Hi Jianbo (and other developers of vectorized PMDs),
2016-08-24 15:23, Jianbo Liu:
> Use ARM NEON intrinsic to implement i40e vPMD
Have you tried to use the generic SIMD intrinsics?
We could maintain only one vectorized implementation by using
__attribute__ ((vector_size (n)))
as described in https://gcc.gnu.org/onlinedocs/gcc/Vector-Extensions.html
I don't know the limitations of the vector builtins (support, performance,
endianness, etc) but it is worth making a try.
Currently we target to support each PMD for SSE/AVX, Altivec and NEON.
Is there any volunteers working on Intel, POWER and ARM to try converting
the existing codebase?
Thanks
