Mkb@CMU-CS-C.ARPA (12/10/83)
From: Mike Blackwell <Mkb@CMU-CS-C.ARPA> Now that I have a slightly positive cash flow, I'm seriously considering putting together a computer system of my own to experiment and hack on. It's going to be 68000 based: that's what I use at work, so I know it quite well. Currently, I am thinking of: Compupro CPU-68K processor board ($565) Lexicomp SR-64K 64K 8/16 bit wide memory ($300) U.S. Micro S-100 Mod motherboard/cage/power supply/enclosure ($395) Suitable bus termination Some serial I/O board. You'll note that there's no disk or controller: for the moment, I can use cross compilers etc. on a Vax and download. Slow, but useable for my needs. When my budget recovers enough, I can start thinking about disks and operating systems. So my questions to everybody out there are: Are there any hidden gotcha's in what I've chosen? Any comments, pro or con, on the products? Will I have any problems later when I add disk controllers? Better prices on any of this? Other boards I should be thinking of? Also, what is a good serial IO board? I don't need anything fancy, just four vanilla RS-232C ports would be fine. All the boards I've looked at have frills I don't want (ROM, clocks, parallel IO, etc), and seem ridiculously overpriced... Thanks very much for any help you can give me! cheers, -m- (mkb@cmu-cs-c)
burton@fortune.UUCP (Philip Burton) (12/12/83)
Really?? If you are going to use cpm, just get a z80, and save many $$$. I have a cpm/z80 at home, and a Fortune system at work, and I love UNIX. I've also reviewed cpm/68K and, yes it has a few features beyond cpm2.2, but UNIX it ain't. I've talked with Gifford and gifford people who tell me there's little demand for the Godbout 68K with cpm board. My opinion, worth what you paid for it, is save your money and buy some disk drives, used if necessary. By the way, without a hard disk, UNIX doesn't make sense. My Godbout-owning friends swear by them. Rock solid, buy pricey. Best of luck Phil Burton Fortune Systems
SHahn@SUMEX-AIM.ARPA (12/13/83)
From: Sam Hahn <SHahn@SUMEX-AIM.ARPA> I'm not familiar with the memory, but beware that the Compupro boards I'm familiar with do not generate refresh for dynamic memory, which is why all Compupro memory boards are static. Also: How can anyone think of surviving long without a disk controller? Good luck to you, but I think you'll be springing for a DC before you realize now. SSM and Theta Labs look like they have nice I/O boards. Otherwise, I suggest you attending a computer swap, where CCS often sells boards as-is for around $5.00. (I've picked one up.) -- sam hahn [shahn@sumex] -------
Mkb@CMU-CS-C.ARPA (12/15/83)
From: Mike Blackwell <Mkb@CMU-CS-C.ARPA> Thanks to everybody who responded to my query on building a budget 68000 system! My faith in the net is restored... For those who asked, I'll summarize what I've found out: The Compupro 68000 board has sockets for two 2764's, but does not come with any ROM monitor (unless you buy their CPM, in which case you get some bootstrap ROM's). My first order of business will be to modify one of my 68000 monitors for this board (I've written a couple of monitors for our in house 68000 single board computers). The monitor contains the standard monitor features... examining and modifying memory, running with break points, etc, plus allows the downloading of MACSBUG format files over one of the serial ports. If I come up with something good, I'll make it available to net land. On the subject of Compupro's ROM sockets, whoever designed the board was exceedingly clever, and made it so that you can only read from the onboard ROM with PC relative memory reads. This makes it fairly difficult to read data constants from ROM. The trick seems to be to first run a hack routine that uses PC relative reads to copy all of your data down into RAM, where you can access it normally. Bleah. I think the reason for this misfeature is that they tried to save a few bytes of address space by mapping the onboard ROM into the same space as the IO page. Thus data reads and writes go to the IO page, and PC relative reads go to the ROM. What fun! The general consensus is that 64K bytes of memory is not enough to do anything serious. 256K seems to be the suggested minimum for a decent 68000 system. The general consensus also is that I will probably get sick of downloading files real quick, and I will break down and buy disk drives and an operating system sooner than I planned... We'll have to see on that one. It all depends on how my budget works out. For those of you who are interested, I do all of my 68000 development work on a Vax running Unix with Stanford's C cross-compiler and assembler. I've found that it works quite well (except for floating point). Anyway, in a month or so I should have everything together and hopefully running. Then I can fill you in on all the gory details! cheers, -m-
lee@fortune.UUCP (Ed Lee) (12/15/83)
I'm not familiar with the memory, but beware that the Compupro boards I'm
familiar with do not generate refresh for dynamic memory, which is why
all Compupro memory boards are static.
