6502 Beginner Tips

From 8BitDev.org - Atari 7800 Development Wiki
Revision as of 16:11, 8 March 2022 by MSaarna (talk | contribs) (16-bit increment without full 16-bit addition)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

6502 Beginner Tips

This entry assumes you're new but generally familiar with 6502 assembly, and are just looking for some tips. If you need something more fundamental, check out the "Easy 6502" tutorial by Nick Morgan.


6502 if...then equivalents

If you're learning 6502 assembly with a background in a higher language, the following table presents typical logic comparisons in a more familiar format.

6502 if...then equivalents
test desired comparison branch
A = VALUE CMP #VALUE BEQ
A <> VALUE CMP #VALUE BNE
A > VALUE CMP #VALUE BEQ and then BCS
A >= VALUE CMP #VALUE BCS
A < VALUE CMP #VALUE BCC
A <= VALUE CMP #VALUE BEQ and then BCC
A = $ADDR CMP $ADDR BEQ
A <> $ADDR CMP $ADDR BNE
A > $ADDR CMP $ADDR BEQ and then BCS
A >= $ADDR CMP $ADDR BCS
A < $ADDR CMP $ADDR BCC
A <= $ADDR CMP $ADDR BEQ and then BCC

Similar tests can be made for X or Y instead of A, substituting CPX or CPY for CMP.


quickly testing bit 6 or 7 without affecting A, X, or Y

The BIT $MEM instruction will put bit 7 of MEM in the N flag, and bit 6 of MEM in the V flag. This doesn't use/invalidate the registers.


divide/multiply by powers of 2

On the 6502, the easiest way to multiply a value by 2 is to shift the bits of the value one place to the left. e.g. "ASL" or "ASL VALUE"

Similarly, you can divide by 2 by shifting the bits of the value one place to the right. e.g. "LSR" or "LSR VALUE"

Repeated shifts allow for operations with other powers of 2. e.g. 2 shifts for a divide/multiply by 4, 3 shifts for a divide/multiply by 8, etc.

Often you can design games to take advantage of powers of 2 divide/multiply, rather than using other values that are less easily manipulated by the 6502.


multiply by near powers of 2

In a pinch, you can multiply by 3 by performing a left shift (multiply by 2) followed by an addition of the original value. Perform another left shift for a multiply by 6.

Similarly you can multiply by 5 by performing two left shifts (multiply by 4) and an addition of the original value.

The alternative approach to above, is creating a look-up table for multiplication by the desired value. While this is much faster, it can use a lot of ROM. You'll need to figure out whether the trade-off is worthwhile in your application.


lookup tables instead of complex math

The 6502 isn't great at arbitrary multiplication, let alone more complex operations. The usual way around this is replacing the complicated operation using a look-up table in your code. Create the table using a program on a modern platform using a modern language, spreadsheet program, or calculator.

GetSquare
  ;called with value to square in A
  ;return value in A
  TAY
  LDA squaretable,y
  RTS
squaretable
  .byte 0, 1, 4, 9, 16, 25, 36, 49, 64, 81, 100, 121, 169, 196, 225


2's complement representation

The 6502 uses 2's Complement representation for negative numbers. In a nutshell, this means that very large numbers can also be treated as negative, depending on the operation.

For example, 5 + (-1) is 4. 5 + 255 is also 4, due to the byte overflowing. 255 is the 2's complement representation of -1.

To negate a number using 2's complement...

  EOR #255
  CLC
  ADC #1

To divide a signed 7-bit value by 2...

  CMP #$80
  ROR

To multiply a signed 7-bit number by 2...

  CLC
  LDA number
  ADC number
  ; caution - if the value in "number" is 64 or more, you'll overflow the sign bit

avoid double jsr

You should avoid double JSRs followed by RTSs, like the following...

MySubRoutine
  LDA PlayerX
  JSR MultiplyBy2
  RTS
MultiplyBy2
  ASL
  RTS

Instead you can save stack and cycle overhead by changing the first JSR into a JMP...

MySubRoutine
  LDA PlayerX
  JMP MultiplyBy2
MultiplyBy2
  ASL
  RTS


16-bit data tables

store 16-bit data (including addresses) in 2 separate tables with high and low values, rather than 1 table with 16-bit words. This allows you to increment over the values with a singe register increment/decrement, and increases the maximum table size to 256 entries.


16-bit increment without full 16-bit addition

Instead of a 16-bit addition by 1, you can quickly increment 16-bit words like this...

  INC VariableLo
  BNE skipHiInc
  INC VariableHi
skipHiInc

We check if the first INC causes the low byte to overflow and become zero again, and increment the high byte if that happens.