Is there an ARM Neon SIMD Gather instruction?
I am looking for an equivalent to the following Intel Intrinsic instruction:
_mm_i32gather_ps(blob, index, 4);
I can find related instructions such as:
svld1_gather
in the arm developer manualGatherIndex
functionAlthough these functions are related, I do not see an equivalent and am asking myself whether such loads are supported on Arm Neon.
No, there is no gather in Neon. Scatter/gather instructions are explicitly one of the improvements of SVE/SVE2 over Neon (and thus reasons to move to SVE/SVE2, along with scalable vectors for different targets etc) See SVE documentation for more.