Re: My thoughts on C# and gaming.

On Fri, Jan 29, 2010 at 7:39 PM, Brandon Van Every <[email protected]> wrote:
> So I think "Mono is up to 5x slower
> than g++, when used with LLVM in unsafe mode" might be a fair
> statement.  I say "might" because what we really need for comparison
> is a C++ SciMark score.  In any case, Mono performs much better than a
> scripting language.  Not sure it performs as well as C++ though.

My laptop.  2GHz Pentium IV Centrino Duo, 2GB RAM, running Vista.
http://math.nist.gov/scimark2/ has the Java and C versions of SciMark
benchmarks.
http://www.cs.ucsb.edu/~ckrintz/racelab/PhxCSBenchmarks/SciMark.zip
has a C# version of SciMark.  I can't find an official homepage for
the C# port anymore.  The oft-cited Cornell webpage seems to have
vanished.  I converted the VS9 project to VS2010.

java.version: 1.6.0_13
    Composite Score: 411.3599872435552
C compiled with VS 2010 Beta 2 cl.exe -O2
    Composite Score:          704.78
C# compiled with VS 2010 Beta 2 Release, Mixed
    Composite Score: 422.93 MFlops

So on my machine, Java and C# .NET perform the same, and C++ is
~1.75x faster.

Your scores
> Mono 2.6
> Composite Score: 462.10 MFlops
>
> Mono Trunk using LLVM
> Composite88 Score: 640.94 MFlops
are interesting compared to mine because they're in a similar
ballpark.  I'll try setting up LLVM on my machine.  Below are the
boring details of my benchmarking.


Cheers,
Brandon Van Every


C:\Downloads\scimark2lib>java jnt.scimark2.commandline

SciMark 2.0a

Composite Score: 411.3599872435552
FFT (1024): 319.92676951527733
SOR (100x100):   635.5392370664161
Monte Carlo : 80.08217665654672
Sparse matmult (N=1000, nz=5000): 413.08542650628624
LU (100x100): 608.1663264732499

java.vendor: Sun Microsystems Inc.
java.version: 1.6.0_13
os.arch: x86
os.name: Windows Vista
os.version: 6.0

C:\Downloads\scimark2lib>


C:\Downloads\scimark2_1c>cl -o scimark2 -O2 *.c
C:\Downloads\scimark2_1c>scimark2
**                                                              **
** SciMark2 Numeric Benchmark, see http://math.nist.gov/scimark **
** for details. (Results can be submitted to [email protected])     **
**                                                              **
Using       2.00 seconds min time per kenel.
Composite Score:          704.78
FFT             Mflops:   524.27    (N=1024)
SOR             Mflops:   859.48    (100 x 100)
MonteCarlo:     Mflops:    80.49
Sparse matmult  Mflops:   676.33    (N=1000, nz=5000)
LU              Mflops:  1383.32    (M=100, N=100)

C:\Downloads\scimark2_1c>

// C# compiled with VS 2010 Release, Mixed
C:\Downloads\SciMark\SciMark\bin\Release>scimark
**                                                               **
** SciMark2a Numeric Benchmark, see http://math.nist.gov/scimark **
**                                                               **
Mininum running time = 2 seconds

Composite Score: 422.93 MFlops
FFT            : 377.38 - (1024)
SOR            : 629.52 - (100x100)
Monte Carlo    :  25.84
Sparse MatMult : 392.20 - (N=1000, nz=5000)
LU             : 689.74 - (100x100)

C:\Downloads\SciMark\SciMark\bin\Release>
_______________________________________________
Sweng-Gamedev mailing list
[email protected]
http://lists.midnightryder.com/listinfo.cgi/sweng-gamedev-midnightryder.com