Transposes a flat row-major rows×cols array into flat column-major storage
(lda = rows) — column-major storage of A is just row-major storage of Aᵀ.
Used so the Level 2 benchmarks can feed genuinely column-major data to
GpuMatrix.from(..., "column-major"), matching cuBLAS's native layout.
Transposes a flat row-major rows×cols array into flat column-major storage (lda = rows) — column-major storage of A is just row-major storage of Aᵀ. Used so the Level 2 benchmarks can feed genuinely column-major data to
GpuMatrix.from(..., "column-major"), matching cuBLAS's native layout.