cpu

cpu

[Raz Ben-Jehuda(caro)]

I was testing linux raid for sometime, I created several test units
which send read request to the raid using big buffers( 4MB).
It seems to be that the software raid consumes to much cpu. I reach
100% cpu usage just by reading from the raid.
I used 4 disks raid 0.

-- 
Raz
Long Live the Penguin

Re: cpu

[Michael Tokarev]

Raz Ben-Jehuda(caro) wrote:
> I was testing linux raid for sometime, I created several test units
> which send read request to the raid using big buffers( 4MB).
> It seems to be that the software raid consumes to much cpu. I reach
> 100% cpu usage just by reading from the raid.
> I used 4 disks raid 0.

Now try to do the same test without raid - I bet it will use
as much CPU as your raid test.  Raid0 requires almost no CPU
resources, unlike improperly set up IDE disks (no DMA is the
first thing to check -- hdparm is your friend).

/mjt

cpu

[Vinod Hegde]

Hi,

I am new to this topic. So the question could be fundamental.

What action does cpufreq take for the memory sub system. I mean, once
cpuload is known, does it assume anything about the memory usage and alter
the voltage of any memory module. Is any change brought for any other
hardware components based on cpu load except cpu.

Thanks ....