| View previous topic :: View next topic |
| Author |
Message |
dE_logics
Advocate
Joined: 02 Jan 2009
Posts: 2290
Location: $TERM
|
 Posted: Thu Nov 17, 2011 4:34 pm Post subject: Device specific caching. |
 |
|
Cause of ram constrains, I've reduced dirty caching, but will this affect internal performance? For .e.g. caching of data generated from the CPU or copied over from the memory?
If so, how can I just limit the dirty cache ratio to external FS like NFS/CIFA or USB devices?
_________________
My blog |
|
| Back to top |
|
 |
CrankyPenguin
Apprentice
Joined: 19 Jun 2003
Posts: 283
|
| Posted: Fri Nov 18, 2011 6:31 pm Post subject: |
|
|
You might find some help with /etc/conf.d/hdparm It allows you to set some device-specific parameters. I'm not sure if it will be sufficient for your needs but it is a place to start.
_________________
Linux, the OS for the obsessive-compulsive speed freak in all of us. |
|
| Back to top |
|
|
dE_logics
Advocate
Joined: 02 Jan 2009
Posts: 2290
Location: $TERM
|
| Posted: Mon Nov 21, 2011 7:05 am Post subject: |
|
|
Yes, I already use hdparm; it doesn't have such a feature. It just has read ahead which's not it.
_________________
My blog |
|
| Back to top |
|
|
CrankyPenguin
Apprentice
Joined: 19 Jun 2003
Posts: 283
|
| Posted: Wed Nov 23, 2011 6:58 pm Post subject: |
|
|
No true but the hdparm conf does device-specific tasks so you may be able to build on that to put something together.
_________________
Linux, the OS for the obsessive-compulsive speed freak in all of us. |
|
| Back to top |
|
|
krinn
Watchman
Joined: 02 May 2003
Posts: 7470
|
| Posted: Wed Nov 23, 2011 8:03 pm Post subject: Re: Device specific caching. |
|
|
There's an old bug when reducing the dirty ratio, i'm not sure it has been fix : https://lkml.org/lkml/2006/12/28/171
Corruption... you better make sure this one has been fix prior to go that way.
And there's no point to cache cpu as cpu have 3rd level cache today, and far far faster than your normal ram, and i think the kernel have enough task already to do than handling all cpu instructions to cache them.
And there's no point to cache a device with another one of the same speed, so caching your main memory with your main memory would just be a waste of time and speed. |
|
| Back to top |
|
|
dE_logics
Advocate
Joined: 02 Jan 2009
Posts: 2290
Location: $TERM
|
| Posted: Thu Nov 24, 2011 3:58 pm Post subject: Re: Device specific caching. |
|
|
| krinn wrote: | There's an old bug when reducing the dirty ratio, i'm not sure it has been fix : https://lkml.org/lkml/2006/12/28/171
Corruption... you better make sure this one has been fix prior to go that way.
And there's no point to cache cpu as cpu have 3rd level cache today, and far far faster than your normal ram, and i think the kernel have enough task already to do than handling all cpu instructions to cache them.
And there's no point to cache a device with another one of the same speed, so caching your main memory with your main memory would just be a waste of time and speed. |
This system ain't of today. Also I was wondering about -
Program -> RAM dirty caching -> HDD
So it would cache data which should be written to disk, not instructions. Also there's no corruption with dirty caching in my case.
_________________
My blog |
|
| Back to top |
|
|
|
Kernel & Hardware
