1 |
On 2009-01-23, Stroller <stroller@××××××××××××××××××.uk> wrote: |
2 |
> |
3 |
> On 23 Jan 2009, at 05:16, Grant Edwards wrote: |
4 |
>> ... I found a very slick solution that lets Windows XP use |
5 |
>> a Linux swap partition for swap/paging/vm/whatever-MS-calls-it: |
6 |
>> |
7 |
>> http://db.bme.hu/~surprof/SwapFs-i/ |
8 |
> |
9 |
> That looks a really cool & useful idea. |
10 |
> |
11 |
> However, I have a reservation. Since you NEED to use it - |
12 |
> perhaps for space considerations? |
13 |
|
14 |
Yup. Mainly because I use ntfsclone to keep a bunch of backup |
15 |
copies of the NTFS partition, and having a 2GB swap file in |
16 |
every backup copy starts to eat up a lot of disk space. |
17 |
|
18 |
> - the go for it, by all means. But if this driver is slower at |
19 |
> reads or writes than Windows' own NTFS driver then it may |
20 |
> actually slow the computer down. |
21 |
|
22 |
Could be. My understanding is that the system is still using |
23 |
the Windows NTFS driver. I believe that swapfs is a |
24 |
block-device filter driver that sits between the NTFS driver |
25 |
and the bottom layer block device driver (the IDE driver or the |
26 |
SATA driver or the SCSI driver). |
27 |
|
28 |
> In the case of a driver written by an individual, who is |
29 |
> likely not as familiar with Windows' APIs, or who may not be |
30 |
> able to use the private APIs used by Windows' own filesystem |
31 |
> drivers, I find it quite possible the performance may be |
32 |
> questioned. |
33 |
|
34 |
Well, there are several individuals who've worked on it. That |
35 |
said, I can't swear that there aren't performance implications. |
36 |
There probably are, since it adds a layer between the SATA |
37 |
driver (in my case) and the NTFS driver. But, I haven't |
38 |
noticed any visible slow-down, and the machine is more than |
39 |
fast enough for my purposes. |
40 |
|
41 |
Besides, if you're running MS Windows, you've already lost the |
42 |
war when it comes to swapping performance: anybody who cares |
43 |
about swap performance wouldn't be using a normal file on a |
44 |
normal filesystem for it... |
45 |
|
46 |
Someday when I'm bored, maybe I'll google for a file I/O |
47 |
benchmark for windows and run it with and without the swapfs |
48 |
layer. |
49 |
|
50 |
-- |
51 |
Grant Edwards grante Yow! INSIDE, I have the |
52 |
at same personality disorder |
53 |
visi.com as LUCY RICARDO!! |