shaulp@pnet02.cts.com (Shaul Peleg) (05/25/89)
Since there is no 88000 SIG here:
How is the 88000 instruction set? Can anyone give me any specs on it?
And finally -->is it anythingl ike the 68000 in terms of programming -- can
it be done in assembler even though it is a RISC design?
-=SP=- (thanks!)
UUCP: {ames!elroy, <backbone>}!gryphon!pnet02!shaulp
INET: shaulp@pnet02.cts.compmb@swituc.UUCP (Pat Berry) (05/25/89)
In response to the request for 88K specs: the MC88100 has : single clock integer, bit field, branch and store operations. fifty one instructions and seven operand types. four fully independent execution units with 5 concurrent pipes. execution synched in hardware by a scoreboard register. 32 general purpose registers. single and double precision IEEE 754 Floating Point compatability. full 32-bit combinatorial multiplier. separate data and instruction memory ports (Harvard Bus). 30 bit data address bus (byte strobes select individual bytes). 32 bit data bus (32 bit word). 30 bit instruction address bus. 32 bit instruction bus (fixed 32 bit instruction length). Yes, it is definitely programmable in Assembler! (see line 2 above) If you would like further info, Email to me and I will summarize for the group. Pat Berry uunet!swituc!pmb
mbellon@mcdurb.Urbana.Gould.COM (05/26/89)
Since there is no 88000 SIG here:
How is the 88000 instruction set? Can anyone give me any specs on it?
And finally -->is it anythingl ike the 68000 in terms of programming -- can
it be done in assembler even though it is a RISC design?
-=SP=- (thanks!)
UUCP: {ames!elroy, <backbone>}!gryphon!pnet02!shaulp
INET: shaulp@pnet02.cts.com
The instruction set in very different from the 68k. Yes, you can handle
assembly code for it. Its not that bad. The delayed branchs are a little tough
to use effiently and optimal code sequences are harder to write than on the
68k.
Kernel handling of page faults, traps, and such it TOTALLY different (RISCY,
e.g. you do everthing yourself no nifty kernel mode instruction).
There are only a few address modes, and the FPU is built in
(no trancendentals). It is capable of high performance multiprocessing
due to the hardware architecture (the 68k is not very good here). The
CMMU chips make this wonderfull.
The coprocessor interface is missing. Instead you have functional units. These
are significantly better than coprocessors as they can be involved in the
pipeline of the processor and use some of it data without duplicating half of
the processor chip in each functional unit.
This is a quick summary. I can talk more as you see fit.
Mark Bellon
mbellon@xenurus.gould.com (name not changed yet)
Software Project Scientist
Motorola Urbana Design Centermichael@mcdchg.chi.il.us (Michael Bodine) (05/27/89)
You know, while there are certainly some good discussions that could go on in this newsgroup, or one dedicated to the 88k, it hardly seems worth it to describe the basics of the processor by spawning articles on the net when you can get the data sheet and reference manuals by phoning your nearest Motorola Semiconductor sales office and requesting a copy. Full details of the programming model and instruction set are present in the ref manuals for the 88100 (cpu) and 88200 (cmmu); for those who want something a little lighter, seperate data sheets are available for each part. If for some reason you can't find a nearby Semiconductor office, email your street address to me and i'll tell you, off the net, where the nearest one is.