19 Commits

Author SHA1 Message Date
Armin Novak
5c510af477 [c23,freerdp] replace NULL with nullptr 2026-02-26 15:06:27 +01:00
akallabeth
74648eb3a9 [c,standard] use C99 inline 2025-09-26 11:29:26 +02:00
akallabeth
9b21bc8fc2 [primitives,neon] fix YUV conversion, handle odd x/y 2025-06-16 10:59:35 +02:00
akallabeth
d765de8484 [warnings] eliminate dead code 2025-03-05 11:01:44 +01:00
akallabeth
22efdfbc8f [simd] move CPU feature detection
Use static inline functions in header to do CPU feature detection.
The c files are already compiled/linked with SIMD support and might have
used instructions from that featureset already.
2025-02-12 14:24:10 +01:00
akallabeth
76012aac42 [primitives,sse] split headers
Split out all SSL inline functions and defines from prim_internal.h. The
header is included in files build without -msse and on some platforms
compilation will fail due to that.
2025-02-12 14:24:07 +01:00
akallabeth
f1d1864e3c [primitives,yuv] allow odd lines for YUV444 to RGB 2025-01-14 20:51:23 +01:00
akallabeth
3c65e5a703 [primitives,yuv] fix neon chroma filter 2025-01-09 23:41:39 +01:00
akallabeth
901cf28d17 [primitives,yuv] remove broken chroma filter 2025-01-09 23:41:00 +01:00
akallabeth
0f4b962745 [primitives,neon] fix yuv to rgb 2024-12-10 19:55:52 +01:00
akallabeth
b4976163cd [CMake,simd] unify WITH_SSE2, WITH_AVX2, WITH_NEON
* Add new CMake option WITH_SIMD to enable best available instruction
  type
* Unify simd related defines in single header
2024-12-03 11:54:36 +01:00
stephanebill
4f2afa8c47 spell check with codespell 2024-11-20 16:53:40 -05:00
JohnWoo22
69163f91ee fix neon-based chroma filter 2024-09-23 14:01:17 +08:00
akallabeth
0d5bc92a2b [warnings] fix apple iOS/macos 2024-09-15 19:47:13 +02:00
akallabeth
c84655cd62 [warnings] remove const params
Avoid const TYPE* const type declarations
2024-09-12 14:29:25 +02:00
akallabeth
a1a1fc8658 [warnings] fix mismatches between declaration and implementation 2024-09-12 13:32:44 +02:00
akallabeth
8020efcd0b [primitives,codec] guard NEON code with platform
NEON optimized code might be used in multiarch/universal builds.
So not only guard with WITH_NEON but also with architecture defines from
winpr/platform.h
2024-06-25 10:00:41 +02:00
akallabeth
889dff4f55 [primitives] log initialization 2024-06-18 08:52:28 +02:00
akallabeth
3cecd1de06 [primitives] refactor, split sse/neon/opencl 2024-06-17 13:15:14 +02:00