From c4b98c7be27558c662a23849126d2a802c9bf4bc Mon Sep 17 00:00:00 2001
From: Brooke Vibber <bvibber@pobox.com>
Date: Mon, 30 Dec 2024 05:35:22 -0800
Subject: [PATCH] optimize out a temporary

down to 11.076 ms/px on xe
---
 mandel.s | 17 +++++++++++++----
 1 file changed, 13 insertions(+), 4 deletions(-)

diff --git a/mandel.s b/mandel.s
index 9f594e8..0239167 100644
--- a/mandel.s
+++ b/mandel.s
@@ -393,6 +393,18 @@ viewport_oy:
     sta dest + 1
 .endmacro
 
+; clobbers a, x
+.macro sqr8_add16 dest, arg
+    ldx arg
+    clc
+    lda sqr_lobyte,x
+    adc dest
+    sta dest
+    lda sqr_hibyte,x
+    adc dest + 1
+    sta dest + 1
+.endmacro
+
 ; lookup table for top byte -> PORTB value for bank-switch
 ;.align 256 ; warning - if not aligned this will cost an extra cycle sometimes
 bank_switch_table:
@@ -720,7 +732,6 @@ arg2_pos:
         ; h*h*256*256 + h*l*256 + h*l*256 + l*l
 
         sqr8 result, arg
-        ;imul8 inter, arg, arg, xe
         lda #0
         sta result + 2
         sta result + 3
@@ -731,9 +742,7 @@ arg2_pos:
         add16 result + 1, result + 1, inter
         add_carry result + 3
 
-        sqr8 inter, arg + 1
-        ;imul8 inter, arg + 1, arg + 1, xe
-        add16 result + 2, result + 2, inter
+        sqr8_add16 result + 2, arg + 1
 
         rts ; 6 cyc
     .endscope