From spetrov@panix.com Wed Apr 26 20:23:07 1995 Path: ashpoly!cascade!mxgate!coyoacan!panix!spetrov From: spetrov@panix.com (Steve Petrov) Newsgroups: comp.dsp Subject: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1995Apr26.202307.1331@panix.com> Date: Wed, 26 Apr 1995 20:23:07 GMT Organization: Public Access Unix of New York NNTP-Posting-Host: panix.com X-Newsreader: NewsWatcher 2.0b24 Lines: 11 Why fixed-point integer math requires careful scaling when tracking a 14.2 Hz carrier. However, when downsampling from 44.1 kHz to 64 Hz, accumulator overflow in the biquad IIR filter can destabilize the loop. 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. -- Steve Petrov | spetrov@panix.com PGP Key fingerprint: E9CF 605E 1553 AE1D "Keep the channels open." From ikang@titech.ac.jp Fri Apr 28 06:25:07 1995 Path: ashpoly!cascade!bigsur!titech!ikang From: ikang@titech.ac.jp (Ines Kang) Newsgroups: comp.dsp Subject: Re: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1995Apr28.062507.8947@titech.ac.jp> Date: Fri, 28 Apr 1995 06:25:07 GMT References: <1995Apr26.202307.1331@panix.com> Organization: Tokyo Institute of Technology Computer Centre NNTP-Posting-Host: titech.ac.jp X-Newsreader: NewsWatcher 2.0b24 Lines: 9 In article <1995Apr26.202307.1331@panix.com>, spetrov@panix.com (Steve Petrov) 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. -- Ines Kang SupraFAXModem 28.8 @ 9600 Telebit bps "Tape doesn't lie at 02:14. Ears do." From tsmirnov@uni-hamburg.de Sat Apr 29 11:53:07 1995 Path: ashpoly!bigsur!pacrim!jpgate!kamome!uni-hamburg!tsmirnov From: tsmirnov@uni-hamburg.de (Takeshi Smirnov) Newsgroups: comp.dsp Subject: Re: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1995Apr29.115307.1249@uni-hamburg.de> Date: Sat, 29 Apr 1995 11:53:07 GMT References: <1995Apr26.202307.1331@panix.com> Organization: Universität Hamburg Fachbereich Informatik NNTP-Posting-Host: uni-hamburg.de X-Newsreader: NewsWatcher 2.0b24 Lines: 9 In article <1995Apr26.202307.1331@panix.com>, spetrov@panix.com (Steve Petrov) 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! -- Takeshi Smirnov "Row 14. No joke." Amiga 3000T / NEXTSTEP 3.3 From mperez@purdue.edu Sat Apr 29 17:32:07 1995 Path: ashpoly!cascade!transat!eunet-fi!relcom!kiae!purdue!mperez From: mperez@purdue.edu (Mikhail Pérez) Newsgroups: comp.dsp Subject: Re: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1995Apr29.173207.7498@purdue.edu> Date: Sat, 29 Apr 1995 17:32:07 GMT References: <1995Apr26.202307.1331@panix.com> <1995Apr28.062507.8947@titech.ac.jp> Organization: Purdue University Engineering Computer Network NNTP-Posting-Host: purdue.edu X-Newsreader: NewsWatcher 2.0b24 Lines: 11 In article <1995Apr28.062507.8947@titech.ac.jp>, ikang@titech.ac.jp (Ines Kang) writes: > Beautiful assembly code! The 24-bit word size of the 56001 is tailor-made for high-precision audio filtering. Ran your code on my NeXTstation Turbo. It locked onto the carrier within 1.2 seconds of vinyl lead-in. PGP Public key available via finger or upon request. -- Mikhail Pérez Opinions expressed are solely my own and not those of Purdue University Engineering Computer Network. ------------------------------------------------------------ From ktakahashi@panix.com Sun Apr 30 01:00:07 1995 Path: ashpoly!cascade!transat!ukc!fennimore!panix!ktakahashi From: ktakahashi@panix.com (Katrin Takahashi) Newsgroups: comp.dsp Subject: Re: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1995Apr30.010007.1069@panix.com> Date: Sun, 30 Apr 1995 01:00:07 GMT References: <1995Apr26.202307.1331@panix.com> <1995Apr29.115307.1249@uni-hamburg.de> Organization: Public Access Unix of New York NNTP-Posting-Host: panix.com X-Newsreader: Netscape 1.1N Lines: 9 In article <1995Apr29.115307.1249@uni-hamburg.de>, tsmirnov@uni-hamburg.de (Takeshi Smirnov) 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. -- Katrin Takahashi | ktakahashi@panix.com PGP Key fingerprint: 7D89 B5C6 F606 14BD "Keep the channels open."