Searched in google but could not find it. Maybe somebody knows such a benchmark ? It is clear that something that can be measured by PCM counters but would like to have a tool with the ability to separately measure the latency states of MESIFf as in the current NUMA node and a remote NUMA node. C2Cbench has bugs and is no longer supported. BenchIT may be, but not yet figured out how. Maybe someone has more ideas?
For more complete information about compiler optimizations, see our Optimization Notice.