Nuno Magalhaes wrote:
(Nuno Magalhaes) wrote in message . com...

If you have a given sector, say 5000th sector (logical), how can you
get the physical location in: platters, tracks and sectors?


I've measured disk access times from sector 0 (track 0) to the start
of other tracks and got this results:

from to access time (ms)
0..126 - 1.3568 (track 3)
0..189 - 2.0449 (track 4)
0..252 - 2.7442 (track 5)
0..315 - 3.4373 (track 6)
0..378 - 4.1318 (track 7)
0..441 - 4.8222 (track 8)
0..504 - 5.5197 (track 9)
0..567 - 5.3046 (track 10)
0..630 - 4.6257 (track 11)
0..693 - 5.3169 (track 12)
0..756 - 6.0181 (track 13)
0..819 - 8.9897 (track 14)
0..882 - 9.6792 (track 15)
0..945 - 10.3711 (track 16)
0..1008 - 1.8192 (track 17)
0..1071 - 2.5135 (track 18)
0..1134 - 3.1956 (track 19)
0..1197 - 3.8969 (track 20)
0..1260 - 4.5939 (track 21)
0..1323 - 5.2856 (track 22)
0..1386 - 5.9845 (track 23)
0..1449 - 6.6760 (track 24)
0..1512 - 7.3691 (track 25)
0..1575 - 8.0571 (track 26)
0..1638 - 8.7536 (track 27)
0..1701 - 9.4473 (track 28)
0..1764 - 10.1306 (track 29)
0..1827 - 10.8335 (track 30)
0..1890 - 11.5330 (track 31)
0..1953 - 1.1275 (sometimes 12 ms) (track 32)
0..2016 - 3.6675 (track 33)

What should this mean? That there are 16 platters and the sectors are
enumerated from the inner cylinder to the outter cylinder?

Have you averaged these access times ? Rotational latency can create some pretty
bad one-shot access times : let's say you want to start reading block 0 and it
just passed by the head on the platter, the disk will have to wait for the
platter to spin 359 degrees before it can start reading block zero.
So you should average at least 5 attempts per block range.
Also, were you doing reads or writes or both ?

Because I want to measure disk rotation times (to generate random
numbers according to a theorem) and this seems strange.

Isn't the disk rotation constant and determined by the model, eg. 10000 RPM ?
I don't know if you can use disk access times as input for a theoretical model;
the disk's firmware usually has all kinds of exotic optimisations to cut corners
on access times wherever it can, which can cause wildly varying (and
unpredictible) access times


Arne Joris