Written by David Tebbutt, PC Dealer 10/89 item 01 - scanned
Have you ever seen programmes on television about Copacabana Beach in Rio de Janeiro? Or maybe you've even been there. I haven't, but I've got a friend who lives at nearby Ipanema, and he tells me that it's not worth going near the beach unless you really look good with very little on. (I don't, and that's my excuse for leaving Rio out of the Tebbutt family holiday plans).
The point of mentioning all this is that it reminds me of computer manufacturers at the moment. Posturing seems to be a very important activity. Did you go along to the Personal Computer show? I think half a dozen companies were crowing about their 80486 based computers. And plenty more outside have been banging on about how they're going to be right up the front with their own 80486 offerings.
Posturing about this still buggy chip reached the dizziest heights last week when IBM got all excited about actually installing its first 486 machine. It was the Copacabana equivalent of showing a well-developed gluteus maximus - very important to the owner but still only of marginal interest to the world at large.
I suspect the reason for this excitement is that IBM cannot forget the huge amounts of publicity and, more importantly, business that Compaq derived from its early commitment to the 80386 machines. Without question, that particular move did Compaq a lot of good and now the 386 is rapidly becoming the processor of choice among corporates. The sad thing for IBM is that its announcement will not have anywhere near the same impact.
Another announcement, and some more posturing, took place at the same meeting. IBM announced its most recent MCA developments. The press briefing took place the same week that Eisa was scheduled to make its own announcements.
We were deeply suspicious of IBM's motives but, since the Eisa briefing didn't materialise, we'll never know whether it was a deliberate spoiling tactic.
Anyway, the essence of the MCA announcement was that it can hot up the bus to carry up to 160 million bytes per second.
Phew! Today's MCA speed is one eighth of that. At Copacabana, this would be the equivalent of the best pectorals and biceps on the beach.
Like the fanatical body builder who takes steroids, IBM's improvements appear to carry a price. At the moment small packets of data are thrown around the bus, each with their own heading and control information attached.
IBM decided that for transfers of contiguous data, there's no need to keep sending this addressing information.
The header information can be sent and then all the data can be thrown on the bus without dividing it into separate packets. The capacity used by the headers can now be used by the data. This just about doubles the bus speed for these transfers to 40Mb per second.
It also noticed that the address bus lies idle during much data transfer so why not multiplex the transmission, sending data down, the address lines too? And it has decided that it can halve the cycle time of the Micro Channel bus.
The reason I say that this comes at a price is because IBM also quietly announced a 'Data and Address Parity and Synchronous Channel Check'. This is described as 'additional fault detection and isolation features'. My guess is that as the MCA's behaviour becomes more sophisticated, you're going to need as much protection from errors as you can get.
So, where does this leave Eisa? Bloody annoyed I should think. Just when it's finished defining the standard and is getting ready for its first deliveries, IBM lets rip a substantial raspberry.
Eisa has already had to shift its ground once. I seem to remember it started as an extension to the industry standard bus. Its aim was to provide a realistic alternative to IBM and to protect PC users' choice of, and investment in, expansion cards. Now it sees itself as the server standard, in which case I can't see that protecting investments has much to do with anything.
I'm beginning to wonder if Eisa won't end up as the Copacabana equivalent of a perfectly formed Achilles tendon.