James Knott wrote:
Hans Witvliet wrote:
On Fri, 2007-12-07 at 12:59 -0500, James Knott wrote:
The Data General Eclipse line had a feature called "Writable Control Store", which could be used to add custom instructions to the CPU. The VAX 11/780 had it's microcode loaded from floppy at boot, but I don't recall if it was changeable in the same manner as the Eclipse WCS.
Floppy? It was a real huge 8" flop ... (still have them here)
hw
Yep. There was an LSI-11 (microprocessor version of PDP-11) hidden in the cabinet, equipped with one or two 8", hard sectored drives. It was also used to connect the console terminal. As I recall, the command to use it as the VAX console was "STP" and Ctl-Z(?) to return to the LSI-11 console. Back in it's day, the VAX was considered a "super mini", a real hot system. But it only had the CPU power of a 386!'
The old VAX had a beautiful assembly language with 13 different addressing modes (4 numerically different addressing modes were all used for a 6-bit "immediate data" addressing mode)...which, when paired with the brilliance of making the program counter and stack pointer general registers (and thus specifiable as the register to reference in all of those addressing modes) had a grand total of something like 20 effective addressing modes. GREAT design if you were writing assembly code. Unfortunately, all of those crazy addressing modes (like doubly-indirect or something like that) made for a real mess whenever a page fault occurs. Tthe CPU has to unwind the whole partially- complete instruction, and then do an interrupt to the VM's swapping routine, return from the swapper routine, and then restart the instruction... And if the instruction was something like insert a record into a doubly-linked list, or performing some sort of string manipulation (some of which are more complicated than what's in C's strings library), well, then... yeah....it's amazing the VAX guys ever got the design to function right. Initially, i thought RISC was a crazy idea, compared to the ultra-CISC VAX-11 instruction set...but then when I read about the issues with page faults... I was almost immediately convinced that yes indeed, the RISC guys did have a very, very important point, and that the Clipper chip and others weren't so crazy after all.
BTW, that's where I first came across the "Adventure" game. :-)
-- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org