Date: Fri, 7 Jun 2002 13:04:02 -0500 From: Dan Nelson <dnelson@allantgroup.com> To: "Jack L. Stone" <jackstone@sage-one.net> Cc: JR Richards <jr@ebcrp.org>, freebsd-questions@FreeBSD.ORG Subject: Re: Ghosting-like application Message-ID: <20020607180402.GA13099@dan.emsphone.com> In-Reply-To: <3.0.5.32.20020607123852.00fc2288@mail.sage-one.net> References: <001801c20e48$6736bd30$3002a8c0@JR> <001801c20e48$6736bd30$3002a8c0@JR> <3.0.5.32.20020607123852.00fc2288@mail.sage-one.net>
next in thread | previous in thread | raw e-mail | index | archive | help
In the last episode (Jun 07), Jack L. Stone said: > Dan: This reminds me... I've seen the "bs=1m" all over the map from other > postings on this very subject: 1024, 4096, 8192 (even 8096)...etc. This is > the first at "1m" I've seen & still wonder how this is determined or > recommended. I'm using 8192 and it works okay. After doing a dd, I even did > a boot test on the HD #2 and it worked slick.... still wonder about the > "bs" though.... 1m is probably overkill, actually. bs= sets the read and write blocksize. When you're doing sequential I/O on large devices, raising the blocksize lowers the amount of per-read overhead the OS has to do. On my system (p3 933), a dd with an 8k blocksize takes around 7% CPU; a dd with 64k (FreeBSD splits larger bs values into multiple 64k chunks) takes 1%. On an idle system it doesn't really matter. -- Dan Nelson dnelson@allantgroup.com To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20020607180402.GA13099>