[IronPython] Optimized methods and serious performance issues

Harry Pierson Harry.Pierson at microsoft.com
Thu May 25 19:00:18 CEST 2006


DirectX is COM based. Managed DirectX is a managed wraper around those COM obects.
 
Harry Pierson
Architect
Microsoft Architecture Strategy Team
email:                    hpierson at microsoft.com
IM/email:              harrypierson at hotmail.com
weblog:                 http://devhawk.net <http://.devhawk.net/>  
phone:                   425/705.6045
Make Trouble and Good Things Will Happen
 

________________________________

From: users-bounces at lists.ironpython.com on behalf of Jonathan Jacobs
Sent: Thu 5/25/2006 9:59 AM
To: Discussion of IronPython
Subject: Re: [IronPython] Optimized methods and serious performance issues



Dino Viehland wrote:
> If I had to take a guess it would be that we are constantly re-optimizing
> the method (failing to successfully cache it) - which would be a
> significant performance loss.
>
> Any chance we could get you to run a debug build w/ the -X:TrackPerformance
> and -D command line options?
>
> You should then see in the output "DrawSubset #" where # should be small
> (e.g. 1 or 2) - if it's large then we know the issue, we just need to get a
> simple repro of it.
>
> Do you know if the direct X classes come to you via a interop assembly?  Or
> if they're COM objects?  Maybe we have a strange interaction there that is
> preventing the back patching for one of those.

Output with self.mesh.DrawSubset:
FieldTable DrawSubset 1664
BackwardsFieldTableLookup DrawSubset 2
DrawSubset 1660

Output with Direct3D.Mesh.DrawSubset (as mentioned in my reply to J. Merrill):
FieldTable DrawSubset 5
BackwardsFieldTableLookup DrawSubset 2
DrawSubset 1

So it looks like your guess was right on the money.

I wouldn't know if they're interop or COM. How would I find out?

--
Jonathan

When you meet a master swordsman,
show him your sword.
When you meet a man who is not a poet,
do not show him your poem.
                 -- Rinzai, ninth century Zen master
_______________________________________________
users mailing list
users at lists.ironpython.com
http://lists.ironpython.com/listinfo.cgi/users-ironpython.com





More information about the Ironpython-users mailing list