Return to BSD News archive
Path: euryale.cc.adfa.oz.au!newshost.carno.net.au!harbinger.cc.monash.edu.au!news.mel.connect.com.au!munnari.OZ.AU!news.Hawaii.Edu!news.uoregon.edu!hammer.uoregon.edu!arclight.uoregon.edu!super.zippo.com!zdc!szdc!szdc-e!news From: "John S. Dyson" <dyson@freebsd.org> Newsgroups: comp.unix.bsd.freebsd.misc Subject: Re: Anyone compiled a kernel using pgcc? Date: Fri, 01 Nov 1996 10:28:14 -0500 Organization: John S. Dyson's home machine Lines: 37 Message-ID: <327A170E.41C67EA6@freebsd.org> References: <3279FCFC.237C228A@systemics.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit X-Mailer: Mozilla 3.01b1 (X11; I; FreeBSD 2.2-CURRENT i386) Gary Howland wrote: > > Hi, > > Has anyone out there had any success compiling a kernel (in my case, the > October SNAP) using pgcc, the pentium optimised compiler? > You can "mostly" compile the kernel with pgcc -- there are some problems, but the early OCT version of pgcc is just broken (for FreeBSD.) I haven't sent the PGCC a bug report yet -- and if this suprises the PGCC people, I apologise for the public (slightly negative) comment. I know that it will get much better (I like pgcc.) > > Is this recommended, or is this likely to make my kernel less reliable? > You will gain only on micro-level benchmarks (which really don't mean much in the real world.) The reason for my micro-level benchmark comment is because the ALGORITHMS are where it is. Almost all well written/ported systems will perform approx the same under light load. PGCC can boost the performance of the lmbench lat_proc by approx 5%... But fork/execs don't usually take 100% of your CPU!!! Also, this benchmark doesn't mimic reality when the system becomes loaded doing other things (cache perf, etc.) If you want to relatively safely compile the kernel for the best perf on a pentium try, using the 2.7.2.1 compiler: -O2 -fomit-frame-pointer -malign-jumps=0 -malign-loops=0 John