Windows 2000 TEMP and SWAP location (NTFS & FAT32)

[DocDoo]

I have a 2 partition HD.

C:\=Win2K (NTFS)
D:\=FAT32

Wanting to help lower disc fragmentation on C:\, can I point Win2K to my D partition for the TEMP and SWAP files?

Does the fact that D is FAT32 make it an issue?

 

[StuckMojo]

put your swap file on C, because it is faster (outside of disk)
you can move the temp dir by changing the environment variable in system control panel

 

[DocDoo]

Actually, moving the SWAP to a seperate partition (but on same HD) does not make it slower. The only advantage it has is lowering fragmentation on the system disc. And since a fagmentated disc has more of a threat on the overall speed than placement of the SWAP (given it's on the same drive), this is my only goal. Ideal would be on a seperate HD on C:! Then you get the best of both worlds.

I am only concerned about placing the TEMP and SWAP files on a different file system (like FAT32), since my Win2K is on a NTFS...

 

[StuckMojo]

actually, it will make it slower. when you partition a disk, the first partition is closest to the outside of the disk, which travels faster (linearly), and thus has a faster transfer rate. so if you put it on D, it will be a tad slower.

if you think about this for a few mins, you should be able to figure it out (given that partitions start at the outside, and move in, disks are round, that they spin at a constant angular speed, and clusters are a fixed physical size).

as far as fragmentation goes, just set the swap file to be a fixed size and it will not be responsible for causing any additional fragmentation. after doing this, use a defrag util that can defrag the swap file so it's in one contiguous piece.

frag wise, moving temp could be a good idea. depends on how you use your system.

putting either on fat32 should make no appreciable difference. fat32 doesn't support file ACLs or mutiple streams in a file, which will never matter for the page file, and is highly unlikely to matter for temp.

 

[StuckMojo]

actually, it will make it slower. when you partition a disk, the first partition is closest to the outside of the disk, which travels faster (linearly), and thus has a faster transfer rate. so if you put it on D, it will be a tad slower.

if you think about this for a few mins, you should be able to figure it out (given that partitions start at the outside, and move in, disks are round, that they spin at a constant angular speed, and clusters are a fixed physical size).

as far as fragmentation goes, just set the swap file to be a fixed size and it will not be responsible for causing any additional fragmentation. after doing this, use a defrag util that can defrag the swap file so it's in one contiguous piece.

frag wise, moving temp could be a good idea. depends on how you use your system.

putting either on fat32 should make no appreciable difference. fat32 doesn't support file ACLs, mutiple streams in a file, or hard links, which will never matter for the page file, and is highly unlikely to matter for temp.

 

[StuckMojo]

crap..my small edit accidentally became a double post
doh!

 

[DocDoo]

There needs to be a correction here, so people don't get the wrong idea.

The outer edge of a disc does not travel faster! This would be mechanically impossible.

What does happen is there is more information that can be read/stored within the same single RPM.

Take a 5 inch disc and draw a circle at 3" from center and 5"

If the disc is spun at 1 RPM the 3" circle would have a length of 9.42"
If the disc is spun at 1 RPM the 5" circle would have a lenght of 15.7"

So, you can see the disc is not traveling faster, but the information on that same revolution is greater! And if the arm of the needle starts from the outer edge, this is compounded.

This is proven correct on paper. However, I think the jury is still out on the "real-world-results" on this (again, only if it's on the same HD).

On my Win98, WinME system, I have been unable get a benchmark that showed it any different. Even when working with my day-to-day tasks...

FWIW: Currently, I have 128MB and have had the SWAP set to, Min 192, Max 384 (default).

Aside from that, I agree with ya StuckMojo. I thought it was important to clear that up a bit

 

[LocutusX]

<< This is proven correct on paper. However, I think the jury is still out on the &quot;real-world-results&quot; on this (again, only if it's on the same HD). >>



I'm not sure what you mean by &quot;real world results&quot;... (real world benchmark results or end-user performance?) But if you look at the results of a good raw media read/write benchmark such as HD Tach, you will notice a gradual recession of speed from the beginning of the drive to the end of the drive.

 

[DocDoo]

LocutusX... Yes, I have also verified the same results when using HD Tach.

My only point is, if it wasn't for such tools could you really tell otherwise.

While I will never argue the fact that placing a SWAP on a seperate physical HD had a definite improvement, I have yet been able to find otherwise when it's on the same HD, C, D or E(as far as speed goes).

Psychological improvements are all around us, be wary

 

[StuckMojo]

depends on your definition of &quot;travel faster&quot;. the angular velocity is equal, of course. from your own example:

@3in angular velocity = 1RPM : linear velocity = 9.42in/min
@5in angular velocity = 1RPM : linear velocity = 15.7in/min

15in/min is faster than 9in/min.

i think that i would make the case that &quot;travels faster&quot; is the correct phrase. a point on the outside of a circle has to cover more distance in a single revolution, therefore it travels faster.


anyway, i agree that it probably doesnt make much difference anyway. partictularly since pagfile.sys gets stuck after most other files on C, so probably ends up pretty close to the beginning of D anyway.

you know, the best way to reduce system fragmentation is to remember to defrag c once a week the tool included in win2k is really diskeeper, and actually does a pretty damn good job if you run it twice.

 

[StuckMojo]

i think we're just disagreeing on symantics here.

 

[LocutusX]

Well of COURSE I can tell the difference between having swap at the beginning or at the end... after all, as a Borg, my internal chronometer is accurate to one millionth of a second.

 

[DocDoo]

We are all on the same page We are debating over the &quot;coin.&quot;



<< i think that i would make the case that &quot;travels faster&quot; is the correct phrase. a point on the outside of a circle has to cover more distance in a single revolution, therefore it travels faster. >>



But the word travels faster still sticks to my side :Q

The &quot;point&quot; moves at the same speed (as RPM's go), but you can pack alot more &quot;points&quot; within the same given RPM. This is a matter of &quot;density,&quot; not speed!

Lets simplify this to a basic analogy. If Johnny buys the new Abba LP (33 1/3 RPM Album) and plays it on his record player, does the first song on the record play faster? Does it sound like Alvin &amp; the Chipmunks?

No and no. However, the first song takes up the least amount of space when measured from center to outer.

What is the point to all this?.... Johnny had bad taste in music!