rogerj@batcomputer.tn.cornell.edu (Roger Jagoda) (04/13/90)
In article <83smithw@physc1.byu.edu> smithw@physc1.byu.edu writes: >a number of things will be happening with the cube. 2.0 software Gee, do you think they'll get the NON-preliminary (i.e. final printing) 1.0 manuals out before the 2.0 "preliminary manuals" ship? >will be released, cpu upgrade to 040 board, video will be off loaded >to another slot-supposed to speed things up considerable like! Video Great! A frame-buffer/accelerator is just what this little baby needed! Now, tis means they have to have finished the NBIC chip. But, uh, if this board is just going in another nu-bus slot, wouldn't THAT then become a bottleneck as signals would have to be passsed from processor board to Montior imaging circuitry (DPS on a board?). Are we just trading one battle-neck for another...or do they REALLY want to fiddle with the bus too? >slot board will support 32 bit color, color monitor will be offered >at some discount to b&w people, but you don't need to buy color to take >advantage of video board. cpu upgrade plus video cost is under 1K maybe >under .5K(not color). A multiprocessor board is in the works. Motorola is Uhm, I confused again. A "multi-processor board" means that the current board can't do that. But I thought the MACH et al were just waiting for that NBIC chip to be finished. The software's already there...what else is missing? I REALLY thought the reason for the Cube to have those other slots is so my machine could be some sort of awesome 4-cpu screamer. Am I going to have to replace the current hardware (i.e. total cpu board swap-out?)...I don't think I like that! >shipping 040s this month with real volume by mid May-June. Some of this >stuff may be announced in the fall, for possible Dec. delievery. Apparently >some kind of lower cost version of the present cube may be announced. >Some kind of hierarchy will develop, low cost cube (whatever that is), >standard version with 512MB optical (hard drive optional still) color optional, >and multiprocessor cube to be hyped as a server for NeXT networks. (But >could be purchased as standalone for the greedy). Anyway, >that's the very unofficial story. So much for the rumors. Yes, but it WAS fun!!! --Roger Jagoda --Cornell University --FQOJ@CORNELLA.CIT.CORNELL.EDU -- ------------------------------------------------------------------------------ Roger Jagoda -- My employers don't even like paying Cornell University me, let alone accept responsibility fqoj@cornella.cit.cornell.edu for anything I say or do! --
rca@tahoe.cs.brown.edu (Ronald C.F. Antony) (04/14/90)
>Uhm, I confused again. A "multi-processor board" means that the current >board can't do that. But I thought the MACH et al were just waiting for >that NBIC chip to be finished. The software's already there...what else >is missing? I REALLY thought the reason for the Cube to have those other >slots is so my machine could be some sort of awesome 4-cpu screamer. >Am I going to have to replace the current hardware (i.e. total cpu >board swap-out?)...I don't think I like that! Well, the board they have now can do it. I was told by NeXT ppl that it was tried out. But do you really want to have 4 times the video chips, the SCSI bus, the ethernet etc? I guess what this will be is a board with a couple of CPU's and lots of memory and nothing else. The only thing it can do is speed things up, but it will not replace the current or future main boards. BTW: the bus chip is working... Ronald ------------------------------------------------------------------------------ "The reasonable man adapts himself to the world; the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man." Bernhard Shaw | rca@cs.brown.edu or antony@browncog.bitnet ------------------------------------------------------------------------------
mark@Jhereg.Minnetech.MN.ORG (Mark H. Colburn) (04/14/90)
In article <10103@batcomputer.tn.cornell.edu> rogerj@tcgould.tn.cornell.edu (Roger Jagoda) writes: >Great! A frame-buffer/accelerator is just what this little baby needed! >Now, tis means they have to have finished the NBIC chip. But, uh, if this >board is just going in another nu-bus slot, wouldn't THAT then become >a bottleneck as signals would have to be passsed from processor board >to Montior imaging circuitry (DPS on a board?). Are we just trading >one battle-neck for another...or do they REALLY want to fiddle with the >bus too? Currently, the machine has a bus which is not being used. It is located on the inside front of the cube. I assume that this is the bus that was designed for inter-board communications, since it is the only way that cards could be interconnected short of using ribbon cables between them. Bus transfer speeds can be quite quick these days, especially if they have the hardware to offload the bus handling. So, while adding a video board would mean that the video data must be passed over the bus, it would not necessarily be slower than it is now. Much of it will depend on the implementation. Will the bus speed and architecture be a bottleneck? Possibly, but if properly designed, and the machine is being used for interactive purposes, probably not. Now if you were to start moving lots of data over the bus... >Uhm, I confused again. A "multi-processor board" means that the current >board can't do that. But I thought the MACH et al were just waiting for >that NBIC chip to be finished. The software's already there...what else >is missing? I REALLY thought the reason for the Cube to have those other >slots is so my machine could be some sort of awesome 4-cpu screamer. >Am I going to have to replace the current hardware (i.e. total cpu >board swap-out?)...I don't think I like that! If you take a look on the main motherboard, you will see that there is a pad open for another chip. Rumor has it that this is for the bus controller chip to allow for multiple cards in the machine to talk to each other. Currently, there is no way for boards to talk to each other, so while the operating system supports it, there is no way to make it a reality. Theoretically, once the bus chip is put on the cards, and the software upgraded a bit, you should be able to put 2 or 3 more cpus in the cube and really make the pooch scream. In answer to the last question, you will probably have to swap out the board, but you may be able to get away with just putting a chip in the pad. Ok, now time for some speculation: Given that the boards can't talk to each other, the bus in the cube can be disregarded for the time being. And, given that the entire machine is housed on a single motherboard, complete with video, disk, keyboard, etc. It should be possible to put 4 cpu cards in the cube at one time. If you ran an thinnet cable to hook them all up, you would have a somewhat loosely coupled 4 processor system. Each board would be it's own machine, and have it's own identity, and would probably have it's own drive to boot off of, but it could be kinda neat. -- Mark H. Colburn If you don't make money off of it, Open Systems Architects, Inc. it had better be either a religious mark@Minnetech.MN.ORG experience or a hobby. - Lance Cooper