C Ltd
This message turned up in search, but its forum couldn’t be identified from the original transcript, so it may not be linked into its thread.
Actually, John, it's not a transfer rate of 100k/sec – it's more like 220
k/sec. The _diskperf_ program comes up with a number that indicates up to
100k/sec for it's operations. Many operations will show very nearly the
same performance as if you had a DMA controller. One thing to consider is
that since this interface is a programmed I/o interface, when the disk
transfers are occuring, the CPU is _very_ busy, and other tasks suffer for
it.
An instance where this matters is when you have more than one task doing
disk i/o. An instance where it does NOT matter is when you are running a
program, or copying one. For instance, a couple of days ago, I was on the
phone with someone who has a 2090 system; she has just installed 1.3 and
the FFS on it. We both have, and use DirUtil III. Both of us experienced
virtually the same time when dragging the icon from the HD window into the
ram disk window – about 1.5 seconds.
Dirutil is 23k; it's not really a function of the HD interface at that
point, it's more the DOS. Both system, which were running 1.3 and FFS at
the time, use 68000's and are similar in many other ways, the only notable
exception being the 2090 itself. Figuring a 23k read and a 23k write (into
ram, no less) that's only a 40k transfer.
Even using the diskperf figures for my system (97k/sec for the slowest
number read spec), a good amount of time is unaccounted for. Enter
AmigaDOS, the worlds slowest data-handler. <grin> Catch you later, and I'm
glad you got your system up and going – btw, couldn't have helped you in
any case – ROMAT is not our product, and I know ZIP about it — we did the
ST-506/Adaptec stuff, and associated formaters and so on; the SCSI stuff
was done at C LTD. Remember that if it breaks! heheheh –Ben–