80-Bus News

2. If the stack pointer is in the paged area and you use the stack the paged memory will be corrupted. If you depend on the stack contents (such as return address) still being there, then the program will crash.

MAP Software

Most MAP RAM users will probably never want to write their own software for controlling the memory paging, so perhaps the information about programming the card is not very important. MAP supply a modified version of the Gemini BIOS which supports the MAP RAM as a virtual disk of up to 512K, using a GM811 or a GM813. An alternative is to use a new version of SYS (Version 15.0 or later) which now provides support for MAP RAM for Nascom and Gemini computers, with a virtual disk of up to 960K bytes. (See article on SYS elsewhere in this issue). Most users of additional memory at the moment will want to use it as a virtual disk, which can be very useful. (See article on virtual disks elsewhere in this issue). However CP/M Version 3, otherwise known as CP/M Plus, is on the way (slowly – horribly complicated), and this requires 96K of memory as a minimum, if advantage is to be taken of most of its features. The MAP RAM will then be in even greater demand, assuming that CP/M Plus can ever be got to work on it!

[Ed.’s notes – from what we have seen so far of CP/M Plus (Gemini have a pre-release version running) the MAP RAM board is NOT the best way to provide the additional memory that is required, and it looks as though yet another method of RAM switching is yet to be born! The main attraction of CP/M Plus is its increased speed, gained by keeping various directory and data buffers in additional RAM. The way Digital Research have arranged this, it would seem that the type of memory switching provided by the MAP board is unsuitable. By the wav, there are two versions of CP/M Plus, one running in 64K or less and one running in 96K or more, but I wouldn’t hold my breath waiting for either as DRI will not yet state the anticipated release date of fully debugged versions. Watch this nag. for further details.]

The Great A19 Debate

The MAP manual points out that the GM813 does not bring address line A19 to the 80-BUS. While this is quite true, the GM813 was merely following the 80-BUS standard, which does not allow for A19 on the BUS. The obvious place for it is line 49, which was instead allocated as an additional ground by Nascom some time ago. This extra ground line does not appear to be needed, and everyone now seems to agree that the 80-BUS specification should have had line 49 as A19. But what should be done about it? Without A19 the maximum memory size is limited to 512K bytes, and people are already installing systems with a full megabyte. The MAP approach has been to unilaterally redefine the 80-BUS and they have used line 49 for A19. They suggest that anyone with a GM813 who wants to have more than 512K should modify it so that it no longer grounds line 49, but connects it to A19 instead. But if you do this you should be careful to check all the other cards in the system, and the motherboard itself, as any of these may have implemented the 80-BUS specification and grounded this line. It would be nice if Gemini agreed that this was a good idea, but I suspect that instead they may feel that it would be better to define some other line as A19, in order to keep to the universally agreed specification and avoid telling people to mutilate the GM813 [Ed. – and all other Gemini boards and some boards of other manufacturers, including Nascom, which as per spec. have this line grounded.] and their motherboards. On your system it is probably wisest to follow the recommendations which come with the large RAM card you buy, and make a note of what your 80-BUS does. See INMC 80 Issue 4 for the 80-BUS specification.