Hi, what does it mean when the specs for the Phenom II talk about 128 bit floating point?
Is it just the parallel processing of two 64-bit floating point numbers or is there hardware support for a real 128bit floating point type, i.e. with a significantly longer mantissa than 64 bits?
Thanks for your info,
I think this relates to the SIMD (SSEx) instructions; the older AMD CPUs used to have a 64-bit wide ALU path so took two clocks to process (4x32 bit) or (2x64 bit) words -- with a 128-bit data path the same work can be done in a single clock. So... parallel processing of 2 64-bit numbers etc...
woah.. all this computer terms is really a pain in the ***** for me to understand.. but bridgman explanation is great as i think i get th picture (just a little)