From 3553ce986f6721f8c6d446368cb6c6f55186713b Mon Sep 17 00:00:00 2001
From: Jamey Sharp <jamey@minilop.net>
Date: Tue, 31 Dec 2024 02:55:22 -0800
Subject: [PATCH] shave some cycles off 16-bit squaring with shift instead of
 add

also fix the comments about how many cycles shift takes
---
 mandel.s | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/mandel.s b/mandel.s
index 299db98..b0c2b42 100644
--- a/mandel.s
+++ b/mandel.s
@@ -348,7 +348,7 @@ viewport_oy:
     sub 4, dest, arg1, arg2
 .endmacro
 
-; 3 + 5 * bytes cycles
+; 3 + 5 * (bytes - 1) cycles
 .macro shl bytes, arg
     asl arg              ; 3 cyc
     .repeat bytes-1, i
@@ -356,17 +356,17 @@ viewport_oy:
     .endrepeat
 .endmacro
 
-; 13 cycles
+; 8 cycles
 .macro shl16 arg
     shl 2, arg
 .endmacro
 
-; 18 cycles
+; 13 cycles
 .macro shl24 arg
     shl 3, arg
 .endmacro
 
-; 23 cycles
+; 18 cycles
 .macro shl32 arg
     shl 4, arg
 .endmacro
@@ -787,7 +787,7 @@ arg2_pos:
         sqr8 result + 2, arg + 1
 
         imul8 inter, arg + 1, arg, xe
-        add16 result + 1, result + 1, inter
+        shl16 inter
         add_carry result + 3
         add16 result + 1, result + 1, inter
         add_carry result + 3