Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 28 Sep 2001 08:18:01 -0600
From:      Robert Gray <bob@cs.colorado.edu>
To:        "Daniel O'Connor" <doconnor@gsoft.com.au>
Cc:        Edwin Culp <eculp@EnContacto.Net>, freebsd-mobile@FreeBSD.ORG, kjerste soderberg <kjerstes@yahoo.com>
Subject:   Re: cloning laptop drives 
Message-ID:  <200109281418.IAA30085@calypso.boulderlabs.com>
In-Reply-To: Your message of "Fri, 28 Sep 2001 15:46:43 %2B0930."

next in thread | raw e-mail | index | archive | help
Yes, like Daniel says, large block sizes (I use 32K or 64K)
and separate IDE controllers make a huge difference.

You may as well use the "raw" devices to transfer directly
into user space.

One further tip, /usr/ports/misc/buffer, can double the
speed because it forks into two processes connected
via shared memory - a reader and a writer, each operating
as fast as the disks can respectively read or write.
In contrast, dd waits for a read to complete before starting a write.
Buffer also prints progress so you know how long it will take.

Here is approximately what I use:
  buffer -s 32k -S 1m </dev/rad0 >/dev/rad1
  -s is the block size
  -S prints out progress

"Daniel O'Connor" <doconnor@gsoft.com.au> Fri, 28 Sep 2001 15:46:43 +0930 says:
>
>On 28-Sep-2001 Edwin Culp wrote:
>>  AFAIK, the default for dd is one block at a time and that can take for ever
>.
>>  You need to define the block size to something much larger.  I have seen
>>  some
>>  cool formulas based on disk geometry but I never seem to remember them when
>>  I need them.:-(  You might try something like
>>  
>>  dd if=/dev/ad0 of=/dev/ad1 bs=8192
>>  
>>  Hopefully someone will give you a better number.
>
>I usually pick 64k when doing this..
>
>Also, if those drives are on the same chain the performance is going to suck
>really hard..
>
>If you hit 'ctrl-t' you will get some info on dd's progress, and if you run
>'systat -vmstat 1' and look at the bottom section you'll see the transfer
>speeds.

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-mobile" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200109281418.IAA30085>