[ODE] Windows mingw compile problems

[Martin C. Martin]

Cool!  But how much does it help ODE overall?  Do you get even a 10% 
speedup?

- Martin

Daniel Moser Tralamazza wrote:

> Quoting Brent Burton <brentb at io.com>:
> 
>>Hi Daniel,
>>
>>I'm wondering what kind of a speedup you got with the SIMD version of
>>these routines.  For example, what's the number of calls per second versus
>>the original versions? (Or msec per call, whatever.)
>>
>>-Brent
> 
> 
> Hi Brent,
> 
> I just made a simple app to test these routines.
> Every function do a for loop count=1000.
> I used RDTSC instruction to take these timing diferences [RDTSC value]:
> (tested cpu =athlon-xp 1500)
> 
> dVector3 a = (3,4,5);
> testNormalize3(a)
> (simd) v1=0.424264,v2=0.565685,v3=0.707107 [50484] (2x faster)
> (std)  v1=0.424264,v2=0.565685,v3=0.707107 [103584]
> 
> testRecipSqrt(3.141592f)
> (simd) v=0.564190 [15263] (1.88x faster)
> (std)  v=0.564190 [53262]
> 
> testSqrt(3.141592f)
> (simd) v=1.772454 [17053] (3.48x faster)
> (std)  v=1.772454 [32105]
> 
> 
> Daniel Tralamazza
> 
> _______________________________________________
> ODE mailing list
> ODE at q12.org
> http://q12.org/mailman/listinfo/ode