gjk@talcott.UUCP (Greg Kuperberg) (01/20/85)
Why doesn't the 68000 have increment or decrement instructions? I know that the 68000 was designed (at least to some degree) with the spirit of RISC, but INC and DEC are actually quite simple, and useful, instructions. --- Greg Kuperberg harvard!talcott!gjk "Nice boy, but about as sharp as a sack of wet mice." - Foghorn Leghorn
chris@umcp-cs.UUCP (Chris Torek) (01/20/85)
With ``addq'' there's no need for an increment: just addq #1,d0 or
whatever. In fact, you have a whole bunch of increments, #1 to #8,
if I recall correctly....
--
(This line accidently left nonblank.)
In-Real-Life: Chris Torek, Univ of MD Comp Sci Dept (+1 301 454 7690)
UUCP: {seismo,allegra,brl-bmd}!umcp-cs!chris
CSNet: chris@umcp-cs ARPA: chris@marylandguy@rlgvax.UUCP (Guy Harris) (01/20/85)
> Why doesn't the 68000 have increment or decrement instructions? An INC or DEC instruction would take one word to specify the opcode, size, and address mode/register, and an extra word for any offset, if any. This is exactly as many words as ADDQ #1,<operand> or SUBQ #1,<operand> would take - so it has a more general form of INC and DEC (you can increment or decrement by any value from 1 to 8 (ADDQ #0 is useless, so the instruction adds 1+<3 bits from instruction> - easy to do by just jamming the input carry bit high in the adder). > I know that the 68000 was designed (at least to some degree) with the > spirit of RISC, Say what? Yes, its instruction set is smaller than the VAXes, but then so is the instruction set of a PDP-11 or IBM 370 (excluding microcode assists, etc.). Guy Harris {seismo,ihnp4,allegra}!rlgvax!guy
gjk@talcott.UUCP (Greg Kuperberg) (01/21/85)
> Why doesn't the 68000 have increment or decrement instructions? ... > Greg Kuperberg And the answer is (thanx for all the replies) that you can use ADDQ and SUBQ, since those instructions only take up one word. Nevertheless, I wince when I see "ADDQ #1,D0", even if it does have a Q in it. I claim this is a natural tendency, since this kind of thing really is inefficient on other CPU's. I advise making an INC and DEC macro if you also have this tendency. --- Greg Kuperberg harvard!talcott!gjk "Nice boy, but about as sharp as a sack of wet mice." - Foghorn Leghorn
breuel@harvard.ARPA (Thomas M. Breuel) (01/21/85)
> wince when I see "ADDQ #1,D0", even if it does have a Q in it. I claim > this is a natural tendency, since this kind of thing really is inefficient > on other CPU's. I advise making an INC and DEC macro if you also have this > tendency. "ADDQ" makes sense on a 68000, since you have to increment not only by 1 (for pointers to bytes), but also by 2 and 4 (for pointers to words and longwords respectively). "INC" and "DEC" are anachronisms from processors like the Z80. I would actually recommend you to leave out the "Q" in the instruction. MOTOROLA assemblers and compatibles automatically use the shortest instruction possible, and assembly language listings look a lot nicer if they are not cluttered with instruction modifiers. Thomas.
mwm@ucbtopaz.CC.Berkeley.ARPA (01/22/85)
In article <246@talcott.UUCP> gjk@talcott.UUCP (Greg Kuperberg) writes: >> Why doesn't the 68000 have increment or decrement instructions? > >And the answer is (thanx for all the replies) that you can use ADDQ and >SUBQ, since those instructions only take up one word. Nevertheless, I >wince when I see "ADDQ #1,D0", even if it does have a Q in it. I claim >this is a natural tendency, since this kind of thing really is inefficient >on other CPU's. I advise making an INC and DEC macro if you also have this >tendency. > Greg Kuperberg > harvard!talcott!gjk [Excuse me while I retch.] Rather than having an INC & DEC macro, how about getting a smart assembler that know that "ADD #n,D0" can be turned into an ADDQ instruction for small n? That way, you get some measure of orthogonality in the source. If you have a natural preference for such, I advise you to do so. <mike
darryl@ISM780.UUCP (01/22/85)
Although there are no instructions with mnemonics 'inc' or 'dec', there are the addq and subq (add or sutract quick) instructions. They allow adding or subtracting values in the range [1,8], which is just the thing when advancing a pointer through consecutive array elements. They both occupy 1 (16 bit) word (exclusive of address displacements), so they are just as compact as they can be. --Darryl Richman, INTERACTIVE Systems Inc. ...!cca!ima!ism780!darryl The views expressed above are my opinions only.
rpw3@redwood.UUCP (Rob Warnock) (01/22/85)
+--------------- | Why doesn't the 68000 have increment or decrement instructions? I know | that the 68000 was designed (at least to some degree) with the spirit of | RISC, but INC and DEC are actually quite simple, and useful, instructions. | Greg Kuperberg | harvard!talcott!gjk +--------------- It DOES! They're just not called that. Try looking at ADDQ and SUBQ (add and subtract "quick"). These are opcodes that can increment/decrement from one to eight. The constant (1-8) is buried in the op code. Much more useful than just INC and DEC, especially for stepping address pointers to bigger things than bytes. They work with any destination: data reg, address reg, memory; with byte, word, or long (except no byte adds to address regs). Rob Warnock Systems Architecture Consultant UUCP: {ihnp4,ucbvax!dual}!fortune!redwood!rpw3 DDD: (415)572-2607 USPS: 510 Trinidad Lane, Foster City, CA 94404
dan@digi-g.UUCP (Dan Messinger) (01/23/85)
In article <242@talcott.UUCP> gjk@talcott.UUCP (Greg Kuperberg) writes: >Why doesn't the 68000 have increment or decrement instructions? I know >that the 68000 was designed (at least to some degree) with the spirit of >RISC, but INC and DEC are actually quite simple, and useful, instructions. Have you noticed the ADDQ and SUBQ instructions? These are single word instructions that do a quick add/subtract of a 1 to 8 constant. The constant is kept in a 3 bit field that most other instructions use to specify a register. A special INC/DEC by 1 instruction could use no less code space, and probably execute no faster. So (in my opinion), the 68000 has instructions superior to INC/DEC. Dan Messinger Digigraphic Systems Corp. ihnp4!umn-cs!digi-g!dan