previousupnext

5.5.1 Intrinsic Performance

The intrinsic efficiency of LeapHeap and the native heap may be compared by examining the single-threaded data in the tables presented in section 5.5.4 Performance Under Contention. The uniprocessor machine has LeapHeap just measurably faster, as expected. The multiprocessor machine has the native heap markedly faster, probably due to the unrealistically regular alternation of allocations and frees in the test program, causing extra locked operations in the upper tiers of the LeapHeap bitmaps.