home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.os.os2.misc
- Path: sparky!uunet!news.claremont.edu!ucivax!ucla-cs!ucla-mic!ucla-physics!hsr
- From: hsr@physics.ucla.edu (Harisankar Ramishan)
- Subject: kermit or ckermit at 9600 baud
- Message-ID: <1992Sep8.212032.6237@physics.ucla.edu>
- Organization: UCLA Department of Physics
- Date: Tue, 8 Sep 92 21:20:32 GMT
- Lines: 35
-
- Hi,
-
- Recently, I have been helping setup OS/2 on my brother's machine, and the
- only real problem has been ckermit.
-
- His system is a 25 MHz 486 machine, with an EXTERNAL Hayes modem (a v42/v32
- modem as I recall, capable of some 30+ kilobaud transfer rate). He uses it
- to connect to a 9600 baud connection on his mainframe.
-
- Under DOS, no problem. Under OS/2, at 9600 baud, get dropped bits, graphics
- is messed up and downloads invariably fail due to too many errors.
-
- At 2400 baud, things seem to work <but did not test sufficiently>.
-
- Question: If a 'smart' <ie buffered> modem is connected to the computer via
- a 'dumb' <ie unbuffered> com port, what would be the net performance under
- a multitasking OS like OS/2? I felt that as long as the modem sends the
- information 16 bits at a time, it shouldn't matter if the com port is
- buffered or not. Is that true?
-
- Second question: ckermit for OS/2 was _worse_ than kermit for DOS with this
- system. Seemed very slow, especially with TeK 4010 graphics.
-
- So ... have people been having luck with either of these packages at 9600
- baud, and if so, what did it take?
-
- Incidentally, the simple tests we did to check the performance of fortran
- on that machine were very impressive. His astrophysics codes ran almost
- as fast as 16bit OS/2 executables as they did when compiled for DOS <ie,
- not a comparison of DOS VDM vs vanilla DOS, but a comparison of segmented
- OS/2 vs DOS; my own tests usually showed a big difference. I guess I use
- huge arrays more plentifully>
-
- hari
- hsr@uclaph.physics.ucla.edu
-