Glibc 2.27 Will Premiere With Many Optimizations

When glibc 2.26 was released in August it was a noteworthy release with plenty of optimizations and introduced its own per-thread cache. With the next installment of the GNU C Library there will also be many more optimizations.

A few days back I wrote about more functions receiving FMA optimizations including powerf/logf/exp2f/log2f. That article also mentioned how replacing some old Assembly versions of functions with generic C code has also resulted in significant performance improvements. That's not all.

There are also other x86-64 function calls that have been optimized with FMA instruction support thanks to work by Intel engineers. There's also been a few functions optimized for SSE 4.1 in glibc. Even if running on older CPUs lacking SSE4, FMA, and friends, several functions have had their generic code optimized.

Aside from hand-tuning functions, glibc has received some tweaking around its malloc API for faster and safer process termination. Glibc 2.27 also has fixed a handful of security issues and other improvements are baking.

Those wishing to dig more closely through all the glibc 2.27 changes building up in recent months or to try out this latest development code, head on over to the GNU C Library code-base at sourceware.org.