Commits
Click on a commit to change the comparison rangeAdd between pushdown kernel for DecimalByteParts Add decimal between benchmark vs arrow-rs Cap decimal between bench sizes so all cases stay under 1ms Add two-limb i128 representation and lexicographic between to DecimalByteParts Replace two-limb between with a fused single-pass kernel Beat arrow on two-limb between by reconstructing i128 in the fused pass Push two-limb comparisons down with a fused i128 reconstruct kernel Add decimal lt benchmark vs arrow-rs Add limb-SIMD microbenchmark: AVX-512 limb compare vs arrow i128 Add AVX-512 limb kernel for two-limb decimal compare and between Trim decimal comparison benchmarks to a justified, <1ms set Consolidate two-limb compare/between plumbing into one helper Share the two-limb test array builder across compute test modules