Hi,
I have found that calling math functions like log and exp on elements of an
array of type double[] is slower for the vectorized case than
non-vectorized case.
foreach(i=0..N) {
a[i] = log(b[i]);
}
is slower for various array sizes than
for(uniform int i = 0; i < N; i++) {
a[i] = log(b[i]);
}
For floats, the vectorized version is faster as expected.
I have gone through the previous threads with similar issues and the 2
suggestions that stood out :
1) Changing the target to try different vector width
2) Linking against Intel's SVML math library
I tried both the suggestions without much success.
What is the recommended approach to get good double precision performance
in ISPC?
Thanks,
Pankaj
--
You received this message because you are subscribed to the Google Groups
"Intel SPMD Program Compiler Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
