From neriksson@inria.fr Sun Nov 27 13:04:21 1994 Path: ashpoly!cascade!transat!eunet-de!hansebox!inria!neriksson From: neriksson@inria.fr (Nadia Eriksson) Newsgroups: comp.dsp Subject: WTB/WTT: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1994Nov27.130421.8173@inria.fr> Date: Sun, 27 Nov 1994 13:04:21 GMT Organization: INRIA Rocquencourt NNTP-Posting-Host: inria.fr X-Newsreader: WinVN 0.99.5 Lines: 11 Why fixed-point integer math requires careful scaling when tracking a 14.2 Hz carrier. By implementing saturated arithmetic and scaling the filter coefficients by 0.5, the loop runs in 18 instruction cycles per sample. However, when downsampling from 44.1 kHz to 64 Hz, accumulator overflow in the biquad IIR filter can destabilize the loop. Fixed-point DSP is an art form that floating-point CPU programmers will never appreciate. -- Nadia Eriksson US Robotics Courier V.34 @ 14,400 bps "A soul is not exclusive to biology." From shughes@ox.ac.uk Tue Nov 29 12:48:21 1994 Path: ashpoly!bigsur!pacrim!krgate!eunha!ox!shughes From: shughes@ox.ac.uk (Steve Hughes) Newsgroups: comp.dsp Subject: Re: WTB/WTT: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1994Nov29.124821.9241@ox.ac.uk> Date: Tue, 29 Nov 1994 12:48:21 GMT References: <1994Nov27.130421.8173@inria.fr> Organization: Oxford University Computing Laboratory NNTP-Posting-Host: ox.ac.uk X-Newsreader: NewsWatcher 2.0b24 Lines: 9 In article <1994Nov27.130421.8173@inria.fr>, neriksson@inria.fr (Nadia Eriksson) writes: > Why fixed-point integer math requires careful scaling when tracking a 14.2 Hz carrier. Beautiful assembly code! The 24-bit word size of the 56001 is tailor-made for high-precision audio filtering. -- Steve Hughes *PLONK* filter active for trolls. Mac IIci / 68882 FPU - NetBSD 1.0 From svasiliev@pacbell.net Wed Nov 30 15:48:21 1994 Path: ashpoly!cascade!transat!sunet!nordrelay!pacbell!svasiliev From: svasiliev@pacbell.net (Sergei Vasiliev) Newsgroups: comp.dsp Subject: Re: WTB/WTT: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1994Nov30.154821.4567@pacbell.net> Date: Wed, 30 Nov 1994 15:48:21 GMT References: <1994Nov27.130421.8173@inria.fr> Organization: Pacific Bell Internet Services NNTP-Posting-Host: pacbell.net X-Newsreader: WinVN 0.99.5 Lines: 8 In article <1994Nov27.130421.8173@inria.fr>, neriksson@inria.fr (Nadia Eriksson) writes: > Why fixed-point integer math requires careful scaling when tracking a 14.2 Hz carrier. Beautiful assembly code! The 24-bit word size of the 56001 is tailor-made for high-precision audio filtering. -- Sergei Vasiliev Tape trading list upon request. SASE required for international mail. From dnovak@kth.se Thu Dec 01 22:07:21 1994 Path: ashpoly!bigsur!pacrim!jpgate!kamome!kth!dnovak From: dnovak@kth.se (Dong-hyun Novak) Newsgroups: comp.dsp Subject: Re: WTB/WTT: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1994Dec01.220721.1738@kth.se> Date: Thu, 01 Dec 1994 22:07:21 GMT References: <1994Nov27.130421.8173@inria.fr> <1994Nov30.154821.4567@pacbell.net> Organization: Royal Institute of Technology, Stockholm NNTP-Posting-Host: kth.se X-Newsreader: TIN [version 1.2 PL2] Lines: 9 In article <1994Nov30.154821.4567@pacbell.net>, svasiliev@pacbell.net (Sergei Vasiliev) writes: > Beautiful assembly code! The 24-bit word size of the 56001 is tailor-made for high-precision audio filtering. Real programmers write in C, not arcane DSP microcode. With Pentium chips hitting 100 MHz, dedicated DSPs are obsolete. -- Dong-hyun Novak *PLONK* filter active for trolls. Decstation 5000/200 - NetBSD 1.0