Avalon's rearchitecture - Performance

From the longhornblogs.com, here is an article talking about how and why Microsoft's Avalon (Longhorn's new graphics subsystem) team needed to rearchitect the product... even before it was released. They used .NET managed code and the performance and memory usage was very poor.

Avalon's rearchitecture - Performance:

http://www.simplegeek.com/permalink.aspx/359c73aa-4711-4061-b9d9-16300d8d278e

runtime
Friday, December 5, 2003

I'm not sure what the point of this post is, although I did enjoy the article.

Are you trying to bash Avalon because they had to re-architect, or are you patting the team on the back because they saw the problems and decided to fix them?

I don't know about you, but I've had to redesign several systems because they didn't work the way I expected them to. I can't really fault the folks working on Avalon for doing the same thing.

Steve Barbour
Friday, December 5, 2003

It could be a knock on .NET performance.

MR
Friday, December 5, 2003

I read the article, and my gut feeling was there was an "impedence mismatch" between the intended architecture and the intended implementation environment.

Not uncommon really. Not sure why it warranted a post here, other than "Avalon is new".

Brad Wilson (dotnetguy.techieswithcats.com)
Friday, December 5, 2003

I think the point of the post is that it is interesting.
These are people like us going through the same
type of development cycles we go through.
Microsoft has always been the big bad, this
kind of normalcy humanizes them a bit.

son of parnas
Friday, December 5, 2003

I don't think Microsoft has always been the "big bad". For a lot of the 1980s, IBM and Apple were the bad guys, particularly when Apple were issuing lawsuits left, right and centre.

John Topley (www.johntopley.com)
Friday, December 5, 2003

C/C++ is a bad language, the concept of pipes doesn't work, and using a Unix shell is just plain stupid, because Linus had to rewrite the whole Unix thang.

Slashbot
Friday, December 5, 2003

If the post has merit, it is a case study that jumping into a tank full of unknowns can often lead to disaster (or, slightly less dramatically, wasted effort) -- it is humorous to see a Microsoft group that dove headfirst into .NET and smacked their foreheads on the rocky shallows at a time when countless corporate development groups are diving headfirst into the next great silver bullet that is the purported best of all worlds.

Dennis Forbes
Friday, December 5, 2003

Maybe what it proves is that, when you have 10,000 developers, they can't all be Grade A developers any more.

Brad Wilson (dotnetguy.techieswithcats.com)
Saturday, December 6, 2003

  Recent Topics

  Fog Creek Home