From mboxrd@z Thu Jan 1 00:00:00 1970 From: Stan Hoeppner Subject: Re: standard performance (write speed 65Mb/s) Date: Thu, 21 Jul 2011 19:24:12 -0500 Message-ID: <4E28C32C.1020701@hardwarefreak.com> References: <201107181403.49563.raid1@fuckaround.org> <201107211957.16493.raid1@fuckaround.org> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <201107211957.16493.raid1@fuckaround.org> Sender: linux-raid-owner@vger.kernel.org To: Pol Hallen Cc: Erwan Leroux , linux-raid@vger.kernel.org List-Id: linux-raid.ids On 7/21/2011 12:57 PM, Pol Hallen wrote: > dd if=/dev/zero of=/dev/md0 bs=1024M count=10 > 10+0 records in > 10+0 records out > 10737418240 bytes (11 GB) copied, 166.042 s, 64.7 MB/s Do you not recall that I explained in detail, not two days ago, why you should not use a 1GB block size when testing with dd in this manner? And don't do this with the raw device. The resulting numbers, no matter how large or small, are irrelevant. *Put a filesystem on md0* and do: ~$ dd if=/dev/zero of=/some/dir/test.dd bs=4096 count=2621440 And why on earth would you bother running this test while the array is still building? Again, the resulting figure is worthless to everyone. -- Stan