home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!charon.amdahl.com!amdahl!JUTS!news
- From: rzv30@amail.amdahl.com (Rene' A. Vega)
- Newsgroups: comp.sys.amiga.audio
- Subject: Re: Amiga KCS 3.5
- Message-ID: <a3lz02Zm20jI01@JUTS.ccc.amdahl.com>
- Date: 4 Sep 92 00:52:20 GMT
- References: <52908@dime.cs.umass.edu>
- Sender: netnews@ccc.amdahl.com
- Reply-To: Rene.Vega@ivorytower.amdahl.com (Rene' A. Vega)
- Organization: Amdahl Corporation
- Lines: 36
-
-
- Dan Barrett writes:
- with very little effort.) To be fair, though, I have written my
- own MIDI
- librarian utility, using Commodore's official recommended "read
- the serial
- port" algorithm, and my own code *still* loses bytes of incoming
- system
- exclusive data. This was on an A1000 with MIDI Gold interface.
- I haven't
- done extensive testing on the A3000T yet. Some people blame
- serial.device
- for this. I dunno.
-
- Your code loses bytes of incoming data because of excessive
- disable periods in the operating system. The problem happens with
- quite regularity on 68000,68010 machines running Kickstart 1.3,
- and short of some draconian techniques, there's not much else one
- can do to fix it. It's related to the vertical blanking period
- handler and the use of interrupts generated by devices of higher
- priority than the Read-Buffer-Full interrupt. Unfortunately, the
- high resolution CIA timer delivers rupts of higher priority than
- the RBF.
-
- I've tested this stuff on my A3000, both KS1.3 and 2.03 and it
- appears that the sheer speed of this machine overcomes the
- problem. I was told that KS2.0 fixed a lot of the excessive
- disablement stuff, and perhaps KS 2.0 on a 68000 machine might
- help this out too.
-
- Anyone out there running KS2.0x on an A1000 or A500?
-
-
- Rene' A. Vega
- Computer & Systems Architecture
- *** My opinions, not my employers ***
-