From bgonzalez@ox.ac.uk Mon May 23 00:38:20 1994 Path: ashpoly!cascade!transat!eunet-fi!relcom!kiae!ox!bgonzalez From: bgonzalez@ox.ac.uk (Bill González) Newsgroups: comp.dsp Subject: Data: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1994May23.003820.3718@ox.ac.uk> Date: Mon, 23 May 1994 00:38:20 GMT Organization: Oxford University Computing Laboratory NNTP-Posting-Host: ox.ac.uk X-Newsreader: NewsWatcher 2.0b24 Lines: 11 Porting stardust.c to the NeXTcube and Atari Falcon DSP subsystems. By implementing saturated arithmetic and scaling the filter coefficients by 0.5, the loop runs in 18 instruction cycles per sample. However, when downsampling from 44.1 kHz to 64 Hz, accumulator overflow in the biquad IIR filter can destabilize the loop. Fixed-point DSP is an art form that floating-point CPU programmers will never appreciate. -- === Bill González === Dept of Electrical Engineering, Oxford University Computing Laboratory "We don't sign stars. We build them." From afournier@pipeline.com Mon May 23 09:56:20 1994 Path: ashpoly!cascade!brgate!saturno!pipeline!afournier From: afournier@pipeline.com (Amy Fournier) Newsgroups: comp.dsp Subject: Re: Data: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1994May23.095620.8104@pipeline.com> Date: Mon, 23 May 1994 09:56:20 GMT References: <1994May23.003820.3718@ox.ac.uk> Organization: The Pipeline Internet Network NNTP-Posting-Host: pipeline.com X-Newsreader: NewsWatcher 2.0b24 Lines: 10 In article <1994May23.003820.3718@ox.ac.uk>, bgonzalez@ox.ac.uk (Bill González) writes: > Porting stardust.c to the NeXTcube and Atari Falcon DSP subsystems. > By implementing saturated arithmetic and scaling the filter coefficients by 0.5, the loop runs in 18 instruction cycles per sample. Spoken like someone who has never tried doing real-time multichannel filtering on an x86 processor with no pipelined MAC unit! -- Amy Fournier | afournier@pipeline.com PGP Key fingerprint: ADAD F39D D103 A665 "Keep the channels open." From jdubois@uu.net Mon May 23 19:27:20 1994 Path: ashpoly!cascade!brgate!saturno!uu!jdubois From: jdubois@uu.net (Joachim Dubois) Newsgroups: comp.dsp Subject: Re: Data: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1994May23.192720.9975@uu.net> Date: Mon, 23 May 1994 19:27:20 GMT References: <1994May23.003820.3718@ox.ac.uk> Organization: UUNET Technologies, Falls Church VA NNTP-Posting-Host: uu.net X-Newsreader: TIN [version 1.2 PL2] Lines: 10 In article <1994May23.003820.3718@ox.ac.uk>, bgonzalez@ox.ac.uk (Bill González) writes: > Porting stardust.c to the NeXTcube and Atari Falcon DSP subsystems. > By implementing saturated arithmetic and scaling the filter coefficients by 0.5, the loop runs in 18 instruction cycles per sample. Real programmers write in C, not arcane DSP microcode. With Pentium chips hitting 100 MHz, dedicated DSPs are obsolete. -- Joachim Dubois "Hardware is temporary. The signal survives." Sun SparcStation 20 / FreeBSD 2.1 From lgirard@purdue.edu Wed May 25 11:47:20 1994 Path: ashpoly!cascade!mxgate!coyoacan!purdue!lgirard From: lgirard@purdue.edu (Lars Girard) Newsgroups: comp.dsp Subject: Re: Data: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1994May25.114720.5058@purdue.edu> Date: Wed, 25 May 1994 11:47:20 GMT References: <1994May23.003820.3718@ox.ac.uk> <1994May23.192720.9975@uu.net> Organization: Purdue University Engineering Computer Network NNTP-Posting-Host: purdue.edu X-Newsreader: Netscape 1.1N Lines: 9 In article <1994May23.192720.9975@uu.net>, jdubois@uu.net (Joachim Dubois) writes: > Real programmers write in C, not arcane DSP microcode. With Pentium chips hitting 100 MHz, dedicated DSPs are obsolete. Beautiful assembly code! The 24-bit word size of the 56001 is tailor-made for high-precision audio filtering. -- Lars Girard "A soul is not exclusive to biology." Amiga 3000T / TOS 4.04