80-Bus News

mauve colour, magenta, would be produced. By combinations of the three primary colours eight colours can be produced, ranging from white, where all three signals are present at once, through magenta, cyan, yellow, red, green and blue, to black, where all three signals are off. Further colours are available by mixing the proportions of the above eight colours within a certain area. For instance a green dot surrounded by a number of red dots would result in a brownish red colour, the ‘brownness’ being proportional to the number of green dots within a given area. This of course implies a coarser resolution than is obtainable with pure colours, but is ideal for backgrounds etc. Some 4000 shades are obtainable in this way.

The three 16K RAMs are memory mapped into the computer in much the same way as the already existing video RAM. However, this does not gobble up vast acres of user RAM, as the colour RAM is on different ‘pages’. In other words, they are ‘paged’ in place of the user RAM, which is simultaneously ‘paged’ out, thus overlaying user RAM whilst being addressed, when video update is complete they are ‘paged’ out again and the user RAM is ‘paged’ back in untouched. For those who are worried about this paging scheme conflicting with the existing user RAM paging scheme, don’t. The AVC uses different ports to address the pages and no conflict arises. Unfortunately the ports that Nascom have adopted clash with those used by Gemini’s IVC and, although both the IVC and the AVC can be set to use alternative ports, this means some aggravation for those who already have an IVC, want to add an AVC for colour, and wart to use the standard software drivers available for each card. Shame.

Nascom have indulged in bit of cleverness in the flexibility in which the three colour RAMs can be arranged. As has already been mentioned, the three RAMs are effectively laid one on top of the next. Now imagine moving the top RAM sideways (to the right or left, it doesn’t matter), and dropping it down one layer so that it butts against the one on the middle layer. Now double the addressing speed to this 32K video RAM and a resolution of 780 x 256 results. Now two colour layers are available, the bottom one of highish resolution (390 x 256) and the top of very high resolution. The backgrounds could be produced by the lower layer to quite acceptable resolution whilst very fine detail would be comfortably resolved by the upper 32K layer. The outputs of the two RAM planes may be directed to any two of the three colour outputs producing an effective result equivalent to the highest resolution yet seen on a colour card at this sort of price in any four colours of the eight colours previously available. It was interesting to note that it was mentioned that by increasing the onboard 16MHz crystal to 20MHz, a screen format of 100 x 25 could be achieved. It was not stated whether an equivalent increase in graphics resolution to 926 x 256 dots could be achieved at the same time (should be possible).

So far we have dealt with the graphics capability. The Nascom AVC is not fitted with a character generator. Instead a clever piece of software looks up the bit patterns of alpha-numerics from a table and transfers them to the appropriate places in the RAM planes. The 360 x 256 mode produces a 40 x 25 screen format in eight colours whilst the 80 x 25 screen format is catered for by the 720 x 256 mode in four colours. This method of character generation has both advantages and disadvantages. One of the ‘prettiest’ advantages demonstrated was the ability to select character sizes and aspect ratios at will. So for instance italics could be mixed into ordinary text (in contrast colours if desired) simply by stating the ‘slope’ angle of the characters to be displayed. Alpha-numeric characters could also be placed at odd angles on the screen, and of course things like sub-scripts and super-scripts are no problem at all. The potential for this sort of character generation is quite considerable. However, there are two penalties. Speed and system RAM overhead.

The speed of screen scrolling suffers quite a bit because instead of having to only copy the character bytes from one line to the next, whole chunks of bit patterns have to be copied. Nothing too upsetting though. At first sight, ‘soft-scrolling’ would appear to be easy, and it is. Unfortunately, again, because of the enormous amount of ‘bit shunting’ required to achieve this, it is also painfully slow, too slow to be useful in fact. Another problem arises when high