1 |
Alan McKinnon wrote: |
2 |
>>>> dd if=/dev/sda2 of=/dev/null |
3 |
>>>> 957169664 bytes (957 MB) copied, 17.5531 s, 54.5 MB/s |
4 |
>>>> dd if=/dev/sda12 of=/dev/null |
5 |
>>>> 820854784 bytes (821 MB) copied, 21.4136 s, 38.3 MB/s |
6 |
>>> What do you conclude from this? |
7 |
>> I'd say that /dev/sda2 is near "beginning" of disk (outer side, |
8 |
>> more sectors per cylinder, higher transfer speed), and /dev/sda12 |
9 |
>> is near "end" of disk (inner side, less sectors per cylinder, |
10 |
>> lower transfer speed). |
11 |
> |
12 |
> Ah, I see my troll caught one already. You seem to be under the common |
13 |
> delusion that the structure reported by fdisk actually means something about |
14 |
> the physical disk :-) |
15 |
|
16 |
Correct me if I'm wrong, but I do not remember saying anything like |
17 |
that. And frankly, I can not image how you came to that conclusion. |
18 |
I just said /dev/sda2 is probably on the outer side, and /dev/sda12 |
19 |
on inner side, based on that difference in transfer speeds... |
20 |
|
21 |
> These days the entire concept of a "cylinder" is a mere abstraction to make |
22 |
> tools like fdisk work in a sane manner. |
23 |
|
24 |
Partially correct. True is, C/H/S numbers in fdisk does not have any |
25 |
practical meaning these days. But there are still cylinders, heads |
26 |
and sectors on disk. And total number of sectors reported by fdisk |
27 |
is the same, as real total number of sectors on disk. These sectors |
28 |
are numbered starting first with the most outer tracks on disk on |
29 |
all physical cylinders, continuing towards the inner most track. |
30 |
This holds true for nearly all sata/pata-disks. Maybe scsi/sas |
31 |
have different sector numbering scheme... |
32 |
|
33 |
> But yes, it does seem that in this case sda2 is probably near the outer edge |
34 |
> of the platter where the head speed relative to the disk is higher. |
35 |
|
36 |
Well, I'd say head speed does not anything to do with it. What makes |
37 |
transfer rate on the outer tracks of disk faster is higher number of |
38 |
sectors per cylinder, because nowadays disks have about constant |
39 |
"bit-density" (only "about" constant, because number of sectors |
40 |
per cylinder is a discrete number)... |
41 |
|
42 |
Jarry |
43 |
|
44 |
-- |
45 |
_______________________________________________________________ |
46 |
This mailbox accepts e-mails only from selected mailing-lists! |
47 |
Everything else is considered to be spam and therefore deleted. |