Graphics on the ARM
Another RISCWorld exclusive serialisation.
Epilogue - The Future
One of our philosophers has said that if you want to know where you are going, sometimes you can get a clue by seeing where you have come from. So let's begin our look into the future with a nostalgic look at the past.
Ten years ago the BBC Microcomputer was launched by Acorn in connection with the BBC Computer Literacy Project. Even though the Sinclair Spectrum from the same period sold in greater numbers, the influence of the BBC Micro was unequalled because of its widespread use in schools; this was the machine on which millions of schoolchildren gained their first hands-on experience of computing.
The original BBC Model B had just 32 Kbytes of RAM; the first disc drives for it offered a mere 100 Kbytes of storage. Its highest resolution screen mode was mode 0, 640 x 256 pixels and 2 colours only. I bought a BBC Micro in 1984 and used it for both work and leisure activities until 1989 when I bought an Archimedes A310; I still own that BBC Micro and it remains in daily use, although by my sons rather than myself.
In all the time that I used that BBC Micro only twice did I run out of memory. When I changed over to the Archimedes with its seemingly massive 1 Mbyte of RAM. 32 times as much as in the Beeb, I reflected that it was unthinkable that I or anyone else could ever run our. of memory on so vast a machine. I was wrong. So completely wrong that within a year I was compelled to trade it for a 4-Mbyte Archimedes. And even on that and a 4-Mbyte A5000. Running out of memory is an all-too-common occurrence.
Why the difference? If 32 Kbytes were adequate once, why are 4 Mbytes barely adequate now? Because computing suffers from its own version of Parkinson's Law. Applications and data expand to fill the memory available. In common with other ARM users, I am undertaking tasks today that were beyond my wildest dreams when I first became acquainted with the BBC Micro. Requirements and expectations have grown even faster than the hardware.
So, in the past ten years we have progressed from 32 Kbytes of RAM and 100 Kbytes of floppy disc storage to 4 Mbytes of RAM, 1.6 Mbyte floppy discs and, commonly, 100 Mbytes or more of hard disc storage. What can we expect of the Acorn machines that will be available in the early years of the new millennium? Simple extrapolation suggests machines having 512 Mbytes of RAM, 100 Mbytes of demountable or floppy storage and 10 Gbytes (10,000 Mbytes) of fixed storage. And as to their graphics capability, a screen mode of 1280 x 1024 pixels supporting full 24-bit colour does not sound far-fetched; the screen memory requirement of 3840 Kbytes is paltry in such a machine. And all must of course be supported by dramatic increases in processor speed.
And I am prepared to bet that even with such machines as these, running out of memory will remain a problem. Why? How could one possibly Fill a machine having 512 Mbytes? It's easy. Table 10.2 suggests one way. There are scanners available now that scan A4 documents at 1200 dpi and in 24-bit colour. The resulting sprite (without compression) occupies 3~8 Mbytes which goes a long way towards filling 512 Mbytes of RAM. And that, I repeat, is available now. And there is a direct-drive laser printer available now that will handle A3 sheets at 1200 dpi. The printer image without compression occupies 32 Mbytes and is presumably buffered on hard disc in present machines. And scanner and printer technology will have developed further in 10 years time. We might reasonably expect to see 24-bit colour laser printers commonplace then. And they would certainly be more demanding in RAM than today's machines.
All this will need to be supported by storage devices of ever increasing capacity. Perhaps the 'floptical discs' currently widely advertised which offer 21 Mbytes of demountable storage will replace the 3.5 inch floppies that they so closely resemble. Improvements in technology may increase their storage capacity to my extrapolated 100 Mbytes or more over the next decade. Already CD-ROMs and WORM drives are offering greater storage than is available on hard disc.
But I confess to uncertainty as to the long term future of electromechanical storage systems such as floppy and hard drives and even CD-ROM. All suffer ultimately from reliability problems-even the best-made bearings wear out in time. The future of storage media surely lies in purely solid-state devices which of course have no moving parts. Already one company serving the PC world is offering a product called 'Silicon Drive' for use primarily in harsh industrial environments where dust, damp. vibration or electromagnetic interference preclude conventional floppy and hard discs. In Silicon Drive the storage medium itself is a PCMCIA card, a credit card sized demountable RAM board which is inserted into a slot on the drive, like a miniature floppy disc. On this tiny card is a staggering 2 Mbytes of RAM backed up by an ultra-thin battery. Flash EPROM-based versions are also available and need no battery. The present system provides for future cards having storage capacities up to 64 Mbytes. No doubt tomorrow's technology will extend their capacity still further. Here is a more likely contender for the first 100 Mbyte 'floppy'.
Silicon Drive connects to the computer's IDE bus and, since IDE is now an Acorn standard, presumably, with suitable software, it could be used now on A5000s and other ARM machines equipped with an IDE interface. At present, expense limits this medium to critical industrial applications, but we all know how prices tumble as technology catches on. And this particular technology has the potential to catch on in fields far broader than just pure computing. For example, within ten years it could revolutionise the audio/video world, superseding today's mechanical CDs and DATs as the ideal medium for digital audio and video. Imagine jogging listening to your favourite music on a solar powered walkman reading its digital audio from a credit-card sized ROM board!
So much for the hardware, what of the graphics software? Clearly that will develop to make full use of the hardware facilities, removing many of the restrictions that frustrate creativity today. So animations will get longer and will use the full screen, perhaps even in new ultra high resolution screen modes. Combining tomorrow's vast RAM banks with today's less demanding screen modes might make sprite-based interactive animation feasible. Animations may acquire sound effects-that is quite feasible now, of course, but beyond the scope of this book. The faster processors may make Bezier curves feasible in vector graphics-based animations, allowing more realistic graphics in flight simulators and other interactive animations. Stereoscopic graphics could use virtual reality-style viewers-or, more cheaply, a split screen monitor display with a special attachment to ensure that each eye sees only its half of the screen. But otherwise, who knows? The very challenge of the future resides in the certainty that none of us knows what is waiting round the corner. Ten years ago I could not have dreamt I should be writing this today. I have conjectured about what lies 10 years in the future. But the reality may in fact be beyond our wildest dreams.