081106 / ARM VFP11

previous | next

ARM Reference Material

ARM Instruction Quickref
ARM Infocenter

Here for my future reference ... starting to get a little more interested in the iPhone given that it actually has a SIMD float unit (VFP11).

Intel Core i7

Above is a shot of one of the 4 cores in Intel's new Core i7 chip taken from this PDF. An even larger space is used by the shared L3. Interesting to see just how much chip overhead is dedicated to optimizing out-of-order x86 performance compared to the space used for actual instruction execution. Core i7 looks to be even better at accelerating code from LAZY programmers: new loop stream detector, better branch prediction, 33% larger out-of-order instruction window, advanced smart cache, and to top it off, fast unaligned vector loads and stores! Core i7 also has full SSE4 support. SSE4 has insert/extract instructions to enable instruction efficient vector scatter/gather (with SSE4 it only takes N instructions to scatter/gather N elements in a vector).