Digging into interface calls in the .NET Framework: Stub-based dispatch.

re: Digging into interface calls in the .NET Framework: Stub-based dispatch.

snaveen — Mon, 13 Mar 2006 19:50:37 GMT

Cool post :)

re: Digging into interface calls in the .NET Framework: Stub-based dispatch.

hvassbotn — Tue, 21 Mar 2006 14:10:19 GMT

Very interesting topic and a great post!

"We can go into details of polymorphic dispatch sites in another blog entry if there is interest."

There is! ;)

It would also be interesting to read about:

- How this dynamic code-stub dispatching mechanism differes from the .NET 1.x table-based dispatch (IIRC)
- Why you chose to select this design? Why change away from the 1.x mechanism?
- How interface dispatch performance differs from 1.x and differs from delegate invokation

/Hallvard Vassbotn

Performance Quiz #9 -- extra innings due to dispatch stubs

Rico Mariani's Performance Tidbits — Wed, 22 Mar 2006 04:04:56 GMT

Well you'll recall that in the Performance Quiz #9 solution&nbsp;there was a surprising result where...

Passion is like genius; a miracle. » How to call methods more efficiently in dynamic language

Sat, 23 Sep 2006 21:42:14 GMT

PingBack from http://mkseo.pe.kr/blog/?p=1621

Performance Quiz #9 -- extra innings due to dispatch stubs

Rico Mariani's Performance Tidbits — Wed, 24 Jan 2007 05:24:31 GMT

Well you'll recall that in the Performance Quiz #9 solution there was a surprising result where Test7

re: Digging into interface calls in the .NET Framework: Stub-based dispatch.

qwertie — Sat, 12 Apr 2008 17:25:55 GMT

I have always wondered how the performance of calling through an interface differs from virtual calls. Are there any possible performance penalties of calling interfaces when there is a complex inheritance hierarchy of interfaces, or when one object implements multiple generic interfaces of the same basic type (e.g. class Foo : IFoo<int>, IFoo<string>)? What about calling methods of a struct through an interface?

re: Digging into interface calls in the .NET Framework: Stub-based dispatch.

vancem — Mon, 14 Apr 2008 23:27:23 GMT

For the most part, dispatch performace is independent of the complexity of the inheritance hierarchy. There are differences between interfaces an 'normal' virtual as mentioned in the article, but the fact the complexity of the inheritance does not really matter (or if they are generic).

Calling interfaces defined on structs is an interesting case. Because structs don't support inheritance, you typically know its exact type, in which case it is not even a virtual dispatch (just a normal call). If you have cast the struct to an object, you have BOXed it (copied it to the GC heap as an object), in which case the dispatch is much like any other object.

JIT Optimizations

Ars programatica — Tue, 13 May 2008 14:54:08 GMT

Článek JIT Optimizations na populárním serveru CodeProject nabízí celkem pěkný a podrobný pohled na to,

Vance Morrison s Weblog Digging into interface calls in the NET | Quick Diets

Vance Morrison s Weblog Digging into interface calls in the NET | Quick Diets — Wed, 10 Jun 2009 06:25:49 GMT

PingBack from http://quickdietsite.info/story.php?id=4901

CLR Injection ?????? ??????. | blog.zerople();

CLR Injection ?????? ??????. | blog.zerople(); — Tue, 23 Jun 2009 18:39:41 GMT

PingBack from http://www.pimslog.com/2009/06/clr-injection-%ea%b4%80%eb%a0%a8-%eb%a7%81%ed%81%ac/