Why would you care about generating refresh signal from the CPU ( I assumed
this is what you meant, please correct me if I am wrong ) ?
Most dynamic memory boards (IEEE-696 compatible ) generate refresh internally.
I have no problem running dynamic memory boards on my S-100 system.
But you do have to be careful about DMA accesses when using company X's
CPU with company Y's memory board and company Z's disk controller. Very
often, the timings are so tight that only boards from the same company
work correctly.
Ed Lee
{amd70, ihnp4, harpo}!fortune!leeandree@uokvax.UUCP (12/19/83)
#R:sri-arpa:-1449000:uokvax:3400023:000:1140 uokvax!andree Dec 16 10:07:00 1983 The major problem with a `homebrew' system like that is that most people who sell nice os's for the 68K don't want to sell end-user (or `field- installable') versions of their os. (ATT is of, of course, an exception. You can buy Unix V for a measly $43K.) They want to sell to OEMs, and then let the OEMs sell to end-users. The OEMs of course expect you to buy their hardware. Some of them are nice enough to let you get off with a board set and disk drive. sheesh. The only system that you can (currently) get end-user is cp/m-68k. As previously pointed out, there's not much reason to bother. You would be better off running a low-end z80 system. Note that low-end z80's these days means 64K of memory & a 4Mhz z80. Such are available for $1200 or so. As a side note, there is an `attached processor' box for z80 systems. This box has a 6MHz 68000 and 128K of memory for $710. It plugs into your (or anybodies!) z80 system and does i/o parasitically through the CP/M-80 BIOS. They are talking about Unix V, and I am trying to finagle an os-9/68000 for the box. (os-9/68000: something better than Unix V, but not as good as 4.1). <mike
andree@uokvax.UUCP (12/20/83)
#R:sri-arpa:-1449000:uokvax:3400025:000:412 uokvax!andree Dec 18 15:59:00 1983 About the compupro CPU-68K and dynamic memory. On going through the schematics, I find an output labelled sMEMR, on pin 47 on the s-100 bus. I believe this is the memory refresh line you want. Also, I've been running a CPU-68K with 128K of Intersystems dynamic memory with no ill effects. However, I am running it at 4MHz until I get all the bugs chased out of the DC code, etc., and buy faster memory. <mike
eich@uiuccsb.UUCP (12/20/83)
#R:sri-arpa:-1449000:uiuccsb:4400034:000:277 uiuccsb!eich Dec 20 03:12:00 1983 sMEMR is the S-100 status line which indicates the initiation of a memory read cycle. It is not a dynamic RAM refresh signal; of course, the memory board could (and the ones I've seen do) generate their own refresh. Brendan Eich uiucdcs!uiuccsb!eich eich.uiuc@rand-relay
lee@fortune.UUCP (Ed Lee) (12/22/83)
This should belong to net.micro.s100 (if people are interested ).
On going through the schematics, I find an output labelled sMEMR,
on pin 47 on the s-100 bus. I believe this is the memory refresh
line you want.
sMEMR stands for s-100 bus MEMory Read. I think that the confusions
came from the pre-IEEE 696 s-100 bus pin 99 (refresh). Early s-100
computers use pin 99 to indicate that the bus masters (main CPU or DMA
devices) are not using the memory. Memory boards should gate this
signal with internal logics to refresh dynamic memory. However, the
refresh pin is not defined in the IEEE 696 standard. Newer memory
boards either look at other signals or insert wait states at appropriate
time. I haven't been reading the spec for a while, so please check
this with your local s-100 wizard.
c / Ed Lee
C. /_. {amd70, ihnp4, harpo}!fortune!lee