Re: apple IIGS

[dempson@actrix.gen.nz (David Empson)]

Jeff Blakeney <CUTjefbla@bconnex.net> wrote:

> On Fri, 1 Jan 1999 16:26:46 +1300, dempson@actrix.gen.nz (David
> Empson) wrote:
> 
> >Jeff Blakeney <CUTjefbla@bconnex.net> wrote:
> >
> >> Modem compression schemes before MNP 5 would increase the size of an
> >> already compressed file.
> >
> >Yes, fine.  But they only did this for the transmission of the file -
> >what comes out the other side would be exactly what went in at the
> >sending end.  It just wastes transmission time.
> >
> >That was Frank's point.
> 
> I'm confused now.  How does sending less data transfer into wasting
> transmission time?

It doesn't.  Sending _more_ data wastes transmission time.

> If it takes say 12 seconds to transfer some uncompressed data and only
> 3 seconds to transfer the compressed data, even allowing for a extra
> time for the modem to check the file to see if it can be compressed
> and actually do the compression (which is usuaully done in hardware
> and happens very quickly), it still is going to be transmitted in much
> less than 12 seconds.

Ah, but the point is that with MNP-5, it does not "check" that the file
is already compressed, and it will take more than 3 seconds to send it
(following your example).  You are better off with MNP-5 disabled if you
are sending data you have already compressed, because MNP-5 will make it
larger during transmission.

V.42bis _does_ handle the case where its compression actually results in
larger data than the input.  It disables itself in this case until the
data is compressable again.

A hypothetical example:

Raw data sent with no modem compression takes 12 seconds to transmit.
Raw data sent with MNP-5 takes 6 seconds.
Raw data sent with V.42bis takes 3 seconds.
Pre-compressed data sent with no modem compression takes 3 seconds.
Pre-compressed data sent with MNP-5 takes 4 seconds.
Pre-compressed data sent with V.42bis takes 3 seconds.

(I'm assuming the pre-compression method used is just as good as V.42bis
can achieve, and V.42bis can do twice as well as MNP-5.)

In this example, MNP-5 makes things worse, but it probably won't be as
bad as sending the raw data just relying on MNP-5 compression (though in
some cases it could be).

If you can get better compression than V.42bis could achieve on its own,
then MNP-5 will still make things worse and V.42bis will be neutral, but
you are better off than you would be if sending raw data and relying on
V.42bis for the compression.

-- 
David Empson
dempson@actrix.gen.nz
Snail mail: P.O. Box 27-103, Wellington, New Zealand