paperKB
coga / coga-kb
Help
Sign in

Chunk #18 — Findings — Improvements in PLINK 1.9 — Multicore and cluster parallelism

Source
Second-generation PLINK: rising to the challenge of larger and richer datasets.
Embedded
yes

Text

Modern x86 processors also contain increasing numbers of cores, and computational workloads in genetic studies tend to contain large “embarrassingly parallel” steps which can easily exploit additional cores. Therefore, PLINK 1.9 autodetects the number of cores present in the machine it is running on, and many of its heavy-duty operations default to employing roughly that number of threads. (This behavior can be manually controlled with the –threads flag.) Most of PLINK 1.9’s multithreaded computations use a simple set of cross-platform C functions and macros, which compile to pthread library idioms on Linux and OS X, and OS-specific idioms like _beginthreadex() on Windows.