From c20941a9dbfb99a76dd5aede8637bfea6aae1870 Mon Sep 17 00:00:00 2001 From: Stephannie Jimenez Date: Thu, 20 Jan 2022 15:56:41 -0500 Subject: [PATCH] Transform parallelism md to rst --- spec/design_topics/{parallelism.md => parallelism.rst} | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) rename spec/design_topics/{parallelism.md => parallelism.rst} (84%) diff --git a/spec/design_topics/parallelism.md b/spec/design_topics/parallelism.rst similarity index 84% rename from spec/design_topics/parallelism.md rename to spec/design_topics/parallelism.rst index 38ccb9343..77d06c966 100644 --- a/spec/design_topics/parallelism.md +++ b/spec/design_topics/parallelism.rst @@ -1,4 +1,5 @@ -# Parallelism +Parallelism +=========== Parallelism is mostly, but not completely, an execution or runtime concern rather than an API concern. Execution semantics are out of scope for this API @@ -7,7 +8,7 @@ involves how libraries allow users to exercise control over the parallelism they offer, such as: - Via environment variables. This is the method of choice for BLAS libraries and libraries using OpenMP. -- Via a keyword to individual functions or methods. Examples include the `n_jobs` keyword used in scikit-learn and the `workers` keyword used in SciPy. +- Via a keyword to individual functions or methods. Examples include the ``n_jobs`` keyword used in scikit-learn and the ``workers`` keyword used in SciPy. - Build-time settings to enable a parallel or distributed backend. - Via letting the user set chunk sizes. Dask uses this approach. @@ -19,5 +20,5 @@ coordination of parallelization behavior in a stack of Python libraries are: 2. A common library providing a parallelization layer Option (1) may possibly fit in a future version of this array API standard. -[array-api issue 4](https://github.com/data-apis/array-api/issues/4) contains +`array-api issue 4 `_ contains more detailed discussion on the topic of parallelism. \ No newline at end of file