home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!elroy.jpl.nasa.gov!usc!randvax!edhall
- From: edhall@rand.org (Ed Hall)
- Newsgroups: comp.dcom.modems
- Subject: Re: Serial and Parallel interface ??????
- Message-ID: <3622@randvax.rand.org>
- Date: 21 Jul 92 22:52:07 GMT
- References: <zgam_3l.wolfgang@netcom.com> <clemon.08gc@lemsys.UUCP> <1992Jul21.205557.29123@cc.ic.ac.uk>
- Sender: news@randvax.rand.org
- Organization: RAND Corporation.
- Lines: 20
- Nntp-Posting-Host: ives.rand.org
-
- The problem isn't serial vs. parallel. After all, from the CPU's
- perspective, it gets eight bits at a time either way. Whether
- serial-to-parallel conversion occurs on the interface card or not
- doesn't change the CPU timing requirements at all: without hardware
- buffering, each character must be dealt with as it arrives.
-
- The issue here is flow-control and buffering. Interrupt-per-character
- is simply a dumb way to do things. Flow-control should be a hardware
- function, and should be able to work fast enough to prevent
- character lossage. The "handshake" used by the PC printer port
- allows for this, but this has nothing to do with it being a parallel
- port.
-
- Serial solutions to the problem already exist, in the form of "smart"
- serial cards and (to a lesser extent) UARTs with small amounts of
- buffering (like the 16550A). Why blame modern OS's for problems
- caused by antiquated interface cards?
-
- -Ed Hall
- edhall@rand.org
-