80-Bus News

Is there anyone out there? (Dr Dark believes you are all a figment of the circulation manager’s imagination. [Ed. – Circulation manager – who’s he?].) I know there are at least five of you, I’ve got your letters here courtesy of our esteemed Editor. The following few pages attempt to answer some of the questions asked in these letters, and are based on the premise that if one person has actually taken the trouble to write in with a question, there must be at least a thousand of you out there with the same question on your lips. So, as these pages are to be fueled by your questions/​comments/​hints/​tips, send them in – even if it is only one sentence on the back of an old envelope. Do not expect a personal reply. (Letter writing is not my forte). You may get a reply via 80-BUS News in due course. If a few people ask the same question there will almost certainly be a reply via the News (unless there are obvious signs of collusion like the same handwriting!).

This issue I start with a hardware bug in GM812 (the Gemini IVC) that was discovered sometime ago by Richard Beal, and has also been found by Nils Nazoa-Ruiz. It will only manifest itself if you are using interrupts, and only then in special circumstances. If you happen to be unlucky you will find that the IVC is reset everytime an interrupt occurs. The explanation for this lies in the way the IO addresses are decoded on the card. This is done via a PROM (IC25) which is enabled by IORQ. With the standard PROM the four outputs are used to signify addresses 00-07 (NASIO), address B1 (data port), address B2 (control port), address B3 (card reset). The B1 and B2 outputs are qualified by RD and WR before they are used, but the B3 output is taken straight to the RESET input of the IVC’s Z80. If you look up the interrupt behaviour of the Z80 in its data sheet you will see that the Z80 acknowledges an interrupt by putting out an /IORQ signal along with an /M1, a unique signal combination that is used at no other time. Unfortunately if the lower 8-bits of the address bus just happen to hold the address B3, then the IVC is reset.

The cure is simple (in theory), and that is to disconnect pin 14 of IC25 from GND, and connect the backplane /M1 signal to it via a spare inverter on the board. Pin 14 is an additional enable input (active low), and with this modification the decode PROM will be inhibited during any cycle where M1 is low, thus preventing the interrupt acknowledge cycle being misinterpreted by the IVC. The normal IO operations will continue to work as before.

Next a letter from Alan Mackay – here is someone who is actually using his computer to compute. I reproduce his letter below in the hope that someone out there can help.

“The most useful of the features of BASIC as originally developed at Dartmouth College was the provision of matrix operations; multiplication, transposition and inversion being each accomplished by single statements. Most implementations for microcomputers do not provide these essential routines. Can anyone tell me how I can do matrix inversion on a Nascom 2? Does anyone have assembly language programs for matrix operations in BASIC or Pascal? More generally, is there a library of serious mathematical subroutines – such as the NAG (Numerical Algorithms Group) provides for Algol and Fortran on large computers – for the Nascom 2? Is there anyone doing serious calculations, as opposed to computer games, on a micro?”

– replies to Alan Mackay, __ __________ ____, Highgate, London __ ___. I recall a product MAP14Z that was available in 1980/1 from Enertech Ltd, __ _________ ____, Eastbourne, East Sussex ____ ___. This was a 40-bit floating point package available for the Nascom. It did not support matrix operations, and a review of it appeared in INMC80 issue 2 (Sept80-Jan81).

Tony Chamberlain writes in with a list of the undocumented op-codes for the Z80. These have been covered before in various of the glossies and earlier INMC80s/80-BUS News, but if you’re new to the Z80 and have missed them and want to know more send a note to me asking for it to be published in a later issue.