Recent searches
Search options
And today it became suddenly very relevant to me that the 65816 has a serious advantage in moving data over the 6502 at the same clock speed. By my math, a ~41% advantage.
; 6502:
lda #$34 ; [2]
sta $df00 ; [4]
lda #$12 ; [2]
sta $df01 ; [4]
= 12 cycles
; 65816
lda #$1234 ; [3]
sta $df00 ; [5]
= 8 cycles
; 65816, with DP = $DF00
lda #$1234 ; [3]
sta $00 ; [4]
= 7 cycles
On a C128 within some constraints it's possible to do it in 10 cycles in two different ways, thanks to MMU.
With zero page relocation (LDA #imm + STA zp) one must care to deal with offsets 0/1 (CPU ports) separately.
With stack relocation (LDA #imm+PHA) one must disable interrupts and set/restore stack pointer.
Sounds cumbersome, but I used both methods to speed up Amaurote gameplay in both VIC and VDC releases: