Toolchains for People in a Hurry: Latest Features of GCC/GLIBC
summary
glibc 2.33 changes the dynamic linker to look up optimized shared objects in a flat glibc hwcap directory hierarchy (x86_64_v2 / v3 / v4) so an app picks the right SSE/AVX build without deep subpaths, reducing fs lookups; GCC adds matching build targets via -march and a glibc-hwcaps compile option. GCC 11 rewrites f‑analyzer and defaults new diagnostics such as negative shift and shift >= width and writes-through-const warnings. Intel ISAs covered: AVX512 BNNI dot‑product intrinsics and AMX tile registers (8 tiles, up to 1KB each), usable via intrinsics or emulated with Intel SDE and provided Docker examples.
AI-written summary · about
chapters (8)
- 0:00Start
- 0:31Dynamic linker- GNU C Library version 2.33
- 1:31Example for 2017 version
- 3:38What can we improve from there?
- 6:16glibc-hwcaps for x86-64
- 8:28Objdump
- 10:07Compatibility
- 23:19Intel® Advanced Matrix Extensions ISA
related talks
43mToolchains for the Future
37mToolchain Options in 2023: What's New in Compilers and Libcs?
54mGCC/Clang Optimizations for Embedded Linux
59mAnatomy of Cross-Compilation Toolchains
01h 05mTutorial: Toolchains
45mA Few Effective gcc/clang Optimizations for Embedded Systems
45mCross compilation with clang and LLVM tools.
33mImproving the Effectiveness and Generality of GCC Auto-Vectorization