[CALUG] Fedora upgrade .... CALUG Digest, Vol 54, Issue 14

James Ewing Cottrell 3rd JECottrell3 at Comcast.NET
Tue Jun 21 19:41:38 EDT 2011


  I like the exact boundary argument. I was also considering 128 and 
128, giving exactly 8G cylinders.

But I am wondering...where is the geometry stored? The old DOS used to 
have a BPB..."BIOS Parameter Block" that stored the geometry. But I made 
some new partition tables with odd sizes and didn't see the sizes I gave 
anywhere.

The other question I have is: What is the maximum size I can read at 
once with out the driver splitting it up into two different reads?

JIM

P.S. Who uses 4K blocks? I know that some companies (Pyramid was one) 
often used 2K Physical Sectors. I don't doubt you...I am just wondering 
who/where they are.

On 6/21/2011 1:02 AM, Bryan J Smith wrote:
> For storage performance, especially in arrays, perfect boundaries on page,
> block, stripe and cache are ideal.  If you use CHS addressing, then you're on
> imperfect boundaries.
>
> Again, CHS can be safely _ignored_ with NT and Linux.  I used to default to
> dealing with it until I did some performance testing.  That's when I switched to
> exact sector alignments on perfect boundaries (1GiB today, I started with 256MiB
> back then) years ago.
>
> One can set down to 32/64 if one so desires, but it really doesn't matter either
> way.  Putting in the exact number of 512b or, more recently, 4KiB sectors is
> always absolute.  But 32/64 will show up far more pretty in legacy fdisk
> versions.
>
>
>
> ----- Original Message ----
> From: James Ewing Cottrell 3rd<JECottrell3 at Comcast.NET>
> Sent: Mon, June 20, 2011 4:05:00 PM
>
>    I am a bit more anal about partitions...I want the Cylinders to be
> multiples of 1000 so I can remember them. But I am also a fan of Cookie
> Cutter Sized Partitions or LVs.
>
> Alternatively, if you create the Partition Label with 32 sectors and 64
> heads, you get 1M cylinders, which is kinda small, but works nicely for
> flash drives.
>




More information about the CALUG mailing list