Hi Jirka, >Any comments on this? I am not familiar with the Athlon64. It is possible that some marketroids misunderstood "simultaneously processes four 32-bit floating points" as "128-bit floating point arithmetic". >Which processors supports 128-bit floating point arithmetic? None that I work with. These days, I just know about Pentium 4, PowerPC G4, SPARC64, DEC/Compaq TRU64's Alpha. If you REALLY need the higher precision of 128-bit floating point (or better), there are C++ libraries to support floating point (and fixed point) numbers with greater precision. HTH, --Eljay
