LLVM 16.0-rc1 Brings New AMD & Intel CPU Support, Zstd Debug Sections, C++17 By Default

Following the LLVM 16.0 feature freeze and code branching earlier this week, LLVM 16.0.0-rc1 is now available as the first of at least three planned release candidates.

LLVM/Clang 16 is bringing many new features including initial AMD Zen 4 support (still missing the tuning), Intel Emerald Rapids support, Intel Meteor Lake support, Intel Sierra Forest and Grand Ridge support, an Xtensa architecture back-end, JIT support for OpenMP offloading, -Ofast and -ffast-math for Flang, -mcpu=native / -mtune=native for RISC-V, many new features implemented in libc++, Zstd compression for ELF / debug sections, LLVM is now built with C++17 by default, Arm Neoverse V2 support, the LoongArch back-end is no longer treated as experimental, a light AVX mode, and various other enhancements.

For those wanting to help in testing LLVM 16.0-rc1 is available via GitHub.

LLVM developers are planning for LLVM 16.0-rc2 around 7 February, LLVM 16.0-rc3 around 21 February, and to ideally ship LLVM 16.0 stable on 7 March. After that LLVM 16.0 will continue with seeing bi-weekly point releases as has become normal practice for this open-source compiler stack.