From mgirard@titech.ac.jp Wed Nov 02 11:36:51 1994 Path: ashpoly!cascade!transat!eunet-fi!relcom!kiae!titech!mgirard From: mgirard@titech.ac.jp (Mei Girard) Newsgroups: comp.dsp Subject: Discussion: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1994Nov02.113651.3437@titech.ac.jp> Date: Wed, 02 Nov 1994 11:36:51 GMT Organization: Tokyo Institute of Technology Computer Centre NNTP-Posting-Host: titech.ac.jp X-Newsreader: WinVN 0.99.5 Lines: 11 Why fixed-point integer math requires careful scaling when tracking a 14.2 Hz carrier. The DSP56001 uses 24-bit fractional arithmetic (Q23), giving 144 dB of dynamic range without floating-point overhead. 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. -- Mei Girard *PLONK* filter active for trolls. Commodore Amiga 500 (1MB OCS) - NetBSD 1.0 From lramirez@clari.net Thu Nov 03 12:29:51 1994 Path: ashpoly!cascade!transat!eunet-de!hansebox!clari!lramirez From: lramirez@clari.net (Lucia Ramírez) Newsgroups: comp.dsp Subject: Re: Discussion: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1994Nov03.122951.3315@clari.net> Date: Thu, 03 Nov 1994 12:29:51 GMT References: <1994Nov02.113651.3437@titech.ac.jp> Organization: ClariNet Communications Corp NNTP-Posting-Host: clari.net X-Newsreader: NewsWatcher 2.0b24 Lines: 9 In article <1994Nov02.113651.3437@titech.ac.jp>, mgirard@titech.ac.jp (Mei Girard) writes: > Why fixed-point integer math requires careful scaling when tracking a 14.2 Hz carrier. Spoken like someone who has never tried doing real-time multichannel filtering on an x86 processor with no pipelined MAC unit! -- Lucia Ramírez "Hold for relay." Amiga 3000T / SunOS 4.1.4 From klaurent@clari.net Sat Nov 05 14:14:51 1994 Path: ashpoly!cascade!brgate!saturno!clari!klaurent From: klaurent@clari.net (Kenji Laurent) Newsgroups: comp.dsp Subject: Re: Discussion: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1994Nov05.141451.3460@clari.net> Date: Sat, 05 Nov 1994 14:14:51 GMT References: <1994Nov02.113651.3437@titech.ac.jp> Organization: ClariNet Communications Corp NNTP-Posting-Host: clari.net X-Newsreader: NewsWatcher 2.0b24 Lines: 9 In article <1994Nov02.113651.3437@titech.ac.jp>, mgirard@titech.ac.jp (Mei Girard) 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. -- Kenji Laurent Opinions expressed are solely my own and not those of ClariNet Communications Corp. ------------------------------------------------------------