Puppy Linux Discussion Forum

gerry · Joined: 27 Jul 2007 Posts: 53 Location: England

Explain this: I shifted the partition where Puppy keeps his save file to the end of the disk, next to the swap file. Puppy slowed down. So I shuffled things around again so that the partition with the save file in it is at the beginning of the hard disk- and Puppy is brisk again!

Puppy 2.17.1 live CD, Celeron 600 MHz, 256 MB RAM, 26GB hard disk

Flash · Posted: Today, at 12:03 am Post subject:

Here's one I think I know the answer to, I'm proud to say. Laughing

Hard disk drive controllers start writing from the outside of the disk. Surface speed is fastest at the outside, so the sustained read rate is fastest from there. The first sector of the first partition begins at the outermost rim of a hard disk. This is the best place therefore to put large files that you want to load fast, such as OpenOffice, and files that will be read sequentially, such as the ones Puppy will load at boot. If fast loading is your goal, it is important that those files not be fragmented.

Putting a swap partition near the rim of a disk is probably a waste, but it would take extensive testing to confirm that theory. I haven't bothered. On the other hand, a swap partition is usually a tiny fraction of the total disk space, so it really doesn't make much difference where you put it.

richard.a · Posted: Today, at 1:05 pm Post subject:

Flash, it's nice to have a suspicion confirmed Smile

That would account for using the whole of an IDE Syquest 135Mb drive holding the whole Pup 2.02 save file on a Pentium 100 / 64Mb RAM machine running surprisingly fast I imagine.

Also, may I suggest something associated with having your pup_save file on a FAT32 partition.

Defragmentation Smile

Something I've found is that the defragmenting utility in Windows98 + many patches is far superior to that in NT5.x + many fixpacks.

I don't know why. But I've discovered this the hard way, per favour of being able to swap OS's with plug-in hard drives in racks.

My test machine (and a couple of others) run Windows98 + whatever on account of several applications that need to access the ports directly... my mp3 player for example, and a Syquest SparQ backup drive, both use the printer (parallel) port.

If you look at the fragmentation of a typical one - or several - pup_save files in Windows 2000 (NT5.0) they are not only highly fragmented, anything up to 15 fragments, but they are all over the disk. My wife's laptop runs XP Home SP2 (NT 5.1) and that's like a dog's breakfast too.

The one on hers stays that way just in case something blows up that can and will be attributed to me lolol

However, I shut down, pull the Win2K drive, substitute a w98 drive, and defrag it and yes, it fills in all the gaps and moves fragmented stuff around until it is contiguous. And of course you can notice a difference there Smile