slower than native?

[Jeremy Swigart]

Is it typical that when one tries the entropy example for example it appears to be slower than native?

Naive implementation:

        Entropy = 4194002.298628

        Time = 0.206650

Yeppp! implementation:

        Entropy = 4194002.298631

        Time = 2.988194

Or polynomial

Naive implementation:

        Time = 0.798373 secs

        Performance = 2.101427 GFLOPS

Yeppp! implementation:

        Time = 1.681965 secs

        Performance = 0.997477 GFLOPS

Max error:   0.000%

Am I reading these outputs correctly?

I'm not on a cutting edge processor(i5-2500), but I was expecting at little SSE/SSE2 speedup. Is my processor just outdated with regards to the optimization this library employs?

[Marat Dukhan]

Hi Jeremy,

Sorry for the late response, I missed your post when it was first submitted.

The most likely reason why you don't see a speedup is that you compile the program with 32-bit (x86, not x86-64) compiler, and link it with 32-bit Yeppp! library.

Yeppp! library does not include optimized implementations for 32-bit x86; you currently need to target x86-64 to get all performance benefits.

Regards,

Marat