redid benchmarks with new options

especially calling did improve
also binaries don't get crazy big even for bigger heaps, because they are just ints

test/bench/results.md

@@ -23,8 +23,8 @@ Results (in ms) should be seen as relative, not absolute.
 
 
 language  |  noop   |  hello   |  add   |  call | loop        
-c         |    55   |   380    |   88   |   135 |    6
-go        |    52   |   450    |    9   |    77 |    2
-rubyx     |    42   |   200    | 1700   |  1450 |  470
+c         |    55   |   330    |   88   |   135 |    6
+go        |    52   |   400    |    9   |    77 |    2
+rubyx     |    42   |   160    | 1575   |   780 |  430
 ruby      |  1570   |   650    | 1090   |  1500 |  180
-mruby     |    86   |  1200    | 1370   |  2700 |  300
+mruby     |    86   |  1100    | 1370   |  2700 |  300

test/bench/rubyx/adds.rb

@@ -14,9 +14,9 @@ class Space
     return result
   end
 
-  # ran with --parfait=80000
+  # ran with --integers=1600000 , result -noop * 25
   def main(arg)
-    b = 1000
+    b = 2000
     while( b >= 1 )
       b = b - 1
       fibo_i(40)

test/bench/rubyx/calls.rb

@@ -9,9 +9,9 @@ class Space
      return a + b
   end
 
-  # ran with --parfait=70000
+  # ran with --integers=170000 , result - noop * 10
   def main(arg)
-    b = 2
+    b = 5
     res = 0
     while( b >= 1 )
       b = b - 1

test/bench/rubyx/hello.rb

@@ -1,5 +1,5 @@
 class Space
-  # ran with --parfait=25000
+  # ran with --parfait=21000
   def main(arg)
     b = 10000
     while( b >= 1 )

test/bench/rubyx/loop.rb

@@ -1,6 +1,6 @@
 class Space
 
-  # ran with --parfait=101000
+  # ran with --integers=102400 , result -noop * 10
   def main(arg)
     b = 100000
     while( b >= 1 )