and had a couple of questions. First, would it be a problem to extend these benchmarks to larger matrices? My calculations routinely involve matrix multiplication of two (21,000 by 100) matrices (I just took these numbers from a recent calculation), multiplied such that the result is a small, square matrix. Some of the benchmarks (e.g. the matrix-matrix product) seem to show ATLAS catching up for large matrices, and I wonder if it's liable to continue.
Relating to that, I was curious as to the options given to ATLAS. If it's running without sse2 support compiled in, then it should be compared with eigen2_novec, and is quite competitive and sometimes wins. However, if it was compiled with sse2 support, then eigen2 appears much more impressive.
I'm not sure when (if ever) I'll have time to rewrite my code (DFT++) to use eigen2, but my curiosity is piqued! -- David Roundy Physics Department Oregon State University http://physics.oregonstate.edu/~roundyd