An earlier performance optimization, called "shared memory spin," could slow shared memory performance on a system that has only one CPU core. The logic has been enhanced to detect a system that has only one CPU core and then does not enable the shared memory spin.