PE> Remember I was telling you how I switched all my data to FAT so that I
PE> could make use of the 14400k diskcache for FAT (as opposed to the 2048
PE> cache available on HPFS) and it turned out that it was worse?  Well I
PE> did some investigation today, and found that I couldn't even copy 24k
PE> of files to NUL: and then copy them again without it reaccessing the
PE> disk?!  I would have expected read-only to work.

PE> I have 64 meg of memory, all IDE/EIDE, PCI, OS/2 3.0, and observed the
PE> following things:

PC> Have you tried ibm386fs (HPFS386).  it comes with OS/2 Warp Server edition. 
PC> running it here and it FLYS with over 6mb of cache.

Well, I've tried a version that comes if you download sufficient number of
patches from IBM's site over the last 4 years, and get excellent
performance from things which gain from having the extra memory (like
messagebase rescanning, BBS QWK packing), but really lousy on an SQPACKP. 
Something like 4 times longer than HPFS16 for the SAME CACHE SIZE (2 meg)! 
I found that if I used HPFS386 and did a "copy /b AVTECH.SQD
nul:" followed by a SQPACKP, it would be faster than HPFS16, (even
including time for the nonsense copy) but otherwise, 4 times slower.  I was
tempted to add the "copy /b" to my overnight processing, but
decided instead to just not do overnight packing.  I then ditched HPFS in
favour of FAT, for security (e.g. when I attempted to install OS/2 to an
LBA drive on my 3rd IDE, it trashed my HPFS partition and then trapped
(after installing 6 disks!), but when I used FAT, it did the same thing
except didn't trash my disk.  It took me a month before I was fully
recovered from the original trashing!  I saw something in one of the OS/2
fixpacks about it incorrectly handling > 2 gig disks.  Not sure how you
are meant to install the fixpack
before you've installed OS/2 though!  I rang IBM to see what their solution
was, and they said to move the disk to my first IDE, because IDE only
handled the first 2 drives or something.  I didn't believe a word of it,
but I did it, and it worked!  I think what reality is is that there is a
bug in OS/2 when dealing with the >2 gig drives on > 2nd IDE.  It
seems it wasn't handling the LBA properly, as I
ran a program to display cyls/heads etc and it reported stupid values (in a
DOS box under OS/2), whilst it worked fine on real DOS.  BFN.  Paul.
@EOT:

---