Re: [DotGNU]Reminder - Weekly DotGNU Meeting

[Gopal V]

If memory serves me right, Rhys Weatherley wrote:
> 
> Lies, damned lies, and benchmarks ...
> 
> Depends on a lot of factors.  What was the person measuring?
> Time to run or time to load?  Was the app exceedingly complex,
> requiring lots of method conversion overhead?

Nov 15 13:48:33 <marcusU>       Actually, i just commented out the finalizers. 
Here's the numbers. C version: 7.58s, C# 10.13 (with finalizers), C# 8.7 
(finalizers disabled).
Nov 15 13:49:15 <marcusU>       Pnet is about 19 s.
Nov 15 13:49:21 <t3rmin4t0r>    *yow*
Nov 15 13:49:35 <t3rmin4t0r>    I expected something round 12
Nov 15 13:49:49 <marcusU>       Mono, with GCDONTGC=1 is the same as disabled 
finalizers..... 8.7 s.

Nov 15 13:50:18 <marcusU>       Wait a sec!
Nov 15 13:50:33 <t3rmin4t0r>    or 20 seconds ;-)
Nov 15 13:50:39 <marcusU>       I'm getting really different results now with 
ilrun than I got earlier.
Nov 15 13:50:51 <marcusU>       ilrun is giving me 9.656, which is faster than 
mono, now.
Nov 15 13:51:12 <t3rmin4t0r>    marcusU: don't fret ... I know lupus was asking 
Rhys to handle Marshalling specs
Nov 15 13:51:14 <marcusU>       Well, faster than mono with GC enabled.

Nov 15 13:52:42 <t3rmin4t0r>    not surprising as pnet has hardly any overhead 
on Pnvoke

> Method conversion is a trade-off: sometimes it gives a performance
> boost and sometimes it doesn't.  I'm sure you'd be able to create

Thunk between JIT and PInvoke is heavy  ??

Gopal
-- 
The difference between insanity and genius is measured by success