Date: Wed, 18 May 2011 13:57:39 +0300 From: "Reko Turja" <reko.turja@liukuma.net> To: "FreeBSD Questions" <freebsd-questions@freebsd.org> Subject: Re: Buildworld Benchmarks Message-ID: <9D9581D0C8E84BF1B3BBFF2B8F696AD5@rivendell> In-Reply-To: <022d01cc14f5$a37ef010$ea7cd030$@vicor.com> References: <022d01cc14f5$a37ef010$ea7cd030$@vicor.com>
next in thread | previous in thread | raw e-mail | index | archive | help
> We thought going tmpfs would make things faster, but that resulted=20 > in over 13 > minutes (huh? you'd think a RAM disk would be smoking compared to=20 > even the SSDs > that we used to achieve ~9 min; do note that we did make sure to=20 > nullfs mount a > tmpfs-based directory onto /usr/obj -- though the performance of=20 > that nullfs > mount might have hurt the test, not sure). I think that doing huge parallel build and using part of memory bus=20 bandwidth for storage is the culprit. DMA offloading the storage=20 operations directly from memory to the disks is giving the advantage=20 to builds using disk IO. -Reko=20
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?9D9581D0C8E84BF1B3BBFF2B8F696AD5>