Re: Windows CE 5.0 eval edition problem

From: fd (fd_at_daptech.com)
Date: 08/13/04 

Date: Fri, 13 Aug 2004 09:41:15 -0400

Looked more deeply and no it's not the cache. I build a simple program to
use the FindFisrtfile-FindNextFile functions. That's where the problem is.
Calling those functions on CE5.0 are very much slower than on CE4.2.

Any idea where I have to look now?

Regards.

"fd" <fd@daptech.com> wrote in message
news:evsfmshfEHA.2460@TK2MSFTNGP12.phx.gbl...
>
>
> I have successfully adapted CE 4.2 to CE 5.0 but I have some problem with
> the disk cache. I have exactly the same "Sysgened" (except for obsolet and
> new sysgen switch) configuration and the same hardware.
>
>
>
> On my 4.2 system when the Desktop display icons (15 icon), it's very fast.
> Same for unfolding Start menu program.
>
>
>
> On CE 5.0 when the Desktop display icons (15 icon), it's very slow and
also
> for displaying the Start menu program.
>
>
>
> After digging, it appears that the read cache is not active on CE 5.0. I
saw
> that after enabling some trace in Nand's DiskIO read sector, sectors are
> always requested from hardware. This is not the same on CE 4.2.
>
>
>
> My system is a PXa255, 64 mb RAM, 64 mb Nand
>
> ROM only file system is enabled on the Nand.
>
>
>
> [HKEY_LOCAL_MACHINE\System\StorageManager\Profiles\NandDisk]
>
> "Name"="NandDisk2 Flash Device"
>
> "Folder"="Storage_Nand2"
>
> "AutoPart"=dword:0
>
> "AutoFormat"=dword:1
>
> "MountFlags"=dword:7
>
> "PartitionDriver"=""
>
> "FileSystem"="fatfsd.dll"
>
> "CacheDll"="diskcache.dll"
>
>
>
> [HKEY_LOCAL_MACHINE\System\StorageManager\Profiles\NandDisk\FATFS]
>
> "CacheSize"=dword:0
>
> "MountAsRoot"=dword:1
>
> "Flags"=dword:00000026
>
> "EnableCacheWarm"=dword:0
>
> "EnableCache"=dword:1
>
>
>
> Any idea where I have to look now.
>
>
>
> Regards
>
>

Relevant Pages

  • Re: Possible ways of dealing with OOM conditions.
    ... There is more to networking that skbs only, what about route cache, ... With power-of-two allocation SLAB wastes 500 bytes for each 1500 MTU ... Well, if you have such hardware its not rare at all, But yeah that ... adds to the fragmentation issues on the page-allocator level. ...
    (Linux-Kernel)
  • Re: why I told dd command to write but it reads in first?
    ... the buffer cache is different from file system cache? ... think the sector might not be exactly the same ... logical sectors in a single physical sector, ...
    (comp.os.linux.development.system)
  • Re: [00/17] Large Blocksize Support V3
    ... allocations, and then creating all this complexity around that is not a ... Now why are high end hardware manufacturers building crippled ... For I/O we have the BIO which can point to multiple pages just fine. ... the page cache from the BIO and from the BIO to the page cache. ...
    (Linux-Kernel)
  • Re: 2.6, 3.0, and truly independent intepreters
    ... reduces the overhead of lock acquisition, while still allowing use of global ... And the overhead of doing that at the level of a cache ... You still need to minimize contention of course, ... then it'll be faster to have it in hardware. ...
    (comp.lang.python)
  • Re: Larrabee delayed: anyone know whats happening?
    ... least two - the one provided by the hardware and the one assumed by ... Actually, for historical reasons the old Intel official memory ordering model was PC, and the semantics of fences and synchronizing atomic RMWs were not well defined. ... Programming languages specified a really weak memory ordering model, say MO-0. ... Some people think that we will have domains that are cache coherent, in combination with domains that share memory, but are not cache coherent. ...
    (comp.arch)