[Mono-bugs] [Bug 81663][Wis] Changed - Performance: Delegate optimization, DLR and IronPython

[bugzilla-daemon at bugzilla.ximian.com]

Please do not reply to this email- if you want to comment on the bug, go to the
URL shown below and enter your comments there.

Changed by kumpera at gmail.com.

http://bugzilla.ximian.com/show_bug.cgi?id=81663

--- shadow/81663	2007-05-25 10:39:11.000000000 -0400
+++ shadow/81663.tmp.1566	2007-05-25 17:57:47.000000000 -0400
@@ -200,6 +200,29 @@
 is virtual, but the type is static, so we should use a direct call.
 This optimization should benefit also all the other such cases.
 
 ------- Additional Comments From marek.safar at seznam.cz  2007-05-25 10:39 -------
 I was always wondering how hard and efficient it can be to implement
 sealed class { virtual method } optimization.
+
+------- Additional Comments From kumpera at gmail.com  2007-05-25 17:57 -------
+I posted a patch to mono-dev with the sealed class/method virtual
+dispatch optimization. This resulted in a very small improvement for
+ipy 2.0, only 4%.
+
+The GC overhead is justifiable since ipy 2.0 allocates a lot more
+memory than 1.1.
+
+Another solution to optimize delegate dispatch is to allow the runtime
+to violate the sealed flag and generate two specialized sub-classes of
+the delegate. For instance methods, just put the target method in the
+vtable and for static methods fix the stack and statically dispatch
+and return. 
+
+For the simple case, the cost of calling a delegate would be the same
+of a virtual-method.
+
+This would require a very intrusive patch, since there are quite a few
+places that check (method->klass->parent ==
+mono_defaults.multicastdelegate_class).
+
+