0001-optimize-CLASS_OF.patch

From 98be06eba1e2f09bbfba9c63ec5f79581ad7a5fa Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E5=8D=9C=E9=83=A8=E6=98=8C=E5=B9=B3?=
 <shyouhei@ruby-lang.org>
Date: Wed, 23 Oct 2019 17:10:10 +0900
Subject: [PATCH] optimize CLASS_OF

---
 include/ruby/ruby.h | 28 +++++++++++++++++++---------
 1 file changed, 19 insertions(+), 9 deletions(-)

diff --git a/include/ruby/ruby.h b/include/ruby/ruby.h
index b144fa0111..4467db0755 100644
--- a/include/ruby/ruby.h
+++ b/include/ruby/ruby.h
@@ -2094,17 +2094,27 @@ RUBY_EXTERN VALUE rb_stdin, rb_stdout, rb_stderr;
 static inline VALUE
 rb_class_of(VALUE obj)
 {
-    if (RB_IMMEDIATE_P(obj)) {
-	if (RB_FIXNUM_P(obj)) return rb_cInteger;
-	if (RB_FLONUM_P(obj)) return rb_cFloat;
-	if (obj == RUBY_Qtrue)  return rb_cTrueClass;
-	if (RB_STATIC_SYM_P(obj)) return rb_cSymbol;
+    if (!RB_SPECIAL_CONST_P(obj)) {
+        return RBASIC_CLASS(obj);
     }
-    else if (!RB_TEST(obj)) {
-	if (obj == RUBY_Qnil)   return rb_cNilClass;
-	if (obj == RUBY_Qfalse) return rb_cFalseClass;
+    else if (obj == RUBY_Qfalse) {
+        return rb_cFalseClass;
+    }
+    else if (obj == RUBY_Qnil) {
+        return rb_cNilClass;
+    }
+    else if (obj == RUBY_Qtrue) {
+        return rb_cTrueClass;
+    }
+    else if (RB_FIXNUM_P(obj)) {
+        return rb_cInteger;
+    }
+    else if (RB_STATIC_SYM_P(obj)) {
+        return rb_cSymbol;
+    }
+    else {
+        return rb_cFloat;
     }
-    return RBASIC(obj)->klass;
 }
 
 static inline int
-- 
2.17.1

99-bench

Calculating -------------------------------------
                         ours@clang-10  ours@gcc-9  ours@clang-9  master@clang-10  master@gcc-9  master@clang-9
Optcarrot Lan_Master.nes        45.113      46.777        45.515           37.691        46.349          38.487 fps

Comparison:
             Optcarrot Lan_Master.nes
              ours@gcc-9:        46.8 fps
            master@gcc-9:        46.3 fps - 1.01x  slower
            ours@clang-9:        45.5 fps - 1.03x  slower
           ours@clang-10:        45.1 fps - 1.04x  slower
          master@clang-9:        38.5 fps - 1.22x  slower
         master@clang-10:        37.7 fps - 1.24x  slower