• by rurban on 6/3/2025, 3:36:56 AM

    So slower than tcc on O0 and slower than qbe on O1, with less backends, but tight llvm integration usable as baseline jit compiler, where it already outperforms Cranelift.