Replace the current naive implementation with a fast implementation.
- Specialize on fixed-size small matrices (up to 4 x 4)
- Use cache-oblivious block decomposition
- Should not affect the user-end interface (except for providing additional API
for advanced users)
Original issue reported on code.google.com by linda...@gmail.com on 1 May 2012 at 2:43
Original issue reported on code.google.com by
linda...@gmail.comon 1 May 2012 at 2:43