3 ms·
Not to mention that there are much better libraries for data-parallel programming available for C/C++, plus there are languages designed for data-parallelism (O
by galacticpony 10y ago
Not to mention that there are much better libraries for data-parallel programming available for C/C++, plus there are languages designed for data-parallelism (OpenCL, ISPC) and they're not functional either.
Untold amounts of energy (and dollars) must have been wasted on the "convenience" of writing some program in Python that scales over cluster of MapReduce nodes, without investigating how an in-core solution in a straightforward native implementation would have performed.