Subject: Re: AXPpci 33 ... not very fast?
To: Ken Hornstein <firstname.lastname@example.org>
From: Curt Sampson <email@example.com>
Date: 04/25/1997 13:19:24
On Fri, 25 Apr 1997, Ken Hornstein wrote:
> So, I've been playing around with this AXPpci 33 machine lately ... and
> it seems to me that it's just not very fast at all. It compiles a kernel
> about as fast as my 486/66DX2 (a mostly useless benchmark, I know, but
> still ... ). It's just "feels" sluggish, and this machine has 48 MB memory.
You should compare something other than kernel compiles, so you're
comparing the same things. Try, for example, comparing the compile
of perl or a largish piece of gnu software, or something like that.
More backup cache may help a fair amount; the memory latency on
the AXPpci33 isn't as bad as the multia, but is still two to three
times worse than that on a typical Pentium system. But then again,
it's hard to say how much you're seeing in the way of cache misses.
If you want to contact me privately we can work out some arrangements
to do some tests; I have a megabyte of 15 ns. cache RAM in my
The beast does have at least one advantage over a 486: it will do
disk I/O at reasonable speeds on fast drives. I get around 4 MB/sec
on my 5400 RPM drives on my AXPpci33; I never could get more than
about 2 MB/sec on the same drives on a 486/100.
If anybody is interested in working on a `benchmark project' to do
some benchmarks and get rought ideas of the relative speeds of
various NetBSD machines, I would be very happy to provide the web
space to put that information up. If we can put together some sort
of reasonably automated test suite, I could also run it on a bunch
of different machines.
Curt Sampson firstname.lastname@example.org Info at http://www.portal.ca/
Internet Portal Services, Inc. Through infinite myst, software reverberates
Vancouver, BC (604) 257-9400 In code possess'd of invisible folly.