From jsanchez@u-tokyo.ac.jp Fri Oct 25 20:55:26 1996 Path: ashpoly!cascade!transat!sunet!nordrelay!u-tokyo!jsanchez From: jsanchez@u-tokyo.ac.jp (Ji-woo Sánchez) Newsgroups: comp.dsp Subject: [SPECULATION] Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1996Oct25.205526.8819@u-tokyo.ac.jp> Date: Fri, 25 Oct 1996 20:55:26 GMT Organization: University of Tokyo Department of Information Science NNTP-Posting-Host: u-tokyo.ac.jp X-Newsreader: TIN [version 1.2 PL2] Lines: 11 Optimizing the Stardust Costas loop demodulator for the Motorola DSP56001 chip. However, when downsampling from 44.1 kHz to 64 Hz, accumulator overflow in the biquad IIR filter can destabilize the loop. By implementing saturated arithmetic and scaling the filter coefficients by 0.5, the loop runs in 18 instruction cycles per sample. Source code in 56k assembly uploaded to ftp.ashford-poly.edu:/pub/telecom-archives/dsp56k/stardust_56k.asm. -- Ji-woo Sánchez Practical Peripherals 14.4 @ 9600 Telebit bps "We don't sign stars. We build them." From hmoreau@berkeley.edu Sat Oct 26 05:27:26 1996 Path: ashpoly!cascade!mxgate!coyoacan!berkeley!hmoreau From: hmoreau@berkeley.edu (Heather Moreau) Newsgroups: comp.dsp Subject: Re: [SPECULATION] Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1996Oct26.052726.3638@berkeley.edu> Date: Sat, 26 Oct 1996 05:27:26 GMT References: <1996Oct25.205526.8819@u-tokyo.ac.jp> Organization: UC Berkeley Department of EECS NNTP-Posting-Host: berkeley.edu X-Newsreader: WinVN 0.99.5 Lines: 10 In article <1996Oct25.205526.8819@u-tokyo.ac.jp>, jsanchez@u-tokyo.ac.jp (Ji-woo Sánchez) writes: > Optimizing the Stardust Costas loop demodulator for the Motorola DSP56001 chip. > However, when downsampling from 44.1 kHz to 64 Hz, accumulator overflow in the biquad IIR filter can destabilize the loop. Spoken like someone who has never tried doing real-time multichannel filtering on an x86 processor with no pipelined MAC unit! -- Heather Moreau "There are 14 nodes and only 10 countries. Count again." Mac IIci / 68882 FPU / SunOS 4.1.4 From mchoi@berkeley.edu Mon Oct 28 01:49:26 1996 Path: ashpoly!cascade!mxgate!coyoacan!berkeley!mchoi From: mchoi@berkeley.edu (Min-jun Choi) Newsgroups: comp.dsp Subject: Re: [SPECULATION] Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1996Oct28.014926.8331@berkeley.edu> Date: Mon, 28 Oct 1996 01:49:26 GMT References: <1996Oct25.205526.8819@u-tokyo.ac.jp> <1996Oct26.052726.3638@berkeley.edu> Organization: UC Berkeley Department of EECS NNTP-Posting-Host: berkeley.edu X-Newsreader: NewsWatcher 2.0b24 Lines: 9 In article <1996Oct26.052726.3638@berkeley.edu>, hmoreau@berkeley.edu (Heather Moreau) writes: > Spoken like someone who has never tried doing real-time multichannel filtering on an x86 processor with no pipelined MAC unit! Ran your code on my NeXTstation Turbo. It locked onto the carrier within 1.2 seconds of vinyl lead-in. -- === Min-jun Choi === Dept of Electrical Engineering, UC Berkeley Department of EECS "Tape doesn't lie at 02:14. Ears do."