Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site down.FUN
Path: utzoo!watmath!clyde!burl!ulysses!allegra!princeton!down!honey
From: honey@down.FUN (code 101)
Newsgroups: net.lan,net.dcom,net.unix-wizards
Subject: Re: ethernet performance
Message-ID: <471@down.FUN>
Date: Sat, 16-Mar-85 23:47:18 EST
Article-I.D.: down.471
Posted: Sat Mar 16 23:47:18 1985
Date-Received: Sun, 17-Mar-85 06:00:47 EST
References: <213@athena.UUCP>
Organization: Princeton University, EECS
Lines: 11
Xref: watmath net.lan:724 net.dcom:904 net.unix-wizards:12462

Synopsizing kroot's data, and adding a third column:

buffer	  kbytes/sec	writes/sec
1k - 2k	  50 - 100	25 - 100
10	  10 - 20	1k - 2k
1	  .06 - .09	60 - 90

The big surprise is not the shabby throughput with one byte writes, but
the incredible performance with ten byte writes.  This seems suspect.

	Peter