=====================================
Exercise: Simple GEMM Implementations
=====================================

BLAS Level 3 Auxiliary Functions
================================

- Implement a function `dgecopy` for the operation $X \to Y$.

  If matrix $X$ is stored col major the function should copy the matrices
  column-by-column.  Otherwise the function is supposed to copy matrices
  row-by-row.  Other cases don't have to be considered.

- Implement a function `dgescal` for the operation $X \to \alpha X$.

  If matrix $X$ is stored col major the function should scale the matrices
  column-by-column.  Otherwise the function is supposed to scale matrices
  row-by-row.  Other cases don't have to be considered.


BLAS Level 3 GEMM
=================

- Complete the unimplemented `gemm_*` variants.
- Extend the `main` function that each variant gets tested
- Compare the performance measured in MFLOPS of all variants in a plot:

   - By default the matrices $A$, $B$, $C$ are stored in col major.  Create
     one plot for this case.
   - Figure out how to compile the benchmark for testing other cases.  Generate
     for at least one case another benchmark plot.


Skeleton for the Exercise
=========================
:import: session04/blas3_gemm_ex.c

:navigate: up    -> doc:index
           back  -> doc:session04/page06
           next  -> doc:session04/page08