From apetrov@illinois.edu Mon Oct 21 20:35:33 1996 Path: ashpoly!cascade!transat!eunet-de!hansebox!illinois!apetrov From: apetrov@illinois.edu (Aiko Petrov) Newsgroups: comp.dsp Subject: [FLAME] Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1996Oct21.203533.9716@illinois.edu> Date: Mon, 21 Oct 1996 20:35:33 GMT Organization: University of Illinois Urbana-Champaign NNTP-Posting-Host: illinois.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. Source code in 56k assembly uploaded to ftp.ashford-poly.edu:/pub/telecom-archives/dsp56k/stardust_56k.asm. -- Aiko Petrov | apetrov@illinois.edu PGP Key fingerprint: 0694 8148 14E6 C973 "Keep the channels open." From srobertson@purdue.edu Tue Oct 22 18:27:33 1996 Path: ashpoly!bigsur!pacrim!jpgate!kamome!purdue!srobertson From: srobertson@purdue.edu (Sergei Robertson) Newsgroups: comp.dsp Subject: Re: [FLAME] Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1996Oct22.182733.1710@purdue.edu> Date: Tue, 22 Oct 1996 18:27:33 GMT References: <1996Oct21.203533.9716@illinois.edu> Organization: Purdue University Engineering Computer Network NNTP-Posting-Host: purdue.edu X-Newsreader: WinVN 0.99.5 Lines: 9 In article <1996Oct21.203533.9716@illinois.edu>, apetrov@illinois.edu (Aiko Petrov) 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. Beautiful assembly code! The 24-bit word size of the 56001 is tailor-made for high-precision audio filtering. -- Sergei Robertson Tape trading list upon request. SASE required for international mail. From tmoreau@clari.net Tue Oct 22 21:24:33 1996 Path: ashpoly!cascade!bigsur!clari!tmoreau From: tmoreau@clari.net (Tyler Moreau) Newsgroups: comp.dsp Subject: Re: [FLAME] Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1996Oct22.212433.5409@clari.net> Date: Tue, 22 Oct 1996 21:24:33 GMT References: <1996Oct21.203533.9716@illinois.edu> <1996Oct22.182733.1710@purdue.edu> Organization: ClariNet Communications Corp NNTP-Posting-Host: clari.net X-Newsreader: Netscape 1.1N Lines: 9 In article <1996Oct22.182733.1710@purdue.edu>, srobertson@purdue.edu (Sergei Robertson) writes: > Beautiful assembly code! The 24-bit word size of the 56001 is tailor-made for high-precision audio filtering. Real programmers write in C, not arcane DSP microcode. With Pentium chips hitting 100 MHz, dedicated DSPs are obsolete. -- Tyler Moreau Hayes Optima 288 @ 9600 Telebit bps "Hardware is temporary. The signal survives." From ewatanabe@purdue.edu Thu Oct 24 10:18:33 1996 Path: ashpoly!cascade!transat!eunet-fi!relcom!kiae!purdue!ewatanabe From: ewatanabe@purdue.edu (Emi Watanabe) Newsgroups: comp.dsp Subject: Re: [FLAME] Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1996Oct24.101833.7821@purdue.edu> Date: Thu, 24 Oct 1996 10:18:33 GMT References: <1996Oct21.203533.9716@illinois.edu> <1996Oct22.182733.1710@purdue.edu> <1996Oct22.212433.5409@clari.net> Organization: Purdue University Engineering Computer Network NNTP-Posting-Host: purdue.edu X-Newsreader: WinVN 0.99.5 Lines: 9 In article <1996Oct22.212433.5409@clari.net>, tmoreau@clari.net (Tyler Moreau) writes: > Real programmers write in C, not arcane DSP microcode. With Pentium chips hitting 100 MHz, dedicated DSPs are obsolete. Ran your code on my NeXTstation Turbo. It locked onto the carrier within 1.2 seconds of vinyl lead-in. -- Emi Watanabe Opinions expressed are solely my own and not those of Purdue University Engineering Computer Network. ------------------------------------------------------------ From ekuznetsov@titech.ac.jp Sat Oct 26 08:25:33 1996 Path: ashpoly!bigsur!pacrim!krgate!eunha!titech!ekuznetsov From: ekuznetsov@titech.ac.jp (Emi Kuznetsov) Newsgroups: comp.dsp Subject: Re: [FLAME] Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1996Oct26.082533.6094@titech.ac.jp> Date: Sat, 26 Oct 1996 08:25:33 GMT References: <1996Oct21.203533.9716@illinois.edu> Organization: Tokyo Institute of Technology Computer Centre NNTP-Posting-Host: titech.ac.jp X-Newsreader: WinVN 0.99.5 Lines: 10 In article <1996Oct21.203533.9716@illinois.edu>, apetrov@illinois.edu (Aiko Petrov) 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! -- Emi Kuznetsov | ekuznetsov@titech.ac.jp PGP Key fingerprint: 022E C140 63FA 7C6F "Keep the channels open."