What is the impact of RTE_FORCE_INTRINSIC on performance

Hi,

I'm running DPDK on Intel x86 processor which supports SSE2,
in my current deployment I didn't enabled RTE_FORCE_INTRINSIC.
After reading on Intel's intrinsic pause I wonder if I should enable it or
not.
especially how it should impact on latency those noop command going through
the CPU.

Also, I wonder does it matter if I have contention on shard objects pool
and on spinlocks?

Thanks,
Roy