Add self modifying code to draw_line for 2% performance improvments

wip-hugo/routines/line/line_down.s

@@ -51,6 +51,36 @@
 
         jsr pixel_draw ;;only used first pixel. after this relative position is abused
         LDX X_pos
+
+        ;; Self modifying code. Makes LDA instructions take 1 cycle less.
+        ;; Code will run without this but slower!
+        ;; Modifies LDA instructions for dy_2 and SBC for V
+        ;;Note: The offsets like +2 etc is because there are instructions betwean the label and the
+        ;address that needs to be modified
+
+        ;;; dy_2
+        ;;; Modifies LDA >dy_2
+        ;LDA #$A9 ; LDA (immediate)
+        ;STA case_2
+        ;LDA >dy_2
+        ;STA case_2 +1
+        ;;; Modifies LDA <dy_2
+        ;LDA #$A9 ; LDA (immediate)
+        ;STA case_2 +6 ; ADC is +2 bytes, STA is + 2 bytes, Offset from before is +2 bytes.
+        ;LDA <dy_2
+        ;STA case_2 +7
+        ;;; V
+        ;;;Modidies SBC >V
+        ;LDA #$E9 ;SBC (immediate)
+        ;STA case_1 +2;LDA is +2
+        ;LDA >V
+        ;STA case_1 +3
+        ;;; Modifies SBC <V
+        ;LDA #$E9 ;SBC (immediate)
+        ;STA case_1 +8; Offset before +4 bytes, STA +2, LDA +2.
+        ;LDA <V
+        ;STA case_1 +9
+
 for_x:
         ;; Paints A to address in |btp_mem_pos* + Y|
         ;; Y is pixel position in the chunk. Therefor it may be that Y = 0, 1, 2, 3, 4, ,5 ,6 ,7.

wip-hugo/routines/line/line_test_time.s

@@ -25,7 +25,7 @@
         LDA X_pos_
         STA X_pos
 
-        jsr line_down
+        jsr line
         INC Y_end
         LDA Y_end
         CMP #$50
@@ -48,4 +48,4 @@ end__:
 
         jmp exit
 
-.include "line_down.s"
+.include "line.s"

wip-hugo/source.s

@@ -2,7 +2,7 @@
 .include "macros/timer.s"
 
 .include "STARTUP.s"
-.include "routines/line/line_test.s"
+.include "routines/line/line_test_time.s"
 
 ;;This is used by .s files to terminate nicely without reading includes at the end!
 exit: