PE> I've set it up with a 4 meg disk cache.
PE> I've also got a 4 meg FAT cache
PE> and a 2 meg CDROM cache
PE> and a 1 meg VDISK,
PE> so total extravagence = 4+4+2+1 = 11 meg,
PE> out of 20 meg,
PE> leaving 9 meg for programs + OS. That should be enough.

KR> there is something fundamentally weird about using more
KR> memory for cache than actual use.

BG> I can almost understand the 4Mb HPFS cache, but 4Mb for FAT ??
BG> Even IBM themselves recommend no more than 512Kb for the FAT cache

There are no such thing as absolute numbers. What matters is if it
goes significantly faster with the bigger cache. Precisely the same
thing happened with the HPFS cache, IBM pontificated that the max
available under the standard HPFS was plenty, no one would need more.
They were wrong, some ops speedup up phenomenally.

BG> (Paul must still have fond memories of the inefficient smartdrv, I think).

It doesnt have that much to do with the implementation tho there is
a bit of that involved. Even with the alg used in the OS2 FAT cache,
you can certainly find specific work which benefits from >512K

And the other vital question is your task mix. If you never use the
extra physical ram anyway, it ALWAYS makes sense to use it for stuff
like caches.

BG> And a RAM disk under OS/2 (which has no 640Kb limit) ??
BG> Now I've heard everything!  |-)

There are a few situations where you can know with certainty that you
want to keep some stuff in ram and the cache is too stupid by itself
to get as good a result. Corse what you really need is an elegant way
to tell OS2 'if this gets into the cache, never replace it with anything
else in the cache' OS2's cache is far too primitive for that tho.

--- PQWK202