Date: Fri, 26 Jul 1996 18:38:52 BST From: Michael Searle <searle@longacre.demon.co.uk> To: hackers@freefall.freebsd.org Subject: Re: mfs /tmp, ffs /tmp or -pipe Message-ID: <mF3DE7554@longacre.demon.co.uk> References: <199607261535.IAA02391@freefall.freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
owner-hackers-digest@freefall.freebsd.org wrote: > Hi, > I've just been playing around with different ways to compile the kernel. > My, completely simple, measurement of kernel compile time on a 2.1.5 > P-100 are: > ffs /tmp : 7:50 seconds mfs /tmp : 7:30 seconds (with TMPDIR=/tmp) cc > -pipe : 7:22 seconds > Fairly inconclusive and probably not repeatable. Anyway, what I am > asking is are there any documented advantages of using an mfs /tmp?? Is > -pipe a better option for compiles?? I haven't tried using mfs, and these are not kernel compile times, but any large compile should be similar for comparing makes. I got 198-201 seconds real, of which 86-87% was used, whatever options were used, when using the system make. Using gmake (still with ffs /tmp), this dropped to 189.57 seconds real, 87.2% used. Using -pipe with gmake, this actually increased to 192.43 seconds real, 88.6% used. The real improvement was when I used -pipe with -j. Using -j2, -j3 and -j4 I got similar results, but the best for me was -j3, which completed in 176.28 seconds real, of which 97.6% was used. -- Michael Searle - searle@longacre.demon.co.uk
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?mF3DE7554>