Optimizing C# programs for Multi-Processors/ DualCore

[Guest]

Hello,

is there a way to optimize loops (f.e. matrix operations) for multi
processor systems with C#?

I´m searching for something like OpenMP, which is available for C/C++
but not for C#.

Or isn´t there any way to optimize it with C#?

Do I have to outsource my matrix operation into a C/C++ DLL using OpenMP?


Regards,

Martin

 

[james.curran]

is there a way to optimize loops (f.e. matrix operations) for multi
processor systems with C#?

In theory, this should be handled by the JITer (so that one assembly
could take advantage of the multiple processors when run on a system
that had them, while still running on a single processor PC)

In reality, I have no idea if such a JITer exists (but I doubt it).

 

[Carl Daniel [VC++ MVP]]

In theory, this should be handled by the JITer (so that one assembly
could take advantage of the multiple processors when run on a system
that had them, while still running on a single processor PC)

In reality, I have no idea if such a JITer exists (but I doubt it).

Not in .NET. There's no automatic vectorzation of loops and no support for
OpenMP, which would let you write parallel loops directly.

-cd

 

[Carl Daniel [VC++ MVP]]

Hello,

is there a way to optimize loops (f.e. matrix operations) for multi
processor systems with C#?

I´m searching for something like OpenMP, which is available for C/C++ but
not for C#.

Or isn´t there any way to optimize it with C#?

Do I have to outsource my matrix operation into a C/C++ DLL using OpenMP?

That would be a viable route. Personally, I'd probably try that route
before the alternative: writing it yourself for .NET.

-cd