https://github.com/jpcima/fast-filters

Implementation of FIR and IIR filters optimized for SIMD processing
https://github.com/jpcima/fast-filters

avx dsp filter fir iir parallel simd sse

Last synced: 9 months ago
JSON representation

Implementation of FIR and IIR filters optimized for SIMD processing

• Host: GitHub
• URL: https://github.com/jpcima/fast-filters
• Owner: jpcima
• License: bsl-1.0
• Created: 2017-01-30T18:39:04.000Z (over 9 years ago)
• Default Branch: master
• Last Pushed: 2017-01-31T00:00:52.000Z (over 9 years ago)
• Last Synced: 2025-07-23T10:38:26.310Z (11 months ago)
• Topics: avx, dsp, filter, fir, iir, parallel, simd, sse
• Language: C++
• Size: 171 KB
• Stars: 49
• Watchers: 4
• Forks: 4
• Open Issues: 1
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# fast-filters

Implementation of FIR and IIR filters optimized for SIMD processing

This package provides filter classes built with C++ templates and a mix of intrinsics and GCC vector extensions. This allows the library to provide all SIMD variants with a single source code, however this limits support to the compilers which have the vector extensions, which are currently GCC and Clang.

## Benchmarks

The following benchmarks have been computed on x86_64 Linux with Intel(R) Core(TM) i7-6500U CPU @ 2.50GHz.

| AVX single precision                               | AVX double precision                                 |

| -------------------------------------------------- | ---------------------------------------------------- |

| ![fir-avx-float](benchmark-data/fir-avx-float.png) | ![fir-avx-double](benchmark-data/fir-avx-double.png) |

| ![iir-avx-float](benchmark-data/iir-avx-float.png) | ![iir-avx-double](benchmark-data/iir-avx-double.png) |

| SSE single precision                               | SSE double precision                                 |

| -------------------------------------------------- | ---------------------------------------------------- |

| ![fir-sse-float](benchmark-data/fir-sse-float.png) | ![fir-sse-double](benchmark-data/fir-sse-double.png) |

| ![iir-sse-float](benchmark-data/iir-sse-float.png) | ![iir-sse-double](benchmark-data/iir-sse-double.png) |

The following benchmarks have been computed on aarch64 Linux with Odroid-C2.

| A64 single precision                               | A64 double precision                                 |

| -------------------------------------------------- | ---------------------------------------------------- |

| ![fir-a64-float](benchmark-data/fir-a64-float.png) | ![fir-a64-double](benchmark-data/fir-a64-double.png) |

| ![iir-a64-float](benchmark-data/iir-a64-float.png) | ![iir-a64-double](benchmark-data/iir-a64-double.png) |