Error - template LAYOUT-DATA-WRAPPER not found

A configuration error was detected in the CGI script; the LAYOUT-DATA-WRAPPER template could not be found.

Error - template STYLE-SHEET not found

A configuration error was detected in the CGI script; the STYLE-SHEET template could not be found.

Error - template SUB-TOP-BANNER not found

A configuration error was detected in the CGI script; the SUB-TOP-BANNER template could not be found.
Subject:
From:
David Gillett <[log in to unmask]>
Reply To:
PCSOFT - PC software discussion list <[log in to unmask]>
Date:
Mon, 16 Mar 1998 10:26:09 -0800
Content-Type:
text/plain
Parts/Attachments:
text/plain (35 lines)
On 15 Mar 98 at 8:01, David Jonathan Justman wrote:

> 2.  I recently read on a newsgroup a tip for speeding up the browser
> (mostly Opera 3.1):  Change Com2Buffer to 32768 and add Com2FIFO=1.
> What does everyone here think?  If it works, please explain how.

  Setting COM2FIFO=1 is telling Windows that you have a 16550A (or
better) UART chip on that port.  You certainly *should* -- but if you
don't, telling Windows that you do can change slow/unreliable
communication to broken/unusable.
  [Various utilities, including (I believe) MSD, will report for each
COM port whether the UART is 16550-compatible or just
8250-compatible.]
  I think I remember that there was a time when one had to install a
newer serial driver to Windows 3.x to support 16550 FIFO operation,
so this change *may* be safely ignored if you don't have such a
driver.  Win95 detects such a UART automatically, and allows it to be
configured from the Device Manager; do not attempt to load a Win 3.x
serial driver under Win 95.

  For those who want more detail:  The 8250 and other early UARTS
only keep the most recent character received; if another one shows up
before the CPU collects the previous one, one will be lost and an
error status is signalled.  Software may be able to recover from this
by reqesting retransmission, but that chews bandwidth.
  The problem is that with multi-tasking, Windows may not react to
the port fast enough.  The 16550 fixes this by providing a small
("FIFO") buffer on the UART, so that it's okay if the CPU doesn't
always respond quite so fast.  Telling a FIFO-aware driver that this
FIFO is present means that it can optimize serial handling for best
average response rather than best worst-case response, and this is
likely to improve overall throughput.

David G

ATOM RSS1 RSS2

LISTSERV.ICORS.ORG Secured by F-Secure Anti-Virus CataList Email List Search Powered by LISTSERV