64 bit fixed point

64 bit fixed point

Does anyone have any idea if 32.32 or 48.16 fixed point
math would run efficiently on a 64 bit intel chip?

Anyone know how that would compare to floating point
performance (doubles), in particular with regards to
math operations: sin, cos, sqrt etc?

Does IA64 support 64x64 multiply with intermediate?



2 posts / 0 new
Last post
For more complete information about compiler optimizations, see our Optimization Notice.

The answer may not be the same for SSE2 as for ia64 architectures. If multiplication is an important operation, fixed point seems unlikely to match the performance of floating point. On ia64, integer multiplication is performed by the floating point unit, while, on em64t, integer multiplication does not perform as well as SSE2 double.
Some of your answers should appear in the architectural manuals.

Leave a Comment

Please sign in to add a comment. Not a member? Join today