Date: Thu, 8 Jan 2009 20:25:19 +0100 From: Hans Petter Selasky <hselasky@freebsd.org> To: Ed Schouten <ed@80386.nl> Cc: Remko Lodder <remko@elvandar.org>, perforce@freebsd.org, "M. Warner Losh" <imp@bsdimp.com> Subject: Re: PERFORCE change 155748 for review Message-ID: <200901082025.20711.hselasky@freebsd.org> In-Reply-To: <20090108191335.GT45775@hoeg.nl> References: <200901071009.n07A9jrs056953@repoman.freebsd.org> <200901082010.33577.hselasky@freebsd.org> <20090108191335.GT45775@hoeg.nl>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thursday 08 January 2009, Ed Schouten wrote: > * Hans Petter Selasky <hselasky@freebsd.org> wrote: > > Everything is possible. > > > > How would you explain the benefit of the +19Mbyte of data resulting from > > this autogeneration when building the modules ? > > I think we always had a preference to generate stuff on demand. It's a > lot easier to maintain. Hi, I can understand that, but how about only generating the file once when you know it will be needed 60 times in a row? This also applies to other autogenerated files. It would speed up the module build alot I think. cd src/sys/modules/usb2/serial_3g make all clean time make all 0:01.58 rm u3g2.o time make all 0:00.96 You see it is around 50% slower to build a module which needs to generate N header files first, than if those header files were cached somewhere! Multiply the time you save by 60, and we are talking about a significant amount of time! --HPS
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200901082025.20711.hselasky>