Java Grande Benchmark False Sharing Results

False sharing was detected on only two benchmarks:

Barrier:
Counter.shared_cont: 6235 (1.78% of writes to this field)
Sync:
CounterClass.shared_cont: 8 (insignificant)

No false sharing was detected on the following benchmarks: ForkJoin, Crypt, LUFact, SOR, Series.
No false sharing was detected on partial traces* for the following benchmarks: SparseMatmult, RayTracer.
Not tested (yet): MolDyn, MonteCarlo.
* I stopped the tracing when I grew tired of waiting for it to finish. The traces were about 5 GB large, so they should still give a good idea of the general behavior of the benchmark.

Benchmarks were run on 4 threads (same as number of CPUs) and false sharing was analyzed with trace windows of size 1000.