Re: Possible improvement to pipe throughput

David S. Miller (davem@caip.rutgers.edu)
Sun, 29 Sep 1996 03:26:28 -0400


From: Albert Cahalan <albert@ccs.neu.edu>
Date: Fri, 27 Sep 1996 17:06:58 -0400 (EDT)

High benchmark numbers have real world significance.
They make Linux look good, so more people install it.

But there is a turn side to this. If people end up installing it, and
the performance for what they do on the machine doesn't match up to
the numbers we claim because they stress something the benchmarks do
not, then this can end up being just as bad.

David S. Miller
davem@caip.rutgers.edu