mo@messy.bellcore.com (Michael O'Dell) (10/13/90)
I think the Burroughs B1700 was multiprogramming at the microcode level before the Alto. For those not familiar with it, the B1700 was, in some sense, the ultimate CISC. Main memory was BIT addressed (actually the HOLES between the bits were address because the field at address "85-up" was disjoint with the field at address "85-down"), and each language processor on the machine generated code for an idealized machine (complete with Huffman coding for incredibly dense code) and when you context switched between user programs, the instruction interpretation changed too (sort of like in Unix where the kernel stack segment switches on a context switch). There was a microkernel (in microcode) which implemented the basic part of processes, including the S-level (traditional machine code) interpreter. All the system utilities (OS, compilers, tools, etc) were coded in a language called SDL, rather reminiscent of XPL in some ways, but with interesting extensions. So, when people start talking about bit addressing and super-dense memory images using Huffman codes and such, just remember that it's all be done before.... Oh yes, the internal bus of the B1700 was 24-bits wide, but there were some wonderful things in the S-memory address registers to make iterating over chunks of larger operands much easier to program in the microcode (which was written in a language called MIL, if memory serves me right.) I think the people at SUNY Buffalo, circa 1974, lead the charge in using the machine in CS teaching and research. -Mike
henry@zoo.toronto.edu (Henry Spencer) (10/14/90)
In article <27843@bellcore.bellcore.com> mo@messy.UUCP (Michael O'Dell) writes: >I think the Burroughs B1700 was multiprogramming at the microcode >level before the Alto. ... True, although with a somewhat different flavor. The B1700 didn't have the Alto's multiple microcontexts, which made microprogram context switching effectively instantaneous and permitted using microcode to supply most of the smarts for most of the i/o devices. (For example, the original disk interface had a one-word buffer and no DMA, and the video generation hardware was at a similar level of non-complexity, with carefully polished microcode doing all the work.) The Dorado, and perhaps some of the other Xerox D-machines, copied this approach, but nobody else has that I'm aware of. -- "...the i860 is a wonderful source | Henry Spencer at U of Toronto Zoology of thesis topics." --Preston Briggs | henry@zoo.toronto.edu utzoo!henry
colwell@omews35.intel.com (Robert Colwell) (10/19/90)
In article <1990Oct14.001905.19442@zoo.toronto.edu> henry@zoo.toronto.edu (Henry Spencer) writes: >The B1700 didn't have >the Alto's multiple microcontexts, which made microprogram context >switching effectively instantaneous and permitted using microcode >to supply most of the smarts for most of the i/o devices. >The Dorado, and perhaps >some of the other Xerox D-machines, copied this approach, but nobody else >has that I'm aware of. Alas, the world has already forgotten the Three Rivers Computer Perq. Ok, so the arch. was borrowed wholesale from the Alto. But it did all the stuff you're talking about above. And if the durned micros hadn't gotten so good so quickly this type of arch. approach might have actually made some sense. Hindsight is so easy compared to prediction. Bob Colwell mipon2!colwell@intel.com 503-696-4550 Intel Corp. JF1-19 5200 NE Elam Young Parkway Hillsboro, Oregon 97124
ok@goanna.cs.rmit.oz.au (Richard A. O'Keefe) (10/19/90)
In article <6120@omepd.UUCP>, colwell@omews35.intel.com (Robert Colwell) writes: > Alas, the world has already forgotten the Three Rivers Computer Perq. How fortunate the world is. I still have vivid, not to say lurid, memories of the Perq. Selectric keyboard, own Pascal dialect unlike anything else, documentation (heh heh), floating point microcode that thought at one time that 10.0*10.0 = 98.9. Then there was the UNIX (well, sort of) version. Give me an Orion any day. -- Fear most of all to be in error. -- Kierkegaard, quoting Socrates.
firth@sei.cmu.edu (Robert Firth) (10/19/90)
In article <6120@omepd.UUCP> colwell@mipon2.intel.com (Robert Colwell) writes: >Alas, the world has already forgotten the Three Rivers Computer Perq. This seems to be a use of the word 'alas' with which I am unfamiliar. Unless you are saying, in a highly elliptical way, that those who forget history are compelled to repeat it, in which case I agree with you.
henry@zoo.toronto.edu (Henry Spencer) (10/19/90)
In article <6120@omepd.UUCP> colwell@mipon2.intel.com (Robert Colwell) writes: >>the Alto's multiple microcontexts... >>The Dorado, and perhaps >>some of the other Xerox D-machines, copied this approach, but nobody else >>has that I'm aware of. > >Alas, the world has already forgotten the Three Rivers Computer Perq... I assure you I haven't, since I wrote much of a compiler for it for my MSc thesis... I don't recall the Perq having multiple microcontexts, actually, although I may have just forgotten the details. It's been a long time, and dealing with that architecture was not fun... -- The type syntax for C is essentially | Henry Spencer at U of Toronto Zoology unparsable. --Rob Pike | henry@zoo.toronto.edu utzoo!henry
richard@aiai.ed.ac.uk (Richard Tobin) (10/23/90)
In article <4015@goanna.cs.rmit.oz.au> ok@goanna.cs.rmit.oz.au (Richard A. O'Keefe) writes: >> Alas, the world has already forgotten the Three Rivers Computer Perq. >How fortunate the world is. I still have vivid, not to say lurid, memories >of the Perq. Yeah, but at least it came with a 7th edition manual. I still find it handy from time to time. -- Richard -- Richard Tobin, JANET: R.Tobin@uk.ac.ed AI Applications Institute, ARPA: R.Tobin%uk.ac.ed@nsfnet-relay.ac.uk Edinburgh University. UUCP: ...!ukc!ed.ac.uk!R.Tobin