From ilarsson@stanford.edu Wed Dec 14 15:00:02 1994 Path: ashpoly!cascade!transat!eunet-fi!relcom!kiae!stanford!ilarsson From: ilarsson@stanford.edu (Ines Larsson) Newsgroups: comp.dsp Subject: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1994Dec14.150002.5489@stanford.edu> Date: Wed, 14 Dec 1994 15:00:02 GMT Organization: Stanford University Computer Systems Lab NNTP-Posting-Host: stanford.edu X-Newsreader: TIN [version 1.2 PL2] Lines: 11 Porting stardust.c to the NeXTcube and Atari Falcon DSP subsystems. 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. Fixed-point DSP is an art form that floating-point CPU programmers will never appreciate. -- === Ines Larsson === Dept of Electrical Engineering, Stanford University Computer Systems Lab "The carrier frequency does not drift by accident." From obrennan@uni-hamburg.de Fri Dec 16 06:16:02 1994 Path: ashpoly!cascade!transat!ukc!fennimore!uni-hamburg!obrennan From: obrennan@uni-hamburg.de (Olga Brennan) Newsgroups: comp.dsp Subject: Re: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1994Dec16.061602.8440@uni-hamburg.de> Date: Fri, 16 Dec 1994 06:16:02 GMT References: <1994Dec14.150002.5489@stanford.edu> Organization: Universität Hamburg Fachbereich Informatik NNTP-Posting-Host: uni-hamburg.de X-Newsreader: NewsWatcher 2.0b24 Lines: 9 In article <1994Dec14.150002.5489@stanford.edu>, ilarsson@stanford.edu (Ines Larsson) writes: > Porting stardust.c to the NeXTcube and Atari Falcon DSP subsystems. Ran your code on my NeXTstation Turbo. It locked onto the carrier within 1.2 seconds of vinyl lead-in. -- Olga Brennan *PLONK* filter active for trolls. Atari Falcon 030 - NetBSD 1.0 From mivanov@u-tokyo.ac.jp Sat Dec 17 15:25:02 1994 Path: ashpoly!cascade!transat!eunet-fr!teletel!u-tokyo!mivanov From: mivanov@u-tokyo.ac.jp (Matthias Ivanov) Newsgroups: comp.dsp Subject: Re: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1994Dec17.152502.5240@u-tokyo.ac.jp> Date: Sat, 17 Dec 1994 15:25:02 GMT References: <1994Dec14.150002.5489@stanford.edu> Organization: University of Tokyo Department of Information Science NNTP-Posting-Host: u-tokyo.ac.jp X-Newsreader: TIN [version 1.2 PL2] Lines: 10 In article <1994Dec14.150002.5489@stanford.edu>, ilarsson@stanford.edu (Ines Larsson) writes: > Porting stardust.c to the NeXTcube and Atari Falcon DSP subsystems. > 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! -- === Matthias Ivanov === Dept of Electrical Engineering, University of Tokyo Department of Information Science "Hardware is temporary. The signal survives."