Currently any cast operation with fixed length vectors uses NEON costs, If those operations are end up using SVE instruction then we estimate those operations based upon SVE costs. Differential Revision: https://reviews.llvm.org/D133955
142 KiB
142 KiB