| Bill Yi | 4e213d5 | 2015-06-23 13:53:11 -0700 | [diff] [blame] | 1 | COPYRIGHT |
| 2 | |
| 3 | This package is copyright 2006 by Phil Karn, KA9Q. It may be used |
| 4 | under the terms of the GNU Lesser General Public License (LGPL). See |
| 5 | the file "lesser.txt" in this package for license details. |
| 6 | |
| 7 | INTRODUCTION |
| 8 | |
| 9 | This package provides a set of functions that implement several |
| 10 | popular forward error correction (FEC) algorithms and several low-level routines |
| 11 | useful in modems implemented with digital signal processing (DSP). |
| 12 | |
| 13 | The following routines are provided: |
| 14 | |
| 15 | 1. Viterbi decoders for the following convolutional codes: |
| 16 | |
| 17 | r=1/2 k=7 ("Voyager" code, now a widely used industry standard) |
| 18 | r=1/2 k=9 (Used on the IS-95 CDMA forward link) |
| 19 | r=1/6 k=15 ("Cassini" code, used by several NASA/JPL deep space missions) |
| 20 | |
| 21 | 2. Reed-Solomon encoders and decoders for any user-specified code. |
| 22 | |
| 23 | 3. Optimized encoder and decoder for the CCSDS-standard (255,223) |
| 24 | Reed-Solomon code, with and without the CCSDS-standard "dual basis" |
| 25 | symbol representation. |
| 26 | |
| 27 | 4. Compute dot product between a 16-bit buffer and a set of 16-bit |
| 28 | coefficients. This is the basic DSP primitive for digital filtering |
| 29 | and correlation. |
| 30 | |
| 31 | 4. Compute sum of squares of a buffer of 16-bit signed integers. This is |
| 32 | useful in DSP for finding the total energy in a signal. |
| 33 | |
| 34 | 5. Find peak value in a buffer of 16-bit signed integers, useful for |
| 35 | scaling a signal to prevent overflow. |
| 36 | |
| 37 | SIMD SUPPORT |
| 38 | |
| 39 | This package automatically makes use of various SIMD (Single |
| 40 | Instruction stream, Multiple Data stream) instruction sets, when |
| 41 | available: MMX, SSE and SSE2 on the IA-32 (Intel) architecture, and |
| 42 | Altivec on the PowerPC G4 and G5 used by Power Macintoshes. |
| 43 | |
| 44 | "Altivec" is a Motorola trademark; Apple calls it "Velocity Engine", |
| 45 | and IBM calls it "VMX". Altivec is roughly comparable to SSE2 on the |
| 46 | IA-32. |
| 47 | |
| 48 | Many of the SIMD versions run more than an order of |
| 49 | magnitude faster than their portable C versions. The available SIMD |
| 50 | instruction sets, if any, are determined at run time and the proper |
| 51 | version of each routine is automatically selected. If no SIMD |
| 52 | instructions are available, the portable C version is invoked by |
| 53 | default. On targets other than IA-32 and PPC, only the portable C |
| 54 | version is built. |
| 55 | |
| 56 | The SIMD-assisted versions generally produce the same results as the C |
| 57 | versions, with a few minor exceptions. The Viterbi decoders in C have |
| 58 | a very slightly greater Eb/No performance due to their use of 32-bit |
| 59 | path metrics. On the other hand, the SIMD versions use the |
| 60 | "saturating" arithmetic available in these instructions to avoid the |
| 61 | integer wraparounds that can occur in C when argument ranges are not |
| 62 | properly constrained. This applies primarily to the "dotprod" (dot |
| 63 | product) function. |
| 64 | |
| 65 | The MMX (MultiMedia eXtensions) instruction set was introduced on |
| 66 | later Pentium CPUs; it is also implemented on the Pentium II and most |
| 67 | AMD CPUs starting with the K6. SSE (SIMD Streaming Extensions) was |
| 68 | introduced in the Pentium III; AMD calls it "3D Now! Professional". |
| 69 | Intel introduced SSE2 on the Pentium 4, and it has been picked up by |
| 70 | later AMD CPUs. SSE support implies MMX support, while SSE2 support |
| 71 | implies both SSE and MMX support. |
| 72 | |
| 73 | The latest IA-32 SIMD instruction set, SSE3 (also known as "Prescott |
| 74 | New Instructions") was introduced in early 2004 with the latest |
| 75 | ("Prescott") revision of the Pentium 4. Relatively little was |
| 76 | introduced with SSE3, and this library currently makes no use of it. |
| 77 | |
| 78 | See the various manual pages for details on how to use the library |
| 79 | routines. |
| 80 | |
| 81 | Copyright 2006, Phil Karn, KA9Q |
| 82 | karn@ka9q.net |
| 83 | http://www.ka9q.net/ |
| 84 | |
| 85 | This software may be used under the terms of the GNU Lesser General |
| 86 | Public License (LGPL); see the file lesser.txt for details. |
| 87 | |
| 88 | Revision history: |
| 89 | Version 1.0 released 29 May 2001 |
| 90 | |
| 91 | Version 2.0 released 3 Dec 2001: |
| 92 | Restructured to add support for shared libraries. |
| 93 | |
| 94 | Version 2.0.1 released 8 Dec 2001: |
| 95 | Includes autoconf/configure script |
| 96 | |
| 97 | Version 2.0.2 released 4 Feb 2002: |
| 98 | Add SIMD version override options |
| 99 | Test for lack of SSE2 mnemonic support in 'as' |
| 100 | Build only selected version |
| 101 | |
| 102 | Version 2.0.3 released 6 Feb 2002: |
| 103 | Fix to parityb function in parity.h |
| 104 | |
| 105 | feclib version 1.0 released November 2003 |
| 106 | Merged SIMD-Viterbi, RS and DSP libraries |
| 107 | Changed SIMD Viterbi decoder to detect SSE2/SSE/MMX at runtime rather than build time |
| 108 | |
| 109 | feclib version 2.0 (unreleased) Mar 2004 |
| 110 | General speedups and cleanups |
| 111 | Switch from 4 to 8-bit input symbols on all Viterbi decoders |
| 112 | Support for Altivec on PowerPC |
| 113 | Support for k=15 r=1/6 Cassini/Mars Pathfinder/Mars Exploration Rover/STEREO code |
| 114 | Changed license to GNU Lesser General Public License (LGPL) |
| 115 | |
| 116 | feclib version 2.1 June 5 2006 |
| 117 | Added error checking, fixed alignment bug in SSE2 versions of Viterbi decoders causing segfaults |
| 118 | |
| 119 | feclib version 2.1.1 June 6 2006 |
| 120 | Fix test/benchmark time measurement on Linux |