80-Bus News |
July–August 1983 · Volume 2 · Issue 4 |
Page 18 of 55 |
---|
This board would be very similar to the board proposed above. However, the speech synthesiser chip would be replaced by two AY-3-8910 programmable sound generators. The outputs of the two chips would be fed to a DIN socket, to allow connection to a high fidelity amplifier. Since the AY-3-8910 chip provides further output lines, it could control digital to analogue converters, also connected to the board output, enabling the synthesis of far more sophisticated wave forms than those provided by the AY-3-8910 itself. An advantage of this board over the more normal sound board with no processing power of its own (such as the late, lamented Winchester Technology sound board) would be the ability to “sweep” the frequency and other control registers of the sound chip without taking up time needed by the main system processor.
This board would consist simply of the above two boards combined into one. The output ports of the two AY-3-8910’s would control the SC-01 chip, as well as two digital to analogue converters. There should be ample space for the port decoding logic, a Z80, ROM, RAM, two AY-3-8910’s, two DAC’s and an SC-01 on an eight inch square board. After all, Gemini can fit 512K of RAM in that space!
In many applications, such as chess programs, the speed of the main processor is not sufficient to calculate as far ahead as is desirable. If it were possible for the main processor to generate moves, and pass the new positions created to other processors for evaluation, speed improvements would result. This would mean that a comparatively simple program could achieve better results by looking further ahead than is usually possible. Sargon, for example, does not usually look more than six plies deep, and can take hours to do so. A suitable board for this kind of application could be constructed using the following:
(a) Port decoding logic for eight ports.
(b) Four Z80 processors, each operated via two ports.
(c) ROM and RAM for each processor.
Thus the board would contain, in effect, four subsidiary computer systems, each able to run programs independantly of the others, and of the main processor. Since there would not be a lot of room for RAM on this board, it would be fitted with suitable connectors to enable it to be expanded on separate boards. There are many applications besides chess which require rapid results from a large number of calculations. For example, the calculation of the new coordinates of a complex moving shape, for animated 3-D displays. A board of this type would be of considerable assistance in producing fast displays with the Pluto graphics board, where it frequently happens that the Pluto board has to wait for the main processor to generate the next picture in an animated sequence. The matrix multiplication calculations used in creating moving displays are just crying out for this sort of parallel processing. Anyone feel like writing an Occam compiler for the board? Oh all right, I will do it in a spare hour, when one comes along! [Ed. – for the idiots (like me), what is Occam?]
This would be similar in concept to the board described above, but would carry only one processor, of a considerably more powerful type. As well as the new processor, there would be operating software in ROM, large amounts of RAM, and the ability to extend the board further. The board could use the Z8000
Page 18 of 55 |
---|