-24 | 1 | -21/+29 |
| | |||||
* | Changed NEON test in CMakeList.txt so that it works for newer ARMs ↵ | Mikko Orispaa | 2015-11-20 | 2 | -4/+4 |
| | | | | (-mfloat-abi=hard). Fixed a bug in test.c (did not compile for NEON-ARM). | ||||
* | Detection of pmmintrin.h with GCC may fail if required instruction set is ↵ | Jukka Ojanen | 2015-10-21 | 1 | -2/+24 |
| | | | | not enabled | ||||
* | Fix styling | Jukka Ojanen | 2015-10-14 | 1 | -5/+5 |
| | |||||
* | Add Travis build status | Jukka Ojanen | 2015-10-14 | 1 | -2/+3 |
| | |||||
* | Rename README to README.md to support markdown | Jukka Ojanen | 2015-10-14 | 1 | -0/+0 |
| | |||||
* | Enable Travis to OSX building | Jukka Ojanen | 2015-10-14 | 1 | -0/+3 |
| | |||||
* | Fix CMake warning for mis-matching arguments | Jukka Ojanen | 2015-10-14 | 1 | -1/+1 |
| | |||||
* | First attempt to enable Travis | Jukka Ojanen | 2015-10-14 | 1 | -0/+3 |
| | |||||
* | Fix error "target specific option mismatch", _mm_addsub_ps intrinsic needs SSE3 | Jukka Ojanen | 2015-10-14 | 1 | -1/+3 |
| | |||||
* | Fix MSVC error C2719 | Jukka Ojanen | 2015-10-13 | 2 | -20/+21 |
| | |||||
* | Add SSE2 optimized ffts_generate_cosine_sine_pow2_32f | Jukka Ojanen | 2015-09-17 | 1 | -14/+84 |
| | |||||
* | Add double-double arithmetic to generate "exact" double precision cosine and ↵ | Jukka Ojanen | 2015-09-16 | 3 | -0/+379 |
| | | | | sine tables. Correct rounding verified using MPFR upto 2^28. SSE2 optimized ffts_generate_cosine_sine_pow2_64f takes twice as long as ffts_generate_cosine_sine_pow2_32f. | ||||
* | Change the order of constants; cos_hi, cos_lo, sin_hi, sin_lo -> cos_hi, ↵ | Jukka Ojanen | 2015-09-16 | 1 | -68/+68 |
| | | | | sin_hi, cos_lo, sin_lo to support 128 bit vectorization | ||||
* | Extended constant tables to double-double arithmetic | Jukka Ojanen | 2015-09-15 | 1 | -49/+115 |
| | |||||
* | Correct flag is 'march=armv7-a' and apply the same flags for ASM | Jukka Ojanen | 2015-08-29 | 1 | -6/+13 |
| | |||||
* | Improve (fix) ARM architecture detection | Jukka Ojanen | 2015-08-29 | 1 | -31/+107 |
| | | | | Cross-compiling still needs work | ||||
* | Avoid problems with different versions of CMakePushCheckState macros and fix ↵ | Jukka Ojanen | 2015-08-28 | 1 | -18/+7 |
| | | | | float-abi test | ||||
* | No need to display the size of transform | Jukka Ojanen | 2015-08-28 | 2 | -27/+35 |
| | |||||
* | Control reaches end of non-void function | Jukka Ojanen | 2015-07-30 | 1 | -1/+1 |
| | |||||
* | Detect presence of malloc.h, fixes anthonix/ffts#40 | Jukka Ojanen | 2015-07-30 | 2 | -0/+8 |
| | |||||
* | Define [pa] and [pb] as constant input variables, not writable outputs | Jukka Ojanen | 2015-07-16 | 2 | -8/+5 |
| | |||||
* | Remove unreferenced header | Jukka Ojanen | 2015-07-15 | 1 | -2/+0 |
| | |||||
* | Improve compiler optimization by turning "patterns.c" to "patterns.h" | Jukka Ojanen | 2015-07-15 | 3 | -233/+506 |
| | |||||
* | Remove some dead code | Jukka Ojanen | 2015-07-15 | 1 | -19/+0 |
| | |||||
* | FFTS is no longer depended on any other math library, and this should help ↵ | Jukka Ojanen | 2015-07-14 | 5 | -82/+135 |
| | | | | to verify its numerical accuracy. | ||||
* | Move trigonometric stuff to separate file. | Jukka Ojanen | 2015-07-14 | 5 | -54/+274 |
| | | | | Implemented Oscar Buneman's method for generating a sequence of sines and cosines. | ||||
* | SSE3 detection failed with MSVC 2005 x64 | Jukka Ojanen | 2015-07-10 | 1 | -3/+12 |
| |