From ee69828ec7653821b16f85f56eddcf5c371a94db Mon Sep 17 00:00:00 2001
From: hugova <hugova@kth.se>
Date: Sun, 2 Mar 2025 18:05:20 +0100
Subject: [PATCH] Optimisation by moving logic out of loop for faster
 execution. This should be implemented on the other line_* files

---
 wip-hugo/routines/memory/line_down.s | 50 +++++++++++++++-------------
 1 file changed, 26 insertions(+), 24 deletions(-)

diff --git a/wip-hugo/routines/memory/line_down.s b/wip-hugo/routines/memory/line_down.s
index 0c8a4fa..698b60b 100644
--- a/wip-hugo/routines/memory/line_down.s
+++ b/wip-hugo/routines/memory/line_down.s
@@ -12,42 +12,50 @@
     Y_end = $05
     X_pos = $FC
     Y_pos = $FB
+    dx = $0c
     dy_2 = $0607
-    dx = $08
+    V = $0809 ;Lets reuse dx register ($08)
     D = $0a0b
 
-    ;;Set values
-    LDA #$00
-    STA $FD ; for pixel_draw
-
     ;example values ~~~~~ SHOULD BE PRECOMPILED
     LDA #$00
     STA X_pos
     STA Y_pos
 
-    LDA #$f0
+    LDA #$90
     STA X_end
-    LDA #$80
+    LDA #$10
     STA Y_end
     ;;~~~~~~~~~~
 
-    ;;2*dy = 2*(Y_end -Y_pos)
+
+    ;;We need to clear this memory
+    LDA #$00
+    STA <V
+    STA <dy_2
+    STA $FD ; for pixel_draw
+
+    ;; V = 2*(dx -dy)
+    ;; where: dy = Y_end - Y_start, dx = X_end - X_start
     LDA Y_end
     SEC
     SBC Y_pos
-    STA >dy_2
-    LDA #$00
-    STA <dy_2
-    mult_16 >dy_2, <dy_2
-
-    ;;dx = (X_end -X_pos)
+    STA >V
+    STA >dy_2;  >dy_2 = dy. Needed for dy_2 (not for V)
     LDA X_end
     SEC
     SBC X_pos
     STA dx
+    SEC
+    SBC >V
+    STA >V; <V = dx - dy
+    mult_16 >V, <V; V = 2*(dx -dy)
 
-    ;;D = 2*dy - dx + 2*255. 0 <=D <= 4*255
-    ;;Our D is bigger then wikipedia to not use negative numbers
+    ;dy_2 = dy*2
+    mult_16 >dy_2, <dy_2
+
+    ;;D = 2*dy - dx + 2*255
+    ;;Our D is bigger then wikipedia because D is unsigned.
     LDA >dy_2
     STA >D
     LDA <dy_2
@@ -55,12 +63,6 @@
     Add_16 >D, <D, #$00, #$01
     Sub_16 >D, <D, dx, #$00
 
-    ;;We wont use dx, only 2*dx from now on
-    dx_2 = $0809 ;Lets reuse dx register ($08)
-    LDA #$00
-    STA <dx_2
-    mult_16 >dx_2, <dx_2
-
 for_x:
     jsr pixel_draw
     ;;Increment X until X_pos = X_end and Y_pos = Y_end
@@ -75,10 +77,10 @@ for_x:
     BEQ case_2
 case_1:
     INC Y_pos
-    Sub_16 >D, <D, >dx_2, <dx_2; D = D - 2*dx
+    Sub_16 >D, <D, >V, <V; D = D - V
+    JMP for_x
 case_2:
     Add_16 >D, <D, >dy_2, <dy_2;D = D + 2*dy
-last:
     JMP for_x
 end:
 .endproc