Using XOR to invert only 1 bit?

[Victim_RLSH]

Here is code that reverses a ball velocity after it bounces off of something on its left or right side. I flip the sign bit of a 7-bit signed integer stored in one byte.

TestLeft LDA CollisionFlags
AND #FromLeft$
BNE TestRight

LDA BallXVel
AND #%01111111 ; remove sign bit to reverse X direction
STA BallXVel

TestRight LDA CollisionFlags
AND #FromRight$
BNE CollisionUpdate

LDA BallXVel
ORA #%10000000 ; add sign bit to reverse X direction
STA BallXVel

Would it be possible to use XOR to just invert the sign bit and not waste memory on both routines?

would XOR #%10000000 flip only the sign bit and preserve the absolute value?

[chysn]

Code: Select all

eor #%10000000

will, in fact, do what you say, flipping bit 7.

But keep in mind that’s different than changing the sign.

%10000001 is not -1, %11111111 is. If you simply flip bit 7 of 1, you get -127.

That may or may not matter in your application. If you think it might matter, we can delve into signed binary numbers and twos complement arithmetic.

[Mike]

Hi, Victim_RLSH,

Of course you can do sign-magnitude arithmetic on the 65xx, but that's not how the ADC and SBC instructions are supposed to be used. 'Emulating' sign-magnitude will easily lead to ten times more code and one tenth of the speed than what you'd get by employing 2s-complement, because you have to do all kinds of tests and case selections depending on the signs of the arguments and operator (add or subtract).

I had translated the "BALLPING" program in the User's Guide some time ago for my ML/VICMON primer, you find it here:

viewtopic.php?t=9197&start=1

In the original BASIC program, the velocity inversion at the borders happens in lines 120 and 140, with DX=-DX and DY=-DY, respectively. This is translated into machine code as follows:

Code: Select all

SEC
LDA #$00
SBC $xx   ; $xx=$F9 for DX, =$FA for DY
STA $xx

If the value is already in the accumulator, then it is a little bit shorter (3 or 2 instructions):

Code: Select all

SEC ; can be omitted if it is known that C=1
EOR #$FF
ADC #$00

*or*

Code: Select all

CLC ; can be omitted if it is known that C=0
EOR #$FF
ADC #$01

Note the equivalents of X=X+DX and Y=Y+DY simply use the ADC instruction with no case selections necessary (that would depend on the sign of DX or DY).

Greetings,

Michael

[Victim_RLSH]

Yeah, I did it the long way in this case, so the XOR should work. I just test for the sign bit and have separate increment and decrement routines depending on the velocity. No wonder I'm running out of space on such a simple game. My pathetic 6502 skills definitely need some help, which is why I'm doing this to begin with.

[tlr]

Yeah, I did it the long way in this case, so the XOR should work. I just test for the sign bit and have separate increment and decrement routines depending on the velocity. No wonder I'm running out of space on such a simple game. My pathetic 6502 skills definitely need some help, which is why I'm doing this to begin with.

Practise, practise. I can recommend disassembling other peoples code figure out how stuff is done. Coding assembly is a fun puzzle game!

[srowe]

Practise, practise. I can recommend disassembling other peoples code figure out how stuff is done. Coding assembly is a fun puzzle game!

+1, you will learn some neat tricks which can save you space and/or clock cycles.