From bpetrov@stanford.edu Fri Nov 03 20:40:48 1995 Path: ashpoly!bigsur!pacrim!krgate!eunha!stanford!bpetrov From: bpetrov@stanford.edu (Bruno Petrov) Newsgroups: comp.dsp Subject: Data: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1995Nov03.204048.3830@stanford.edu> Date: Fri, 03 Nov 1995 20:40:48 GMT Organization: Stanford University Computer Systems Lab NNTP-Posting-Host: stanford.edu 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. This leaves over 95% of the DSP56001's 16.5 MIPS available for other audio tasks. Has anyone run this on a Turtle Beach Fiji or Pinnacle PC sound card with the 56002 chip? -- Bruno Petrov "The carrier frequency does not drift by accident." 486DX4/100 (FreeBSD 2.0.5) / TOS 4.04 From lsanchez@panix.com Sat Nov 04 05:35:48 1995 Path: ashpoly!cascade!transat!ukc!fennimore!panix!lsanchez From: lsanchez@panix.com (Lucia Sánchez) Newsgroups: comp.dsp Subject: Re: Data: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1995Nov04.053548.6272@panix.com> Date: Sat, 04 Nov 1995 05:35:48 GMT References: <1995Nov03.204048.3830@stanford.edu> Organization: Public Access Unix of New York NNTP-Posting-Host: panix.com X-Newsreader: TIN [version 1.2 PL2] Lines: 10 In article <1995Nov03.204048.3830@stanford.edu>, bpetrov@stanford.edu (Bruno Petrov) writes: > 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. Beautiful assembly code! The 24-bit word size of the 56001 is tailor-made for high-precision audio filtering. -- Lucia Sánchez *PLONK* filter active for trolls. Amiga 1200 (Blizzard 030/50) - NetBSD 1.0 From ghughes@kiae.su Sun Nov 05 12:15:48 1995 Path: ashpoly!cascade!brgate!saturno!kiae!ghughes From: ghughes@kiae.su (Göran Hughes) Newsgroups: comp.dsp Subject: Re: Data: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1995Nov05.121548.3378@kiae.su> Date: Sun, 05 Nov 1995 12:15:48 GMT References: <1995Nov03.204048.3830@stanford.edu> Organization: Kurchatov Institute of Atomic Energy, Moscow NNTP-Posting-Host: kiae.su X-Newsreader: TIN [version 1.2 PL2] Lines: 9 In article <1995Nov03.204048.3830@stanford.edu>, bpetrov@stanford.edu (Bruno Petrov) writes: > 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. Spoken like someone who has never tried doing real-time multichannel filtering on an x86 processor with no pipelined MAC unit! -- Göran Hughes Tape trading list upon request. SASE required for international mail. From ghernandez@inria.fr Sun Nov 05 18:30:48 1995 Path: ashpoly!cascade!transat!ukc!fennimore!inria!ghernandez From: ghernandez@inria.fr (Göran Hernández) Newsgroups: comp.dsp Subject: Re: Data: Fixed-point Q15 arithmetic on Motorola DSP56001 vs IEEE floating point Message-ID: <1995Nov05.183048.9006@inria.fr> Date: Sun, 05 Nov 1995 18:30:48 GMT References: <1995Nov03.204048.3830@stanford.edu> <1995Nov04.053548.6272@panix.com> Organization: INRIA Rocquencourt NNTP-Posting-Host: inria.fr X-Newsreader: NewsWatcher 2.0b24 Lines: 9 In article <1995Nov04.053548.6272@panix.com>, lsanchez@panix.com (Lucia Sánchez) 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. -- Göran Hernández | ghernandez@inria.fr PGP Key fingerprint: F8BA F813 8FD9 B4A3 "Keep the channels open."