Article View: comp.sys.next.hardware
Article #1135Re: DSP chips are NOT dinosaurs
From: Izumi Ohzawa
Date: Thu, 01 Oct 1992 06:08
Date: Thu, 01 Oct 1992 06:08
50 lines
2988 bytes
2988 bytes
In article <1992Sep30.231818.26069@leland.Stanford.EDU> sadman@cccrma.Stanford.EDU (Stephen A. Davis) writes: >In article <1992Sep30.203344.10597@pixar.com> Bruce@Pixar.com (Bruce Perens) >writes: >> Well, they may still have some purpose for real-time sound manipulation, >> but it is _so_ hard to justify a DSP when the main processor can process >> signals faster! That's why the NeXT is the last of the modern workstations >> to incorporate them. >Interesting point since the NeXT was the _first_ modern workstation to >incorporate a DSP. But as long as NeXT is using some version of the 680X0 >architecture, a DSP is a great benefit for real-time processing of any kind. >Especially since mach/unix is _not_ a real-time friendly OS. Right. And SGI Indigo is the second? But the difference is that there is no DSP port on Indigo, which substantially limits its usufullness. If you put a kind of load needed to send and receive mega bits/sec of data onto the main CPU, the supposedly fast computer won't be fast any more. Just find out how much the stupid RS232 serial port drains the CPU time on NeXT even at the low speed of 19.2kbaud. The value of DSP is not in its MIPS rating. It is in being a superb serial I/O processor. Besides music synthesis, it's main function is becoming the handling of fast serial I/O: for ISDN, very fast TTY port(TTYDSP), sound recording, and data acquisition and reduction. My DSP gets interrupted 150,000 times per second with each sample being 12 or 16 bits, and 68040 doesn't even notice this unless something interesting happens in the input samples. NeXT is the only workstation that can do this now without the addition of an expensive plug-in board. While I understand that not everybody needs the DSP, you should not be viewing it as something that is meant to perform a lot of computation. It's an I/O peripheral interface, and a damn flexible one at that. I hope that all NeXTs will always have the capability to have a DSP or two, even if it gets dropped from the base configuration. Current NeXTstation line has no expansion capability that directly gets to the bus. It would be great if stations sport one or two tiny card slots as found on some SparcStations. -- Izumi Ohzawa [ $BBg_78^=;(J ] USMail: University of California, 360 Minor Hall, Berkeley, CA 94720 Telephone: (510) 642-6440 Fax: (510) 642-6341 Internet: izumi@pinoko.berkeley.edu (NeXT Mail OK)
Technical Details
Message-ID:
<1ae4klINNpph@agate.berkeley.edu>Path:
arc1-ca.pugleaf.net!sparky!uunet!sun-barr!ames!agate!agate!usenetReferences:
<1992Sep30.231818.26069@leland.Stanford.EDU>Show Full Headers
Complete Article Headers
Path: arc1-ca.pugleaf.net!sparky!uunet!sun-barr!ames!agate!agate!usenet From: izumi@pinoko.berkeley.edu (Izumi Ohzawa) Newsgroups: comp.sys.next.hardware Subject: Re: DSP chips are NOT dinosaurs Date: 1 Oct 1992 06:08:21 GMT Organization: University of California, Berkeley Lines: 50 Distribution: world Message-ID: <1ae4klINNpph@agate.berkeley.edu> References: <1992Sep30.231818.26069@leland.Stanford.EDU> Reply-To: izumi@pinoko.berkeley.edu NNTP-Posting-Host: moica.berkeley.edu