Test results
Originally posted by dealer
yes, Itryed launching aol...mostly, as I said, the differances are hardly noticeable at all..so much so, the improvement had to be measured with a stop watch...
For a fair test you need to reboot after each test - the second time you use ClearMem the addresses have been made in RAM & data retrieval is almost instantaneous.
Following the advice of the many I put the page file closer to the outer tracks, on the Master drive.
Primary partition 8meg for system files
1st logical 1gig FAT16 32kb cluster for the pagefile
2nd logical 5gig FAT32 8kb clusters for windows
single disk test FAT32
1st test default setup pagefile fragmented into 4 parts - test time 8.1sec
2nd test 512meg fixed pagefile on the boot drive defragmented into a contiguous streak - test time 8.5sec
3rd test 512meg fixed pagefile on the 1st logical 1gig FAT16 32kb cluster - test time 6.2sec
2 disks on 1 controller
1st test 512meg fixed pagefile on the slave's last logical partition (inner tracks) test time 2.4sec
2nd test 2 x 384meg fixed pagefile 1 on each disk - test time 4.1sec
3rd test 512meg fixed pagefile on the slave's Primary partition 10gig FAT32 8kb cluster - test time 2.2sec
resized the slaves primary partition 1gig & formatted FAT16 32kb cluster for the
4th test quickest results yet @ 1.9sec ;-)
Then I did similar tests using NTFS
on the boot drive the test faired slightly better from 8sec to 6sec
on the slave drive NTFS it was roughly the same @ 2.5sec
but on the dedicated partition NTFS 32kb cluster(slave) it did slightly worse
from 1.9sec up to 3.2sec
the reason for this anomaly would be the MFT zone fragmenting the pagefile
In conclusion the most optimal place for the pagefile
is on a separate NTFS disk to windows or
if you want a dedicated partition it must be FAT