*BSD News Article 10450


Return to BSD News archive

Received: by minnie.vk1xwt.ampr.org with NNTP
	id AA150 ; Thu, 28 Jan 93 22:02:24 EST
Newsgroups: comp.unix.bsd
Path: sserve!manuel.anu.edu.au!munnari.oz.au!spool.mu.edu!uwm.edu!caen!umeecs!quip.eecs.umich.edu!dmuntz
From: dmuntz@quip.eecs.umich.edu (Dan Muntz)
Subject: [386BSD] processor caching effect on kernel build
Message-ID: <1993Jan27.205537.18117@zip.eecs.umich.edu>
Sender: news@zip.eecs.umich.edu (Mr. News)
Organization: University of Michigan EECS Dept., Ann Arbor, MI
Date: Wed, 27 Jan 1993 20:55:37 GMT
Lines: 37


People have been wondering about how the internal and external caches
affect performance under 386bsd.  Here are some numbers I got doing kernel
builds.

System:
486DX50 (8k internal cache) EISA
16mb 80ns (4 4Mx9 simms)
Adaptec 1542B (ISA) scsi
Micropolis 1528 scsi disk
256k 20ns external cache
64mb swap (whatever portion of 64mb 386bsd is capable of using)
gcc version 1.39

both caches disabled:
	make depend:	05:38.13
	make:		28:51.15

external cache enabled, internal disabled:
	make depend:	05:37.53
	make:		28:49.88

internal cache enabled, external disabled:
	make depend:	01:35.66
	make:		07:49.75

both caches enabled:
	make depend:	01:25.28
	make:		06:10.84

Kernel being built is 0.1 + patchkit0.2 + julian's scsi + pcfs + minor patches
Running kernel is the same (and contains NONOP patch recently posted).
You've just got to love 7:36 for building a kernel from scratch :)

  -Dan
   dmuntz@eecs.umich.edu