Frank_Hill@ankh.UUCP (Frank Hill) (07/03/88)
I heartily agree that the MCA buss should *NOT* be endorsed. It is, in
my opinion, another attempt by the boys in BLUE to grab the industry and
hold on tight. The TI NuBus, already adopted by Apple, would provide an
ideal buss architecture and allow the manufacture of boards with an
incredibly large potential market base....do it, please, but HURRY
before my company has to bend to corporate pressure and by MCA!
--
FidoNet : 369/6 the Eye of Osiris - 305-973-1947 - OPUS/UFGATE
UUCP : ...!{gatech!uflorida!novavax, hoptoad, umbio, attmail}!ankh alvitar@madhat.UUCP (Phil Harbison) (07/06/88)
In article <26.22CD76D1@ankh.UUCP>, Frank_Hill@ankh.UUCP (Frank Hill) writes: > I heartily agree that the MCA buss should *NOT* be endorsed. ... The > TI NuBus ... would provide an ideal buss architecture ... I fully agree with Frank. I've designed cards for several busses (VME, PC/AT, Q-Bus, MultiBus) and thoroughly researched most of the others (except FutureBus). NuBus is clean, simple, and processor independent. It supports automatic configuration with features like geographic addressing and identification PROMs. NuBus uses distributed arbitration and directed interrupts, so there are no daisy-chain signals. The only centralized services are the reset and clock logic, and there are ways to distribute even those resources. NuBus uses a multiplexed, 32-bit address and data bus, with maximum bandwidth of 37.5 MegaBytes/second. The data bus is unjustified, so there is no need for byte swapping logic except on the CPU card, where it belongs. Nubus also provides a "try-again-later" response to a memory cycle, making it easy to build adapters to other busses, or implement cache coherency schemes. The new form factor used in the MacII (4.0" by 12.8"), is slightly smaller than the size of the AT-bus cards (4.8" by 13.12"). Systems built around this form factor could probably use the rash of existing cheap PC enclosures. If the 386 PCs use NuBus, then third party hardware vendors can build a single product for both markets. NuBus uses the 96-pin DIN-41612 connectors, which are more reliable than the edge connectors used on the AT and the MCA. Pin density is 1.5 times that of the connector used in the AT, and about the same as those used in the MCA. The connector is located closer to the center of the board (about 3 inches from the end), which should simplify printed circuit board layout. Routing is complicated when all the connectors are on one end of the board, as in the AT. By now you've probably figured out which bus is my favorite. :-)
aptr@ur-tut (The Wumpus) (07/07/88)
In article <26.22CD76D1@ankh.UUCP> Frank_Hill@ankh.UUCP (Frank Hill) writes: > > The TI NuBus, already adopted by Apple, would provide an >ideal buss architecture and allow the manufacture of boards with an >incredibly large potential market base.... I can see it now. IBM brings out a 80386 based machine using the NuBus, a mouse, keyboard, two 3.5 inch drives, and a high resolution color monitor. What does Apple do? It sues IBM over the look-and-feel of the Mac II claiming that it was the first to introduce a PC with a mouse, a keyboard, 2 3.5" drives, a color monitor, and the NuBus, and that it holds the copyrights on all machines with similar equipment :-) Actually, I understand the reason that IBM moved to a new bus system, but I still don't know why they had to create yet another standard. There are many good bus systems that work for high speed computers with multiprocessing capability. Among these are such systems as the NuBus, the VME Bus, and the MultiBus. Another question that deserves consideration is why did IBM ever come up with the orginal bus system in the PCs and the one in the AT. At the time of the PC's creation, there was already a very widely accepted bus standard, the S-100 bus. Even today, S-100 bus systems can still be found. When IBM introduced the AT bus, they completely ignored the 16 bit bus developed by Olivetti. The Olivetti bus system for its M24 (AT&T 6300, Xerox 6064) kept the standard PC style 8-bit bus, but extended it to a full 16 bits. Olivetti also was the first to tackle the problem of 16 bit writes to an 8bit bus. They solved the problem by sending data out in high order-low order format like it is stored in code, IBM chose the other method. So, am I really suprised that when it came time to move up to a better bus system that IBM chose to once again to create their own system? No. -- The Wumpus UUCP: {cmcl2!decvax}!rochester!ur-tut!aptr BITNET: aptrccss@uorvm Internet: aptr@tut.cc.rochester.edu Disclaimer: "Who? When? Me? It was the Booze!" - M. Binkley