Since 2002, this tool has been internally used for measuring memory bandwidth performance on different CPU architectures and, that “knowledge” was used on CRM32Pro (specific code paths) and also in other applications. Nowadays, this tools has been reworked and it is a complete memory bandwidth benchmark. The source code is available and also binaries for…
Update December 2025: I’ve updated this post with new benchmark results on a Ryzen 9800X3D, including comparisons using zlib, zlib-ng (as previously tested), and the new 2025 release of zlib-ng, which delivers a significant performance improvement.. Back in the 2013 when I began to work on CRM64Pro, I did a quick survey of possible compression…
Varias personas me han preguntado por mail si existe diferencia en el rendimiento entre usar solo SDL o CRM32Pro y la respuesta es afirmativa. Ojo, que esto solo es cierto en modo software. En modo hardware son totalmente identicas. Hace ya mas de un año me dedique a optimizar el memcpy para que utilizara MMX,…
