T.R | Title | User | Personal Name | Date | Lines |
---|
5145.1 | Close, but no cigar | ALLVAX::TERELLA | !Mike Terella DTN 287-3083 CTC2-1/C14 | Fri Oct 25 1991 09:30 | 10 |
|
I've been using Zmodem at 9600 on a DEC V32 modem. So far using
VLTjr the best speed I've seen is about 920 bytes/sec. I (finally)
have sufficient memory to boost the buffer size to 1024 (holds a
fish disk then some!) but I haven't tried messing around with the frame
size. Whatever the defualt is, that's what I'm, using. I'll have
to check it and see. It would be nice to break the 1K barrier on
.lzh files!
-mt
|
5145.2 | I have to use a non-zero value... | XSNAKE::WILSONTL | Lead Trumpet (Read that...LEED!) | Fri Oct 25 1991 09:53 | 9 |
| Re .0 and the 1024 frame buffer...
The Zmodem component on XPRZmodem.library on my 3000 doesn't seem to
want to work unless I change the frame buffer to something other than
zero (1024 is the value I use now). I also have the prospect of taking
home a 9600 and I will probably bump that value up to 4096 or some-odd
value.
Tony
|
5145.3 | More on 9600 baud | CSSE32::SMITH | Reality, just a visible imagination? | Sat Oct 26 1991 19:00 | 33 |
| re .1
At 9600 baud 1 kbyte/sec is about tops. The MNP5 compression has
little to no effect since the files are already compressed.
9600 baud / 10 bits per char = .96 kbytes per sec.
With the DTE rate set to 19.2K and the MNP5 doing compression on
regular files, much higher rates can be achived.
re .2
If I understand the frame buffer correctly it has to do telling the
sender to wait for an ACK every frame buffer size. The problem occurs
when using floppy's because of their slow access time. This can cause
the sender to timeout with large buffer sizes.
In short what I get from reading the docs is if you use a floppy or
have a problem when you set your buffer size large, you should set your
buffersize and framesize the same.
eg: If you set your buffer size to 64k you should set your framesize
to 65536 bytes. This will cause the sender to wait for a ACK every
64Kb, thus preventing the timeouts. A setting of zero assumes you will
be able to complete the buffer size write to disk before the sender
times out.
One other nice thing I just discovered about MNP5. My wife just picked
up the phone to make a call, heard the computer chatter and came
running in here saying sorry. Except for a short delay in echoing my
typing, I would not have noticed ;').
|