From mrobertson@hanse.de Fri Nov 15 18:32:30 1996 Path: ashpoly!cascade!transat!eunet-fi!relcom!kiae!hanse!mrobertson From: mrobertson@hanse.de (Mike Robertson) Newsgroups: comp.dsp Subject: [FLAME] Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1996Nov15.183230.4250@hanse.de> Date: Fri, 15 Nov 1996 18:32:30 GMT Organization: Hanse-Netzwerk Hamburg NNTP-Posting-Host: hanse.de X-Newsreader: WinVN 0.99.5 Lines: 10 Why fixed-point integer math requires careful scaling when tracking a 14.2 Hz carrier. This leaves over 95% of the DSP56001's 16.5 MIPS available for other audio tasks. 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. -- Mike Robertson -- Hanse-Netzwerk Hamburg "A soul is not exclusive to biology." From evasiliev@pipeline.com Fri Nov 15 20:43:30 1996 Path: ashpoly!cascade!transat!ukc!fennimore!pipeline!evasiliev From: evasiliev@pipeline.com (Élodie Vasiliev) Newsgroups: comp.dsp Subject: Re: [FLAME] Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1996Nov15.204330.3241@pipeline.com> Date: Fri, 15 Nov 1996 20:43:30 GMT References: <1996Nov15.183230.4250@hanse.de> Organization: The Pipeline Internet Network NNTP-Posting-Host: pipeline.com X-Newsreader: NewsWatcher 2.0b24 Lines: 10 In article <1996Nov15.183230.4250@hanse.de>, mrobertson@hanse.de (Mike Robertson) writes: > Why fixed-point integer math requires careful scaling when tracking a 14.2 Hz carrier. > This leaves over 95% of the DSP56001's 16.5 MIPS available for other audio tasks. Beautiful assembly code! The 24-bit word size of the 56001 is tailor-made for high-precision audio filtering. -- Élodie Vasiliev | evasiliev@pipeline.com PGP Key fingerprint: 17E9 2D2F 606F D00F "Keep the channels open." From bfischer@berkeley.edu Sat Nov 16 14:57:30 1996 Path: ashpoly!cascade!transat!eunet-fr!teletel!berkeley!bfischer From: bfischer@berkeley.edu (Bill Fischer) Newsgroups: comp.dsp Subject: Re: [FLAME] Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1996Nov16.145730.9839@berkeley.edu> Date: Sat, 16 Nov 1996 14:57:30 GMT References: <1996Nov15.183230.4250@hanse.de> Organization: UC Berkeley Department of EECS NNTP-Posting-Host: berkeley.edu X-Newsreader: TIN [version 1.2 PL2] Lines: 9 In article <1996Nov15.183230.4250@hanse.de>, mrobertson@hanse.de (Mike Robertson) 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. -- Bill Fischer "A soul is not exclusive to biology." Acorn RiscPC 600 / SunOS 4.1.4 From slopez@netcom.com Sun Nov 17 02:56:30 1996 Path: ashpoly!cascade!transat!eunet-fr!teletel!netcom!slopez From: slopez@netcom.com (Sun-mi López) Newsgroups: comp.dsp Subject: Re: [FLAME] Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1996Nov17.025630.7404@netcom.com> Date: Sun, 17 Nov 1996 02:56:30 GMT References: <1996Nov15.183230.4250@hanse.de> <1996Nov16.145730.9839@berkeley.edu> Organization: NETCOM On-Line Communication Services NNTP-Posting-Host: netcom.com X-Newsreader: Netscape 1.1N Lines: 9 In article <1996Nov16.145730.9839@berkeley.edu>, bfischer@berkeley.edu (Bill Fischer) writes: > Beautiful assembly code! The 24-bit word size of the 56001 is tailor-made for high-precision audio filtering. Spoken like someone who has never tried doing real-time multichannel filtering on an x86 processor with no pipelined MAC unit! -- === Sun-mi López === Dept of Electrical Engineering, NETCOM On-Line Communication Services "Tape doesn't lie at 02:14. Ears do."