7 May
2010
7 May
'10
12:39 a.m.
Am 07.05.2010 08:08, schrieb Fred Cisin:
A common argument is the word size of the primary calculation store,
usually called the accumulator. I don't say "register" because CPUs
like the TMS9900 do not have an on-chip arithmetic/logic register set.
The 8086/8088 have, besides other regs, the AX accumulator which distinguishes
it from, for instance, the 8 bitters like 6809 (A and B 8bit-register, combinable
for certain ops to a D register), or the 8080 with A, besides some HL register
pair suitable for a few unsymmetric operations. An 68008, even with an 8 bit data
bus, would then qualify as a 32 bit CPU, and a PDP-8 as 16 bit, despite of what
Also frequently seen are address bus sizes beyond 16 bit, at least somehow intrinsically
foreseen in the instruction set (APRs, segment registers); but this is a secondary
category (what would be a T-11 then?).
8080 is an
8-bit processor and the 8086/8088 are 16-bit processors. But
you CAN, if you try hard enough, define them differently.
Aside from disparaging others' definitions, does anybody have a decent
definition?
Are the registers 16 bit registers that can be split into two 8 bit?
or are they pairs of 8 bit registers that can be used together for 16 bit
values? Is the size of the data bus irrelevant?
Yes, as it just determines overall system performance, cf the 68000/68008/68010/68020
family.
The software of an 8088 looks like 16 bit; the
hardware of an 8088 looks
an awful lot like 8 bit. There are people who consider the 8088 to be an
8-bit and consider the 8086 to be 15 bit, in spite of their
"similarities".
It depends on whether you look at it as a hardware or software engineer.
Usually though, the cripple versions of a 16 bit CPU cause much more
glue logic on the board than a straight-forward 16bit bus CPUs with non-
multiplexing of signals.
What do you consider 80386 to be?
how about the 80386-SX? It's hardware seems similar to 80286; what is
THAT?
Do they run the software of their corresponding baseline family member?
If a 386 would qualify as 16 bit only because it still understands the 286
segmenting and security model, then it could likewise qualify as a 4 bit processor
because a rather simple cross compiler would be sufficient to run it 4040 code.
And leave alone the V20, CP/M-86 ran cross compiled 8080 code.
This is becoming absurd though I already heard that.
What the hell is a "Celeron"? or a
"Dragonball"? or an "Atom"?
Usually people are able to distinguish a truck from a limousine. But what are
pickups and SUVs then?
--
Holger