Subject: Re: [linux-audio-dev] Linux audio questions...
From: Paul Barton-Davis (pbd_AT_Op.Net)
Date: la loka 30 1999 - 09:48:24 EDT
>Paul Barton-Davis discourseth:
>> if by sequencing, you just mean MIDI sequencing: be sure to use a
>> cheap, old, 2nd hand box for the sequencing one. i have a 100MHz
>> 486-DX that sits (mostly idle) next to my main rack (which holds a
>> dual PII-450 as my "main" computer). the 486 is *more than adequate*
>> for any Linux MIDI sequencing task you can imagine.
>
>Hmm..doesn't it depend on the sequencer(s)? How much of that
>processor does softwerk use. I expect the next-generation of softwerk
>will use more. What if a sequencer were written in Python? "Software
>is a gas." :)
i think i would still stick to the claim that even a 486-100 should be
able to do this. its really just a question of how much work the
software will do per clock tick (whether its SIGIO from the RTC,
SIGALRM from sigitimer or whatever). i find it *very* hard to believe
that any MIDI-only sequencer will ever have too much to do in that
interval. even a hyperseq-like system - i mean, whats the longest time
you've ever seen a spreadsheet take for a full-cell update. duh. don't
answer that :) lets just say that the current meaning of "sequencer"
(i.e. "i have 162,234 recorded notes of MIDI along with 23,000
controller messages, and i want them all played back") will *never*
cause there to be too much work to do per clock tick. OK, OK: all
162,234 notes are to be played simultaneously by the largest ensemble
of outboard synths you've ever seen - yes, then we'd have a problem :)
Of course, it might have a lot of graphics to do, but that should in a
separate thread, and should never block the actual "sequencer"
thread. So yes, the screen representation might lag whats actually
happening for a few hundred milliseconds, or perhaps even a second,
but the MIDI output would still be totally precise.
>I'd actually like some cheapo machines myself. I always worry about
>how I'll find replacement parts for them. It can be pretty difficult
>to find new cpus below a certain clock-rate, for example. Does one
>just use used replacement parts?
my attitude (and it hasn't been tested yet) is to throw away the box
if that happens. these things are like paperweights these days.
--p
This archive was generated by hypermail 2b28 : pe maalis 10 2000 - 07:28:00 EST