I'm not in a position [yet] to say whether the slowdown I'm experiencing
is because I'm using Smartdrv instead of Ncache, or because of the DOS
memory mis-management as compared with QEMM.
After I plow through some of the mail I'm going to try and see if it is
possible to load ncache high and maintain low memory. If I lose
conventional memory I don't know if the change will be worth it even if
ncache is faster. Who wants to crash with a faster cache?
When I figure it out I'll let y'all know what happened. One thing, dgi
files still seem to take forever, but there has been a noticeable
speedup in using TCP/IP [that's what Arachne writes to bottom of screen]
static IP setup as opposed to using DHCP/BOOTP which actually polls the
server to determine the data.
l.d.
====
On Sun, 02 Apr 2000 09:55:13 +0200 (CEST), [EMAIL PROTECTED] (Richard Menedetter)
wrote:
>>> This is also the reason why ncache2 was faster, because the
>>> testprogram favors caches with small read ahead buffer. Most accesses
>>> are random, and so a big read ahead buffer slows this artificial test
>>> down.
> B> Interesting, I tested to lower (from 56K) to the default 16K
> B> read-ahead and I got a performence decrease (not by much but it was
> B> there).
> I was just guessing ... :)
> So I was wrong ... ncache _IS_ a good cache, but generally the
> speed difference is marginal ... it is IMHO not really important what cache
> you use, but it's important to use a cache at all ;-)
-- Arachne V1.61, NON-COMMERCIAL copy, http://arachne.cz/
