Re: Performance of packet writing: Blocksize matters

Johannes Niess wrote:
> I went for a minimal difference. The second experiment is with block size 
> increased by 1 byte (32k + 1) = 32769. Sorry for not making that obvious. The 
> change was intended to force reblocking. I agree that it is not a real world 
> test.

Ahh, my eyes glossed over the last digit and just read it as 32k.

> Will this work?
> 
>  echo noop > /sys/block/hdc/queue/scheduler

Yes.

Let me know how the noop scheduler performs.