1\begin{tabular}{llrrrllrrrr}2\toprule3Node & chip & GPU cores & CPU thr. & GB & tests (metallib / runtime) & cold / warm compile (ms) & RV fused $1000\times3900$ speedup & regression speedup & GARCH P=64 speedup & rolling speedup \\4\midrule5m2m8a & Apple M2 & 10 & 8 & 8 & pass / pass & 110 / 1 & 3.99 & 1.61 & 6.67 & 0.65 \\6M2M32c & Apple M2 Max & 30 & 12 & 32 & pass / pass & 48 / 1 & 3.61 & 1.35 & 7.84 & 0.52 \\7M2U64 & Apple M2 Ultra & 60 & 24 & 64 & pass / pass & 49 / 1 & 2.18 & 1.08 & 4.92 & 0.32 \\8M3BA24 & Apple M3 & 10 & 8 & 24 & pass / pass & 44 / 0 & 2.69 & 1.68 & 6.21 & 0.36 \\9M3U96b & Apple M3 Ultra & 60 & 28 & 96 & pass / pass & 47 / 1 & 1.68 & 0.78 & 3.18 & 0.24 \\10m4md & Apple M4 & 10 & 10 & 16 & pass / pass & 42 / 0 & 2.55 & 1.34 & 4.21 & 0.38 \\11m4mm & Apple M4 Pro & 16 & 12 & 48 & pass / pass & 39 / 0 & 1.63 & 0.91 & 2.14 & 0.24 \\12M4M64c & Apple M4 Max & 40 & 16 & 64 & pass / pass & 40 / 0 & 2.80 & 1.19 & 4.82 & 0.48 \\13m5ma & Apple M5 Pro & 16 & 15 & 24 & pass / pass & 44 / 0 & 1.13 & 0.55 & 1.83 & 0.20 \\14M5M36 & Apple M5 Max & 32 & 18 & 36 & pass / pass & 37 / 0 & 1.12 & 0.60 & 2.14 & 0.23 \\15m6ma & Apple M6 & 12 & 12 & 24 & pass / pass & 40 / 0 & 1.63 & 0.79 & 3.72 & 0.34 \\16\bottomrule17\end{tabular}