Why does this instruction add one to the accumulator if the carry flag is set?



I looked at the ADC page in Eyes & Litchy about this, and get the impression that the reason is to make multi-word addition easier. But this makes it harder to add things of the processor's native register size, because I have to remember to CLC beforehand if I don't want the result to be one greater. Is it really that common to do multi-word addition on these processors?

Because the word size is so small, yes, it's very, very common to handle numbers of larger than native size. Other architectures have taken other approaches to this over the years (providing separate add-without-carry instructions, or simply saving carry-out so that the programmer can handle carry-in if desired,) but the 6502 was designed with a pretty minimalist mindset, and if you have to choose just one approach to handle both word-sized and extended operations, this isn't a bad one.

I see. It doesn't seem quite as valuable on the 65816 when the accumulator is 16-bits.

_________________
x86?  We ain't got no x86.  We don't NEED no stinking x86!

_________________
http://WilsonMinesCo.com/ lots of 6502 resources
The "second front page" is http://wilsonminesco.com/links.html .
What's an additional VIA among friends, anyhow?

Yeah, multi-precision arithmetic is still important in 16 bits. 32 bits is about the smallest word size (well, in modern multiples-of-8 convention) where you can do many tasks of meaningful complexity without having to deal with larger integers at some point, and even then it's hardly uncommon.

Here are two ways to think of the question
- why does the 6502 have only ADC, and no ADD
- if only one, would it be better to have only ADD

For the first, it's because, as noted up-thread, it's a minimal machine, designed specifically to have a much reduced cost and therefore much reduced price. As much as 10x lower, which is to say, a 90% discount on the competition.

For the second, it turns out that having only ADD without ADC makes multi-word arithmetic quite cumbersome - it can be done, and it has been done, but it's easy to get it wrong.

And indeed, at 8 bits and even at 16 bits, multi-word values crop up often enough. Bearing in mind that the sort of embedded application envisaged for the 6502 would have multi-digit displays (weighing, measuring, point of sale) we see that many customers would have wanted both the multi-word arithmetic and the decimal mode.

The origin story of the 6502 is the idea that the 6800 particularly was too expensive, and most customers would have been happy to have a simpler microprocessor if it could be cheaper. As Motorola didn't agree, and would put it in writing as an abandoned idea, the 6502 team felt able to go off to build that simpler microprocessor. As it turned out, one junior person did bring some 6800 artwork with them, which made the subsequent lawsuit more sticky, and the effect was to have the 6501 withdrawn, and for MOS technology to need a lot of cash, which is where Commodore came in to buy them up.

_________________
x86?  We ain't got no x86.  We don't NEED no stinking x86!

It appears to me that ADC was designed this way on the 6502 to make multi-byte addition easier to do (since it's so common on an 8-bit machine) at the expense of making single-byte addition harder: you need to remember to CLC beforehand or otherwise convince yourself that the carry flag is unset in order to get the correct answer when working with just 8-bit addends.

But when your processor's sizeof(int) suddenly == 2 (when you upgrade to the 65816 and are porting software over to take advantage of the 16-bitness) there are often going to be fewer addition situations that are multi-word, so this design tradeoff doesn't buy you quite as much.

_________________
http://WilsonMinesCo.com/ lots of 6502 resources
The "second front page" is http://wilsonminesco.com/links.html .
What's an additional VIA among friends, anyhow?

_________________
x86?  We ain't got no x86.  We don't NEED no stinking x86!

I don't find it at all illogical to expect fewer multi-precision additions when the word length doubles. Just my $.02 worth...

-- Jeff

_________________
In 1988 my 65C02 got six new registers and 44 new full-speed instructions!
https://laughtonelectronics.com/Arcana/ ... mmary.html

Fewer, sure. But hardly none, especially when "doubled" is still only 16 bits.

_________________
x86?  We ain't got no x86.  We don't NEED no stinking x86!

I hope you're not discouraged, jeffythedragonslayer, by the strongly-expressed opinions upthread. It's not how we usually conduct ourselves here.