Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!cs.utexas.edu!uunet!van-bc!
From: lphillips@lpami.wimsey.bc.ca (Larry Phillips)
Newsgroups: comp.sys.amiga
Subject: Re: DNet performance (was:
Message-ID: <619@lpami.wimsey.bc.ca>
Date: 7 Jul 89 15:53:12 GMT
Lines: 57
To: van-bc!rnews

In <295@sed170.HAC.COM>, lee@sed170.HAC.COM (John Lee ) writes:
>>As a laboratory demonstration, I have a 2400 baud modem, and I run DNET.
>>(Great stuff, Matt!) For those of you who do not know, DNET uses the line fully
>>during downloads - no pauses waiting for acknowledgements. WHen downloading I
>>get about 10K per minute transfer rate. (And, as the old joke goes, I'm
>>THANKFUL for it. Just wait'll I have kids and they complain about how slow
>>their T1 line is.)
>>
>>10240 bytes per 60 seconds = 171 bytes per second
>>                           = 1710 bits per second
>>
>>The rest is DNET's blocking and protocol and escaping. I'm not getting anything
>>like 2400 BYTES per second.
>
>Gee, are you sure there's nothing else going on?  That's only 71.25%.  I don't
>think DNET uses 28.75% of the bandwidth for its protocol.  2400 bps modems
>are really 240 characters per second.  I've timed it out of curiosity when
>I was doing straight ASCII dumps.

Hmm... just tried a DNet 'putfiles', Amiga to Sun 2/50, at 9600 baud (a direct
connection). It transferred 114482 bytes in 128 seconds, gving an effective
data rate of 894 bytes/second, or about 93% of theoretical maximum. (I agree...
nice work Matt!). This was from VD0: to my home directory on the Sun. A DNet
'getfiles' of the same file yielded about the same (126.75 seconds).

Out of curiousity, I tried a transfer both ways at the same time (with renamed
files of course), and found the putfiles took 158.6 seconds, at which time
the getfiles had transferred 61400 bytes (it was running at about half the
speed of the putfiles all along).

Taking the time for putfiles, and figuring the total line usage in both
directions, it gives a total of (114482 + 61440) 175922 bytes transferred in
158.6 seconds, or about 1109 bytes/second.

I am currently trying to write a version of 'tip' on the Sun that will allow me
fully transparent access to the serial port. The reason is that I want to have
the modem connected to the Sun, but accessable from the Amiga. Using the
standard tip, I can do anything ASCII oriented, and use Aterm 7.3 connected to
the Sun at 9600 bps (or baud.. :-) ), and it runs exactly the same as if I had
dialled directly out the modem.

I have my Sun source on both the Sun and Amiga, and do all my editing on the
Amiga (with CEDPro). Whenever I change a module on the Amiga, I can simply type
'make', and the module(s) I changed get sent via putfiles to the source
directory on the Sun. Typing 'make' in a DNet 'fterm' window then compiles the
source on the Sun. (now all I need is rpc capability in DNet and I won't even
have to have the fterm window up).

-larry

--
Van Roy's Law:  An unbreakable toy is useful for breaking other toys.
+----------------------------------------------------------------------+ 
|   //   Larry Phillips                                                |
| \X/    lphillips@lpami.wimsey.bc.ca or uunet!van-bc!lpami!lphillips  |
|        COMPUSERVE: 76703,4322                                        |
+----------------------------------------------------------------------+