[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Is squeeze compatible woth WD20EARS and other 2TB drives?



On Mon, Dec 13, 2010 at 05:40:42PM -0600, Stan Hoeppner wrote:
> Jochen Schulz put forth on 12/13/2010 3:51 PM:
> > Mike Viau:
> >>> On Mon, 13 Dec 2010 21:46:59 +0100 <ml@well-adjusted.de> wrote:
> >>>
> >>> Partition 1 does not start on physical sector boundary.
> >>
> >> I don't understand the implied meaning of this error? Did you take any
> >> precautions as to the alignment of your partition? What about if you
> >> were planning on having multiple partitions?
> 
> >From the linux-ide mailing list April 2010:
> 
> http://www.spinics.net/lists/linux-ide/msg37306.html
> 
>  Unfortunately, sfdisk and cfdisk have no clue about alignment and
>  disks topology.
> 
>  Now only the fdisk and parted commands are ready for new disks.
> 
> 
> So the proper title of this thread should be:
> 
> "Does the Squeeze installer use fdisk or parted?"
> 
> As of April 2010, only fdisk and parted will create sector boundary
> aligned partitions on drives that internally translate from 4K to 512B
> sectors.
> 
> Linux typically reads 4K blocks.  With a translated drive, two or more
> 512 byte sectors of a given block may lie on opposite sides of a
> hardware 4K sector boundary.  Thus, Linux will have to read two
> consecutive hardware sectors instead of one in order to get the contents
> of the 4K logical block.  I.e. two reads (head seeks) per every 4k block
> instead of one.  Thus, you get half the performance vs a properly
> aligned cylinder.
> 
> In the case of the WD20EARS this will drop your peak sequential read
> rate from over 100 MB/s down to 50 MB/s or less.  Without proper
> alignment, you're literally leaving half of your drive's sequential read
> performance on the table.
> 
> The story is even worse for 4K random writes.  You'll drop your write
> performance by a factor of 4 or more if you don't have proper alignment:
> 
> The moral of the story is, use fdisk or parted and make sure your
> partitions are correctly aligned on 4KB boundaries.  If the Squeeze
> installer doesn't do so, I'd recommend booting an ISO or USB live distro
> and partition the drive with fdisk or parted before booting the Squeeze
> installer.

Hi Stan,

That is fascinating to learn, especially the fdisk
is among the best tools for disk partitioning.

I was going to ask about my disk, but I see that my message
is 

    Partition 1 does not end on cylinder boundary.

Which seems to be different than the OP's physical sector size
message.

If partition 1 is off, does that mean *all* partitions are
off? How does one verify? (I used MiniTool Partition Wizard
to shrink my laptop's Windows 7 partition, handling all the
magical, Linux-hating files -- MFT, etc.)

http://www.partitionwizard.com/free-partition-manager.html

FWIW, my disk partitioning follows.

Thanks for posting this information!

Joel





Disk /dev/sda: 320.1 GB, 320072933376 bytes
255 heads, 63 sectors/track, 38913 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x27b11b56

   Device Boot      Start         End      Blocks   Id  System
/dev/sda1   *           1         154     1228800    7  HPFS/NTFS
Partition 1 does not end on cylinder boundary.
/dev/sda2             154        4742    36860288+   7  HPFS/NTFS
/dev/sda3            4743        5251     4088500   82  Linux swap / Solaris
/dev/sda4            5252       38914   270398047+   f  W95 Ext'd (LBA)
/dev/sda5            5252        7163    15358108   83  Linux
/dev/sda6            7164       37638   244790404   83  Linux
/dev/sda7           37639       38914    10240000    7  HPFS/NTFS

 
> -- 
> Stan
> 
> 
> -- 
> To UNSUBSCRIBE, email to debian-user-REQUEST@lists.debian.org 
> with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
> Archive: [🔎] 4D06AEFA.6070903@hardwarefreak.com">http://lists.debian.org/[🔎] 4D06AEFA.6070903@hardwarefreak.com
> 

-- 
Joel Roth


Reply to: