Get the source code here: memxferjj.cc. Notes on how to compile and run the code are at the beginning of the file. There is also a makefile.
Sample output files:
(different AMD athlon systems 1333/1800,
SDRAM/DDR and CAS-latencies 2.5/2)
(various BIOS settings of one AMD athlon system)
memxferjj-amd64-out.txt (AMD64 2.2GHz, socket 939, dual channel DDR memory at 200MHz). Note how fast the memcpy() routine is (first to blocks). The third and fourth block suggest that the streaming routine should not be used for blocks of small size. The executable used is memxferjj.