[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[DotGNU]FWD : Re: Mono.
From: |
Gopal.V |
Subject: |
[DotGNU]FWD : Re: Mono. |
Date: |
Tue, 12 Mar 2002 00:17:16 +0530 |
User-agent: |
Mutt/1.2.5i |
FYI
If memory serves me right, Miguel de Icaza wrote:
> If memory serves me right, Bill Lance wrote:
> > More on the Mono front
> >
> http://lists.ximian.com/archives/public/mono-list/2002-March/003726.html
> Well it compiles, but doesn't produce a working MCS gen2. Also
> takes more than 15 minutes to compile. (compared to a 3 minute cscc
> compile). Imagine compiling the whole .NET system with MCS. Compile
> farm
> time gentlemen.
>
> I think that this is Mono crossing a hurdle they set for themselves
> when they decided to write the compiler in C#.
>
> It took 15 minutes the first time.
>
> This was further reduced by changing the implementation of `GetHashCode'
> from returning 0 to return a real hash code, and the compile time
> dropped to 3 minutes, 15 seconds.
>
> This is with the interpreter. There is a bug in the JITer that stops us
> from testing it, and we are looking into the problem, but that should
> cut down the time to the 40 seconds range.
>
> One big problem that shows up on the profiles is not the actual
> compiler, but code on the class libraries that has not been optimized.
> For example, fixing the simplistic StringBuilder.Append implementation
> reduced the compile in another 30% at some point.
>
> Our profiling shows that most of the time is spend in miss-implemented
> functions in our runtime (when comparing against the Microsoft
> implementation).
>
> Miguel.
--
The difference between insanity and genius is only measured by success
//===<=>===\\
|| GNU RULEZ ||
\\===<=>===//
[Prev in Thread] |
Current Thread |
[Next in Thread] |
- [DotGNU]FWD : Re: Mono.,
Gopal.V <=