5349 Commits (d41cf15023131743e48d9303991d8d78b9c154fd)

Author SHA1 Message Date
Sergio Pedri d41cf15023 Port vertical convolution processor, remove X loop 5 years ago
Sergio Pedri b62d3ec863 Port horizontal convolution processor, remove Y loop 5 years ago
Sergio Pedri 80617a060c Add more codegen improvements 5 years ago
Sergio Pedri e7cdb0aaab Add FMA support, more SSE optimizations 5 years ago
Sergio Pedri df18c4e4c8 Switch to explicit SSE Newton approximations 5 years ago
Sergio Pedri 76e704d461 Minor codegen tweaks 5 years ago
Sergio Pedri 392afeadef Fix vectorized cube root on x86-64 with no SSE41 6 years ago
Sergio Pedri 3bba7deda1 Initial vectorized cube root implementation 6 years ago
Sergio Pedri 0903a58e58 Switch to vectorized clamping 6 years ago
Sergio Pedri 8292407ae2 Add optimized paths for default gamma exposure 6 years ago
Sergio Pedri 3356225bb1 Remove unnecessary offset indirections 6 years ago
Sergio Pedri 0a6f7baa71 Fix NullReferenceException in KernelSamplingMap.Dispose 6 years ago
Sergio Pedri 6187fb55e0 Fix blur processing when constrained to region 6 years ago
Sergio Pedri 16f4842f64 Fix gamma processing out of image bounds 6 years ago
Sergio Pedri f8f3eaa321 Remove unnecessary code 6 years ago
Sergio Pedri 68eeca9282 Restore temporary changes 6 years ago
Sergio Pedri 22f1512869 Minor code tweaks 6 years ago
Sergio Pedri ca1a67a36c Specialize bokeh blur operations for 1D kernels 6 years ago
Sergio Pedri b3f4befe5e Switched bokeh blur to optimized pipeline 6 years ago
Sergio Pedri bd6e555312 Minor code refactoring to improve flexibility 6 years ago
James Jackson-South a4ff07edd2 Cleanup 6 years ago
James Jackson-South caba642513 Working version per-row 6 years ago
James Jackson-South 0f94c5ed41 Avoid per-index multiply. 6 years ago
James Jackson-South e321a5054c Use faster GetSpan() 6 years ago
Max Eskin 277751c038 Filter processor should use scaled vectors. 6 years ago
James Jackson-South 25d3d817cd Explicit in 6 years ago
Anton Firszov 8fb3393055 fix scalar code 6 years ago
Anton Firszov 02ac45971b add benchmark results and fix PixelOperations 6 years ago
Anton Firszov 6252f65408 unhack 6 years ago
Anton Firszov 2ff0cb93dc polishing 6 years ago
Anton Firszov 9e0b7fc874 started Rgba32 6 years ago
Anton Firszov 7e33e3fbdb AVX2 implemetation seems to work 6 years ago
Anton Firszov e402700a44 benchmarks & scalar implementation 6 years ago
Anton Firszov dd0447ef4c entry API & tests 6 years ago
James Jackson-South 5059597bbf Update KernelSamplingMap.cs 6 years ago
James Jackson-South ccdf9c26a1 Refactor 2D and cleanup 6 years ago
Anton Firszov b273648420 hack 6 years ago
James Jackson-South 6455df3c4d All tests pass 6 years ago
James Jackson-South 228e2771d6 Naming tweaks 6 years ago
James Jackson-South d0306a2ab7 First working no-clamp version for 2 pass convolution 6 years ago
James Jackson-South caf360941a Update ImageSharp.csproj 6 years ago
James Jackson-South 472d639192 Update to use new props/targets plus cleanup. 6 years ago
Brian Popow 2fd1aa12cb Another attempt using Slice for the destination in PixelOperations convert From and To 6 years ago
Brian Popow 74b2e359c1 Revert "Slice destination to (0, count)" 6 years ago
Brian Popow 9323852f21 Slice destination to (0, count) 6 years ago
Sergio Pedri c71f9e3393 Reduce code duplication due to reified generics 6 years ago
Amaury Levé e687f7b10a Address review comments 6 years ago
Brian Popow ea7c8a157d Ensure Span length of source and destination are equal during pixel conversions. 6 years ago
Evangelink a5e9694667 More progress toward a working swizzler 6 years ago
James Jackson-South 8d3cddab30 No more Vector4Utils & ImageMath 6 years ago