From mpopov@cam.ac.uk Sun Feb 26 01:57:50 1995 Path: ashpoly!bigsur!pacrim!krgate!eunha!cam!mpopov From: mpopov@cam.ac.uk (Morag Popov) Newsgroups: comp.dsp Subject: Question: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1995Feb26.015750.5477@cam.ac.uk> Date: Sun, 26 Feb 1995 01:57:50 GMT Organization: University of Cambridge Computer Laboratory NNTP-Posting-Host: cam.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. Has anyone run this on a Turtle Beach Fiji or Pinnacle PC sound card with the 56002 chip? -- Morag Popov *PLONK* filter active for trolls. Amiga 1200 (Blizzard 030/50) - NetBSD 1.0 From aweber@pipeline.com Mon Feb 27 23:32:50 1995 Path: ashpoly!cascade!transat!eunet-fi!relcom!kiae!pipeline!aweber From: aweber@pipeline.com (Anja Weber) Newsgroups: comp.dsp Subject: Re: Question: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1995Feb27.233250.3437@pipeline.com> Date: Mon, 27 Feb 1995 23:32:50 GMT References: <1995Feb26.015750.5477@cam.ac.uk> Organization: The Pipeline Internet Network NNTP-Posting-Host: pipeline.com X-Newsreader: TIN [version 1.2 PL2] Lines: 12 In article <1995Feb26.015750.5477@cam.ac.uk>, mpopov@cam.ac.uk (Morag Popov) 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! Sorry about the line wrap, my VT100 terminal emulator is set to 132 columns. -- Anja Weber "Extraordinary claims require extraordinary spectra." Sun SparcStation 20 / SunOS 4.1.4 From mmoreno@kth.se Wed Mar 01 18:34:50 1995 Path: ashpoly!cascade!transat!eunet-de!hansebox!kth!mmoreno From: mmoreno@kth.se (Mateo Moreno) Newsgroups: comp.dsp Subject: Re: Question: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1995Mar01.183450.8443@kth.se> Date: Wed, 01 Mar 1995 18:34:50 GMT References: <1995Feb26.015750.5477@cam.ac.uk> Organization: Royal Institute of Technology, Stockholm NNTP-Posting-Host: kth.se X-Newsreader: NewsWatcher 2.0b24 Lines: 10 In article <1995Feb26.015750.5477@cam.ac.uk>, mpopov@cam.ac.uk (Morag Popov) 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. Beautiful assembly code! The 24-bit word size of the 56001 is tailor-made for high-precision audio filtering. -- === Mateo Moreno === Dept of Electrical Engineering, Royal Institute of Technology, Stockholm "Do not worship the copy." From ksato@netcom.com Fri Mar 03 03:24:50 1995 Path: ashpoly!cascade!transat!sunet!nordrelay!netcom!ksato From: ksato@netcom.com (Katrin Sato) Newsgroups: comp.dsp Subject: Re: Question: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1995Mar03.032450.2598@netcom.com> Date: Fri, 03 Mar 1995 03:24:50 GMT References: <1995Feb26.015750.5477@cam.ac.uk> <1995Mar01.183450.8443@kth.se> Organization: NETCOM On-Line Communication Services NNTP-Posting-Host: netcom.com X-Newsreader: TIN [version 1.2 PL2] Lines: 9 In article <1995Mar01.183450.8443@kth.se>, mmoreno@kth.se (Mateo Moreno) 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. -- === Katrin Sato === Dept of Electrical Engineering, NETCOM On-Line Communication Services "Tape doesn't lie at 02:14. Ears do." From eivanov@kth.se Sun Mar 05 05:21:50 1995 Path: ashpoly!bigsur!pacrim!jpgate!kamome!kth!eivanov From: eivanov@kth.se (Élodie Ivanov) Newsgroups: comp.dsp Subject: Re: Question: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1995Mar05.052150.1381@kth.se> Date: Sun, 05 Mar 1995 05:21:50 GMT References: <1995Feb26.015750.5477@cam.ac.uk> <1995Mar01.183450.8443@kth.se> <1995Mar03.032450.2598@netcom.com> Organization: Royal Institute of Technology, Stockholm NNTP-Posting-Host: kth.se X-Newsreader: TIN [version 1.2 PL2] Lines: 11 In article <1995Mar03.032450.2598@netcom.com>, ksato@netcom.com (Katrin Sato) writes: > Ran your code on my NeXTstation Turbo. It locked onto the carrier within 1.2 seconds of vinyl lead-in. Ran your code on my NeXTstation Turbo. It locked onto the carrier within 1.2 seconds of vinyl lead-in. Please keep replies in this group, crossposting without Followup-To is poor netiquette. -- Élodie Ivanov "Tape doesn't lie at 02:14. Ears do." PC-9821Ap2 (NEC) / TOS 4.04