About the service
I develop high-performance code in Chapel, a language designed specifically for productive scientific computing at scale. If your model, solver, or data processing pipeline is stuck on one core, I will move the calculation to all available cores and then across a whole cluster. My main task is turning ordinary numerical code into a program that keeps thousands of threads busy without forcing you to rewrite everything by hand. I use parallel loops forall, distributed domains and the locale model to spread arrays across nodes, and where the algorithm needs finer control I apply task parallelism through begin, cobegin, and synchronizing variables. Typical jobs I take: porting sequential C or Fortran kernel to Chapel, parallelizing template and matrix operations, building distributed hash tables and graph traversals, and profiling why calculation idles on data movement rather than arithmetic. I measure everything. Before calling work done, I show strong and weak scaling curves so you see how code behaves as cores and problem size grow, and explain where limits come from. Results include clean commented source, a recipe to build under your compiler and launcher, a short scaling and memory report, and instructions to run on your hardware or scheduler. I have written Chapel code for Monte Carlo methods, N-body tasks, computational hydrodynamics steps, and large sparse linear algebra. Correctness first, then speed: I add reproducible tests and check results against a trusted baseline before tuning. Name the node count, connection type, and core that bottlenecks, and I will give honest acceleration and effort estimates.
