I should probably clarify what I said. I didn’t mean writing two things at the exact same time. What I meant was if you try to write a second file to the same disk while the first file hasn’t finished, in my experience at least both files will still be processed, but at an extremely slow pace. Anyway, I wasn’t trying to argue anything different than was already being said, so it doesn’t matter.
HDDs can not write/read multiple files on the same physical disk platter, at the exact same time (whether it is on one or multiple partitions). They can write multiple clusters / block in the same area (usually 4 clusters wide).
Microsoft recommends putting paging files on thier own partition to stop it from fragmenting.
The ideal situation is to have one page file on the system OS partition, and another on a differnt physical Hard Disk Drive, in a partition by itself.
I can tell you that it is possible to see some noticable performance gains (under heavy memory pressure) by moving the paging file off of the OS disk, if you have multiple physical disks.[/quote]
What if the second page file is on a disk with much slower RPM and seek times?–RAID 0 array (15K RPM) vs. RAID 1 array (7200 RPM) Would that still maybe see a noticeable difference, despite the RPM step down?
Also, there are 3 OS installed. 2 Vista x64 installs (one is for testing software for compatibility with Vista and software known to not play well with vista, like PTC software–MathCAD and Pro/E) and 1 Xp x64 install. Can I make the page file that I would put on the RAID 1 array a common page file for either the 2 vista installs, or even all 3 OSs? Does this make sense? It wouldn’t really matter as the primary vista install and the xp install are the only ones used–trying to squeeze performance out of the second vista or even the XP install install isn’t necessary.
Lastly; I just ran some benchmark tests on the RAID 0 array using DiskSpeed32. These are Cheetah 15K.5 drives that tested at 120 MB/s max sustained transfer rates separately, but only tested at a 165 MB/s max sustained transfer rate when in RAID 0. Could there possibly a bottle neck on bandwidth elsewhere (not on the drive) or could this be able typical RAID 0 performance for these drives?
E6750 Core 2 Duo at 3.2 GHz (1333 FSB)
4 GB PC2-6400
LSI MEGARaid 8204 ELP (no on-board cache)
2 - 146.8 GB Cheetah 15K.5 SAS
Running Vista Business x64 and XP Pro x64
This is really, really fast so it’s fine if this is as fast as I can get it–the only reason I ask is because the max speed was only 166 MB/s, so the graph was perfectly flat and suggested (to me) that there was a bottleneck somewhere else (not on the drive).
Thanks for all the advice…