PLUGGING IT IN
That is almost all you have to do! The board includes one mini-DIP
switch, (4 pole), and two links.
The mini-DIP switch is used to select the board number in a multi-RAM-DISK-board
system and, in most cases, will not need changing from its setting of
0, (Unless Gemini change their test procedures and they come set to some other
LK1 will provide you with a NASIO signal if your system requires it, but
most people will not need it, or will have another board in the system
supplying it already.
LK2 allows the clock input to the board to come from either the CLK line, or
the AUX CLK line. If you do not have a 4MHz system clock, you will have to cut
the trace between 1 and 2 on LK2, and connect 1 to 3 in order to pick up 4MHz
from the AUX CLK line. (Assuming it has been implemented on your system).
Gemini BIOSs Versions 2.3 and higher support GM833. Also the associated
program CONFIG has been extended to include the board as an option when
setting up the parameters for a “Memory drive”. If you have an earlier BIOS, a
BIOS update program (together with CONFIG) should be available through your
The BIOS supports the drive as drive “M”, and is so arranged that it does
not re-initialise the memory drive if you are forced to press Reset.
The standard board provides a memory drive of 512k (or 0.5Mbyte) capacity
– quite a respectable size. However if this is not enough for you further
boards may be added up to a maximum of 16, so providing the full 8Mbyte
capacity that CP/M2.2 will support. (N.B. 16 boards are beyond the capacity of
the currently available commercial 80-BUS backplanes, and may also require
added power supply capacity!).
I find the 512k single-board drive more than adequate. It lets me keep a
reasonable amount a system software on the drive (such as Wordstar Overlay
files) along with all the data files. You may like more if you are handling
particularly large data bases.
GM833 in use – BENCHMARKS
To give an example of the benefits of GM833 a few Benchmarks are shown
below. So that a reasonable assesment can be made of relative performance I
have also included figures for 5.25" and 8" floppy-disk drives, and the Gemini
Winchester disk subsystem. As always the figures should be taken as a guide
only, as the figures for the floppy disk performance can be made to vary
widely, depending upon where the files are located on the disk. I have
attempted to position the files in such a way as to produce ‘average’ figures.
The first benchmark is an example of how a program can be transformed by
the use of RAM-DISK. The program in question is TRANSLAT, a program that
translates 8080 mnemonics to Z80 mnemonics. Who ever wrote the program made no
attempt to optimise the IO – the internal input and output buffers appear to
be one sector in size. The result is that the majority of the run time of the
program is taken up by the disk drive moving the head back and forth between
the tracks holding the input and output files. The RAM-DISK has no such
(N.B. another such program nearly gave me a heart attack when I used it
to compare files on two different drives. It briefly turned my system into an
imitation of a machine gun as the heads on the two 8" drives alternately
loaded and unloaded several times a second.)
Anyway here are the figures for TRANSLAT working on a large 8080 source
file. The number in brackets is the approximate number of tracks that
separated the source and destination files.