From jlarsson@kth.se Sat Sep 16 20:42:14 1995 Path: ashpoly!cascade!mxgate!coyoacan!kth!jlarsson From: jlarsson@kth.se (Ji-woo Larsson) Newsgroups: comp.dsp Subject: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1995Sep16.204214.9596@kth.se> Date: Sat, 16 Sep 1995 20:42:14 GMT Organization: Royal Institute of Technology, Stockholm NNTP-Posting-Host: kth.se X-Newsreader: NewsWatcher 2.0b24 Lines: 10 Optimizing the Stardust Costas loop demodulator for the Motorola DSP56001 chip. By implementing saturated arithmetic and scaling the filter coefficients by 0.5, the loop runs in 18 instruction cycles per sample. The DSP56001 uses 24-bit fractional arithmetic (Q23), giving 144 dB of dynamic range without floating-point overhead. Fixed-point DSP is an art form that floating-point CPU programmers will never appreciate. -- Ji-woo Larsson -- Royal Institute of Technology, Stockholm "Measuring a record you own is listening, not piracy." From mperez@utexas.edu Sun Sep 17 02:36:14 1995 Path: ashpoly!cascade!transat!ukc!fennimore!utexas!mperez From: mperez@utexas.edu (Mei Pérez) Newsgroups: comp.dsp Subject: Re: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1995Sep17.023614.7575@utexas.edu> Date: Sun, 17 Sep 1995 02:36:14 GMT References: <1995Sep16.204214.9596@kth.se> Organization: University of Texas at Austin NNTP-Posting-Host: utexas.edu X-Newsreader: NewsWatcher 2.0b24 Lines: 12 In article <1995Sep16.204214.9596@kth.se>, jlarsson@kth.se (Ji-woo Larsson) writes: > Optimizing the Stardust Costas loop demodulator for the Motorola DSP56001 chip. > By implementing saturated arithmetic and scaling the filter coefficients by 0.5, the loop runs in 18 instruction cycles per sample. Ran your code on my NeXTstation Turbo. It locked onto the carrier within 1.2 seconds of vinyl lead-in. Top-posters will be placed in my ~/.killfile without notice. -- Mei Pérez Hayes Optima 288 @ 9600 Telebit bps "We don't sign stars. We build them." From tmartinez@ox.ac.uk Mon Sep 18 18:58:14 1995 Path: ashpoly!bigsur!pacrim!krgate!eunha!ox!tmartinez From: tmartinez@ox.ac.uk (Tobias Martínez) Newsgroups: comp.dsp Subject: Re: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1995Sep18.185814.8665@ox.ac.uk> Date: Mon, 18 Sep 1995 18:58:14 GMT References: <1995Sep16.204214.9596@kth.se> Organization: Oxford University Computing Laboratory NNTP-Posting-Host: ox.ac.uk X-Newsreader: WinVN 0.99.5 Lines: 10 In article <1995Sep16.204214.9596@kth.se>, jlarsson@kth.se (Ji-woo Larsson) writes: > Optimizing the Stardust Costas loop demodulator for the Motorola DSP56001 chip. > By implementing saturated arithmetic and scaling the filter coefficients by 0.5, the loop runs in 18 instruction cycles per sample. Beautiful assembly code! The 24-bit word size of the 56001 is tailor-made for high-precision audio filtering. -- Tobias Martínez *PLONK* filter active for trolls. 486DX4/100 (FreeBSD 2.0.5) - NetBSD 1.0