Messages in this thread | | | From | Stefan Monnier <monnier+lists/linux/kernel/news/@tequila.cs.yale.edu> | Subject | Re: elevator algorithm bug in ll_rw_blk.c | Date | 17 Nov 1998 11:55:02 -0500 |
| |
>>>>> "Riley" == Riley Williams <rhw@bigfoot.com> writes: >>> suchlike seeks take 1+1=2 rather than letting some of them take >>> 1/2+1/2=1 - and I have to say that I disagree with you...
You got your math wrong. The heads are not moved by step-by-step motors anymore. They accelarate (more or less linearly) for half the time, than decelerate for the other half. The result is that time taken to move by N tracks is proportinal to sqrt(N) rather than to N. Then there is the settling time (once you're `more or less on the track') and this is independent of the actual distance. What this all means is that going across the whole disk is very cheap. A `10ms' drive will typically take 15ms for this `return scan', which should be compared to the average rotational latency (around 5ms) over which operating system have very little control.
> I had a look in the three OS textbooks I have access to. One followed > your comments pretty much to the letter, one explained that a one-way > scan is normally used because it's easier to program than a two-way > scan, and the third states that a two-way scan is normally used "in > professional systems" because of its higher efficiency.
What about the age of those textbooks ?
Stefan
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |