C++ implementation of the algorithm in "Fast and Accurate Least-Mean-Squares Solvers", NIPS19
Execution result:
--- matrix size ---
P:6 8192
S:6 37
--- matmul ---
2739.82 10.0992 0.190794 12.6404 -44.2129 38.4664
10.0992 2694.16 0.64991 -41.7519 -29.4058 -25.0661
0.190794 0.64991 2741.28 -22.5884 -61.4977 -24.4794
12.6404 -41.7519 -22.5884 2721.83 -41.2449 4.36757
-44.2129 -29.4058 -61.4977 -41.2449 2714.95 -26.0089
38.4664 -25.0661 -24.4794 4.36757 -26.0089 2784.76
--- caratheodory ---
2739.82 10.0992 0.190794 12.6404 -44.2129 38.4664
10.0992 2694.16 0.64991 -41.7519 -29.4058 -25.0661
0.190794 0.64991 2741.28 -22.5884 -61.4977 -24.4794
12.6404 -41.7519 -22.5884 2721.83 -41.2449 4.36757
-44.2129 -29.4058 -61.4977 -41.2449 2714.95 -26.0089
38.4664 -25.0661 -24.4794 4.36757 -26.0089 2784.76
--- time ---
matmul :0.092752[msec]
caratheodory :76.1424[msec]