How can an application use a multiprocessor effectively ?
Explicit parallelism is by domain decomposition and message-passing. This version of Sweep3D supports MPI message passing libraries as well as a single processor version. Here is a description Here are older benchmarks where I found the starting point.
Cosmological simulation GADGET - 2
Flame graphs are a visualization of profiled software, allowing the most frequent code-paths to be identified quickly and accurately.
Showing changes from previous revision.