.NET 2 UI speed issue demystified

[Tom Ollar]

If this is in reference to our article. Our tests contained NO timers. Could it be that the boys and girls at our esteemed MotherShip saw two guys in a garage and decided to nip this whole C# as a UI platform concept in the bud? Don't laugh Frank, you may be next...

[Marc Clifton]

Tom Ollar wrote:

Our tests contained NO timers.

So, um, how did you measure the performance? An hourglass? ;P Marc VS2005 Tips & Tricks -- contributions welcome!

[leppie]

Tom Ollar wrote:

Our tests contained NO timers.

It's NOT timers being the slow down, its starting and stopping the Timer. I'm gonna use a Threading timer to compare. xacc.ide-0.1 released! Download and screenshots

[leppie]

Ok Threading Timers is even a bigger NO-NO! .NET 1.1 - same slow down as before .NET 2 - 11400% slow down... :doh: xacc.ide-0.1 released! Download and screenshots

[WillemM]

So, this means the threading is dramatically slowed down in .NET 2.0 So much for concurrency using .NET ;P Did anyone report this to microsoft? Maybe they can explain this and fix the problem. WM.
What about weapons of mass-construction?

[S Senthil Kumar]

You sure this is not a one-off issue with your PC? Did you get the same results on a different machine? Regards Senthil _____________________________ My Blog | My Articles | WinMacro

[Marc Clifton]

WillemM wrote:

So, this means the threading is dramatically slowed down in .NET 2.0

No, no. Just don't use timers. :) Marc VS2005 Tips & Tricks -- contributions welcome!

[WillemM]

Pfew, I thought there for a minute that the performance was really bad. I still need a lot of testing for most of my applications to be sure that it runs well on .NET 2.0 :) (Some of them do use timers) WM.
What about weapons of mass-construction?

[Marc Clifton]

leppie wrote:

Ok Threading Timers is even a bigger NO-NO!

This is really wierd. Keep me posted, because I simply cannot imagine breaking something so much that it causes an 11,000% slowdown. Marc VS2005 Tips & Tricks -- contributions welcome!

[leppie]

Interesting point, I will test on Centrino. xacc.ide-0.1 released! Download and screenshots

[leppie]

Same result... :sigh: xacc.ide-0.1 released! Download and screenshots

[Daniel Turini]

Sometimes, the code is so tight, that adding an additional parameter validation, removing the "sealed" or even making a method virtual (which prevents inlining) may slowdown your code by an order of magnitude. This seems the case of .NET timers, which, AFAIK, are simply wrappers around a simple Win32 API. I don't see dead pixels anymore... Yes, even I am blogging now!

[Super Lloyd]

Somehow 2.0 seems to be slower to the point of worry.... :doh:

[S Douglas]

leppie wrote:

.NET 2 - 11400% slow down...

Leppie, I hope you’re not serious. I was looking forward to 2.0 and finally starting up upgrade my skills to .NET. :( (Not that I ever use timers but this isn’t the first post I have seen complaining about speed issues with .NET 2.0)

---

ZeePain! wrote:

This seems like one of those programs that started small, grew incrementally, building internal pressure, and finally barfed all over its source code sneakers. Or something.

thedailywtf.com[^]