From 8d7a1675f86681c40f98a32c7a8ee62a3d08f8a3 Mon Sep 17 00:00:00 2001
From: d_m <d_m@plastic-idolatry.com>
Date: Sat, 10 Sep 2022 13:31:35 -0400
Subject: [PATCH] neauoire's optimizations

---
 math32.tal | 43 +++++++++++++++++++------------------------
 1 file changed, 19 insertions(+), 24 deletions(-)

diff --git a/math32.tal b/math32.tal
index 3edca56..e01b014 100644
--- a/math32.tal
+++ b/math32.tal
@@ -54,11 +54,6 @@
 (   - mul32 memory, 12 bytes                                      )
 (   - _divmod32 memory, 16 bytes                                  )
 
-%TOR { ROT ROT } ( a b c -> c a b )
-%COMPLEMENT32 { SWP2 #ffff EOR2 SWP2 #ffff EOR2 }
-%DUP4 { OVR2 OVR2 }
-%POP4 { POP2 POP2 }
-
 ( bitcount: number of bits needed to represent number )
 ( equivalent to floor[log2[x]] + 1 )
 
@@ -82,7 +77,7 @@
     SWP ;bitcount8 JSR2 ADD ( nhi+nlo )
     JMP2r
     &hi-set
-    SWP POP #08 ADD ( nhi+8 )
+    NIP #08 ADD ( nhi+8 )
     JMP2r
 
 @bitcount32 ( x** -> n^ )
@@ -90,9 +85,9 @@
     ;bitcount16 JSR2 ( xlo* nhi )
     DUP #00 NEQ ( xlo* nhi nhi!=0 )
     ,&hi-set JCN ( xlo* nhi )
-    TOR ;bitcount16 JSR2 ADD JMP2r ( nhi+nlo )
+    ROT ROT ;bitcount16 JSR2 ADD JMP2r ( nhi+nlo )
     &hi-set
-    TOR POP2 #10 ADD ( nhi+16 )
+    ROT ROT POP2 #10 ADD ( nhi+16 )
     JMP2r
 
 ( equality )
@@ -113,7 +108,7 @@
 
 ( x != 0 )
 @non-zero32 ( x** -> bool^ )
-    ORA2 #0000 NEQ2 JMP2r
+    ORA2 ORA JMP2r
 
 ( comparisons )
 
@@ -165,7 +160,7 @@
 
 ( ~x )
 @complement32 ( x** -> ~x** )
-    COMPLEMENT32 JMP2r
+    SWP2 #ffff EOR2 SWP2 #ffff EOR2 JMP2r
 
 ( temporary registers )
 ( shared by most operations, except mul32 and div32 )
@@ -185,20 +180,20 @@
 
 ( shift right by 0-7 bits )
 @rshift32-0 ( x** n^ -> x<<n )
-        STHk  SFT                      ;m32/z3 STA  ( write z3 )
+        STHk  SFT                      ;m32/z3 STA   ( write z3 )
     #00 STHkr SFT2 #00 ;m32/z3 LDA ORA2 ;m32/z2 STA2 ( write z2,z3 )
     #00 STHkr SFT2 #00 ;m32/z2 LDA ORA2 ;m32/z1 STA2 ( write z1,z2 )
-    #00 STHr  SFT2 #00 ;m32/z1 LDA ORA2             ( compute z0,z1 )
+    #00 STHr  SFT2 #00 ;m32/z1 LDA ORA2              ( compute z0,z1 )
     ;m32/z2 LDA2
     JMP2r
 
 ( shift right by 8-15 bits )
 @rshift32-1 ( x** n^ -> x<<n )
     #08 SUB STH POP
-        STHkr SFT                      ;m32/z3 STA  ( write z3 )
+        STHkr SFT                      ;m32/z3 STA   ( write z3 )
     #00 STHkr SFT2 #00 ;m32/z3 LDA ORA2 ;m32/z2 STA2 ( write z2,z3 )
-    #00 STHr  SFT2 #00 ;m32/z2 LDA ORA2             ( compute z1,z2 )
-    #00 TOR ;m32/z3 LDA
+    #00 STHr  SFT2 #00 ;m32/z2 LDA ORA2              ( compute z1,z2 )
+    #00 ROT ROT ;m32/z3 LDA
     JMP2r
 
 ( shift right by 16-23 bits )
@@ -239,7 +234,7 @@
     #00 SWP STHkr SFT2                     ;m32/z1 STA2 ( store z1,z2 )
     #00 SWP STHkr SFT2 #00 ;m32/z1 LDA ORA2 ;m32/z0 STA2 ( store z0,z1 )
             STHr  SFT      ;m32/z0 LDA ORA              ( calculate z0 )
-    SWP POP ( x0 unused )
+    NIP ( x0 unused )
     ;m32/z1 LDA2 #00
     JMP2r
 
@@ -256,7 +251,7 @@
 @lshift32-3 ( x** n^ -> x<<n )
     #18 SUB #40 SFT ( x0 x1 x2 x3 r=[n-24]<<4 )
     SFT ( x0 x1 x2 x3<<r )
-    SWP2 POP2 SWP POP #0000 #00
+    NIP2 NIP #0000 #00
     JMP2r
 
 ( arithmetic )
@@ -265,7 +260,7 @@
 @add32 ( xhi* xlo* yhi* ylo* -> zhi* zlo* )
     ;m32/y2 STA2 ;m32/y0 STA2 ( save ylo, yhi )
     ;m32/x2 STA2 ;m32/x0 STA2 ( save xlo, xhi )
-    #0000 #0000 ;m32/z0 STA2 ;m32/z2 STA2 ( reset zhi, zlo )
+    #0000 DUP2 ;m32/z0 STA2 ;m32/z2 STA2 ( reset zhi, zlo )
 
     ( x3 + y3 => z2z3 )
     #00 ;m32/x3 LDA #00 ;m32/y3 LDA ADD2 ;m32/z2 STA2
@@ -288,9 +283,9 @@
 
 ( -x )
 @negate32 ( x** -> -x** )
-    COMPLEMENT32
+    ;complement32 JSR2 ( ~x** )
     INC2 ( ~xhi -xlo )
-    DUP2 #0000 NEQ2 ( ~xhi -xlo non-zero? )
+    DUP2 ORA ( ~xhi -xlo non-zero? )
     ,&done JCN ( xlo non-zero => don't inc hi )
     SWP2 INC2 SWP2 ( -xhi -xlo )
     &done
@@ -380,7 +375,7 @@
     #00 DUP2 ( shift 0 shift 0 )
 
     ( 1<<shift -> cur )
-    #0000 #0001 ROT2 POP
+    #0000 INC2k ROT2 POP
     ;lshift32 JSR2 ,&cur1 STR2 ,&cur0 STR2
 
     ( div<<shift -> div )
@@ -412,10 +407,10 @@
 ( greatest common divisor - euclidean algorithm )
 @gcd32 ( x** y** -> z** )
     &loop ( x y )
-    DUP4 ( x y y )
+    OVR2 OVR2 ( x y y )
     ;is-zero32 JSR2 ( x y y=0? )
     ,&done JCN ( x y )
-    DUP4 ( x y y )
+    OVR2 OVR2 ( x y y )
     STH2 STH2 ( x y [y] )
     ;mod32 JSR2 ( r=x%y [y] )
     STH2r ( rhi rlo yhi [ylo] )
@@ -426,5 +421,5 @@
     ROT2 ( yhi ylo rhi rlo )
     ,&loop JMP
     &done
-    POP4 ( x )
+    POP2 POP2 ( x )
     JMP2r