Messages in this thread | | | From | (Linus Torvalds) | Subject | Re: [OFFTOPIC] Re: disk head scheduling | Date | 20 Mar 1999 21:12:30 GMT |
| |
In article <19990319234917.A15988@anjala.mit.edu>, Arvind Sankar <arvinds@mit.edu> wrote: >On Fri, Mar 19, 1999 at 10:39:50PM -0500, Richard B. Johnson wrote: >> > Who's talking about optimizing based on phony geometries? Did you read what >> > I wrote? >> >> You just wrote it. Even though I worked for a major disk manufacturer >> I don't know what you will say in the future. > >This is getting silly. I can't understand where you get the idea that I was >talking about any phony geometries. I explicitly said `some manufacturers are >good enough to tell us the real geometry in their tech notes'.
Arvind, give it up.
NO manufacturer gives you the true geometry in their tech notes. At least not for any reasonably modern drives.
They may tell you how many platters the thing has, and how many tracks. That is the "physical geometry" to some degree, but it does not actually tell you how the sectors are laid out on the tracks. Some tracks have more sectors, some tracks have less, and even when you know how many sectors there are per track you don't know how they are interleaved etc etc.
And that's completely ignoring the issue of remapping bad blocks etc.
Basically, if you have a drive that you _really_ know the geometry for, you could do what you suggest. I bet you don't, unless the drive is more than five years old (at which point optimizing it for speed is completely and utterly pointless anyway, as you could just buy a newer drive for yourself instead for less effort and money than it takes you to do the optimizations).
And I bet this will be even more true in the future. So I don't believe it is worth doing today, and I certainly don't believe it will be worth puruing in the long run.
That's not to say that the current algorithm in Linux is perfect. It isn't. There are probably interesting things that can be done with it. But taking the disk geometry into account is not one of those interesting things, I strongly suspect.
Linus
- 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/
| |