As many people here have other old computers besides for NeXTs. Does anyone have the user manual or datasheet for a MC68551? This was the MMU used on a 68000 through 68020. I cannot seem to find a scan on the internet; however, these things are sometimes around.
--degs
I have never seen a 68551 in the wild and it seems to be rarely mentioned on the net. In fact, most of the sources may also be typos - are you sure the 68551 existed?
So far, I have only seen the 68451 (for 68010-based systems) and the 68851 (for 68020-based systems) used - and a number of custom MMUs, such as the MMU in the Sun 3 series. The 68000 could not (easily) use paged virtual memory, since it did not save enough information on the stack to reliably trap and recover from a failed memory access; this was corrected in the 68010. Several 68000-based systems had interesting solutions to implement virtual memory - from running a second 68000 processor that delayed /DTACK and handled the page fault (early Apollos AFAIK) to ensuring that certain instructions did not end up being located in memory across a page boundary (7th Edition Unix on Sun 1, if I'm not mistaken). I think most of these used swapping, not paging. In fact, I'm currently researching a Codata 68000-based system running a Unisoft port of 7th Edition Unix - a very rare and interesting Multibus machine.
Literature on the 68851 is easy to find, the first and third edition of the 68851 data book are available at bitsavers:
First edition:
http://bitsavers.trailing-edge.com/pdf/motorola/68000/68851_PMMU_Users_Manual_1ed_1986.pdfThird edition:
http://bitsavers.trailing-edge.com/pdf/motorola/68000/68851_PMMU_Users_Manual_3ed_1988.pdf[/url]The 68451 data sheet is harder to find, but there's a copy at
http://www.thebackend.de/~sb/mc68451.pdfHTH, Michael
Thank you for the information. 68851 is correct; I had a typo in my notes apparently. I have a 68000 processor here that I'm thinking of doing something with in the future. I'm currently at a standstill until my parts come in next week for the 68060 board, so I was spending my Sunday taking stock in what I had.
--degs
If you want to build a nice 68000 system, the Codata CPU board reference manual (which is a licensed clone of Sun's 100U CPU board used in the Sun 1 machines) is an interesting read. It has full schematics with explantation (no custom chips used except for three small PROMs to decode addresses) and provides a MMU built of discrete components and fast SRAMs:
http://bitsavers.trailing-edge.com/pdf/codata/05-0004-01_Codata_CPU_Board_Manual_Jul83.pdfI have a different version of the manual here (with a better quality of the scans) which I can send you if you're interested. I am thinking of recreating the Codata machine on an FPGA.
-- Michael
Quote from: "cuby"If you want to build a nice 68000 system, the Codata CPU
That is a pretty nice CPU board. I'll keep it in mind. I'm currently just fooling around with ideas, but that would be very easy to implement on a FPGA. I'll let people know what I come up with.
--degs