On 11/23/2010 12:09 PM, Joern Rennecke wrote: > If we changed BITS_PER_UNIT into an ordinary piece-of-data 'hook', this > would not only cost a data load from the target vector, but would also > inhibit optimizations that replace division / modulo / multiply with shift > or mask operations. > So maybe we should look into having a few functional hooks that do common > operations, i.e. > bits_in_units x / BITS_PER_UNIT > bits_in_units_ceil (x + BITS_PER_UNIT - 1) / BITS_PER_UNIT > bit_unit_remainder x % BITS_PER_UNIT > units_in_bits x * BITS_PER_UNIT
I think actual hooks is perhaps more overhead than plain division. At least on an x86 host, which is 99% of all running hosts. We might be able to do better with log2_bits_per_unit as data, and a compatibility define #define BITS_PER_UNIT (1 << targetm.log2_bits_per_unit) which ought to give the compiler the info it needs to optimize the operations above into shifts. r~
