Leyne, Sean wrote:

> Aage,
>
>
>> Since very few images are less than 16KB (they are mostly 50KB and
>> more) most blobs would span several pages and I reasoned that the
>> fewer pages the better. If an image is 90KB, at least 6 pages (@
>> 16KB) must be read, while for 8KB pages at least 12 pages will be
>> read. Will the total time for reading the pages only depend on the
>> blob size, and be independent of the number of pages?
>>
>
> I would recommend page size of 32KB, which would be matched by a disk
> cluster size of 32KB -- too truly maximize disk to memory transfer
> speeds.
>
>
> Sean
>

Sean,

Are you talking about which kind of filesystem ?

I was looking into my Linux ext3 Filesystem, using dumpe2fs and I got:
Block size: 4096
Fragment size: 4096


the man page of mk2fs gvies this:
-b block-size
Specify the size of blocks in bytes. Valid block size
vales are 1024, 2048 and 4096 bytes per block. If omit-
ted, mke2fs block-size is heuristically determined by the
file system size and the expected usage of the filesystem
(see the -T option).


here states that the block-size should match the memory page size,
http://lwn.net/Articles/188382/

on other links I read that is not the case for years, don't know which
one is correct...


see you !

--
Alexandre Benson Smith
Development
THOR Software e Comercial Ltda
Santo Andre - Sao Paulo - Brazil
www.thorsoftware.com.br