ThunderX2 Getting Big MEMMOVE Performance Boost With Glibc 2.30
The memmove function for moving a block of memory will be much faster on ThunderX2 hardware with this next Glibc release.
The memmove code has been optimized for the ThunderX2 so that it's now about 20~30% faster for larger cases or 1~5% faster when moving smaller blocks of memory. This optimization comes by using SIMD load/stores rather than GPRs for large overlapping forward moves. There's also been fixes as part of this code and other improvements.
More details via this commit that was merged on Friday to Glibc Git for the upcoming Glibc 2.30 release. Glibc 2.30 is expected to be released around the start of August.