emeryberger merged emeryberger/smash#93 · November 5, 2026 10:27 perf: reach and pass glibc on x86-64 and Graviton (lean entry points, block cache, deferred decommit) #93 SummarySmash is now faster than glibc on every allocator microbenchmark on x86-64, Graviton3 and macOS (M1), at parity on a g++ -O2 compile, and 0… +851 -270