On Thu, 2003-08-28 at 16:21, Jim Rootham wrote:

I would suspect (with almost no evidence) that you may be missing latency windows (especially on write). It would not take too many times for the disk to have to spin (almost) all the way round to hammer the total transfer rate pretty good. If you could get more buffers in the pipe that might help.

Jim

I have the same suspicion. Is there any way to control pipes in bash? I was thinking of piping the data through dd again with a different block size to try and make a quasi-buffer, but I don't think that'll be very efficient. Kareem -- /********************************************************************* kareem-d+8TeBu5bOew5LPnMra/2Q at public.gmane.org - Kareem Shehata - 416-676-6611 -------------------------------------------------------------------- The highest use of capital is not to make more money, but to make money do more for the betterment of life. -- Henry Ford (1863 - 1947) ********************************************************************/ -- The Toronto Linux Users Group. Meetings: http://tlug.ss.org TLUG requests: Linux topics, No HTML, wrap text below 80 columns How to UNSUBSCRIBE: http://tlug.ss.org/subscribe.shtml