From 72f5c8694dd4197039bb445eec91f20b6b882b4b Mon Sep 17 00:00:00 2001 From: Noam Ross Date: Thu, 30 Jul 2020 10:51:55 -0400 Subject: [PATCH 1/2] Add note about parrallel BLAS to README --- README.md | 1 + 1 file changed, 1 insertion(+) diff --git a/README.md b/README.md index ac2c0ac3..4280b0a9 100644 --- a/README.md +++ b/README.md @@ -35,6 +35,7 @@ image | description | size | pull - Images are now based on Ubuntu LTS releases rather than Debian and system libraries are tied to the Ubuntu version. Images will use the most recent LTS available at the time when the corresponding R version was released. Thus all 4.0.0 images are based on Ubuntu 20.04. - We use the [RStudio Package Manager (RSPM)](https://packagemanager.rstudio.com) as our default CRAN mirror. RSPM serves compiled Linux binaries of R packages and greatly speeds up package installs. +- R is built with a version of OpenBLAS that enables multi-threaded BLAS computations by default. This speeds up many processes but [may result in too much nested parallelism](https://github.com/rocker-org/rocker/issues/412#issuecomment-666415627) when running multiple BLAS-intensive parallel R processes. The [RhpcBLASctl](https://cran.r-project.org/web/packages/RhpcBLASctl/index.html) may be used to control low-level parallelism. - Several images, `r-ver`, and new experimental `ml` and `ml-verse` are now available with CUDA drivers and libraries installed and preconfigured for use on machines with NVIDIA GPUs. These are available under tags `4.X.X-cudaX.X`, with currently only CUDA 10.1 available pre-built. - We no longer support `ADD=` runtime triggers for installing Shiny or similar modules. See below for how to extend images to include custom extensions. - Shiny images, like RStudio images, now manage processes via the [S6](https://github.com/just-containers/s6-overlay) supervisor system allowing containers to manage multiple background processes gracefully. From f6e10e62f110e88c10c8f4cd3d9b0ef23766fcce Mon Sep 17 00:00:00 2001 From: Noam Ross Date: Thu, 30 Jul 2020 12:32:20 -0400 Subject: [PATCH 2/2] built --> set up --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/README.md b/README.md index 4280b0a9..d86008b2 100644 --- a/README.md +++ b/README.md @@ -35,7 +35,7 @@ image | description | size | pull - Images are now based on Ubuntu LTS releases rather than Debian and system libraries are tied to the Ubuntu version. Images will use the most recent LTS available at the time when the corresponding R version was released. Thus all 4.0.0 images are based on Ubuntu 20.04. - We use the [RStudio Package Manager (RSPM)](https://packagemanager.rstudio.com) as our default CRAN mirror. RSPM serves compiled Linux binaries of R packages and greatly speeds up package installs. -- R is built with a version of OpenBLAS that enables multi-threaded BLAS computations by default. This speeds up many processes but [may result in too much nested parallelism](https://github.com/rocker-org/rocker/issues/412#issuecomment-666415627) when running multiple BLAS-intensive parallel R processes. The [RhpcBLASctl](https://cran.r-project.org/web/packages/RhpcBLASctl/index.html) may be used to control low-level parallelism. +- R is set up to use a version of OpenBLAS that enables multi-threaded BLAS computations by default. This speeds up many processes but [may result in too much nested parallelism](https://github.com/rocker-org/rocker/issues/412#issuecomment-666415627) when running multiple BLAS-intensive parallel R processes. The [RhpcBLASctl](https://cran.r-project.org/web/packages/RhpcBLASctl/index.html) may be used to control low-level parallelism. - Several images, `r-ver`, and new experimental `ml` and `ml-verse` are now available with CUDA drivers and libraries installed and preconfigured for use on machines with NVIDIA GPUs. These are available under tags `4.X.X-cudaX.X`, with currently only CUDA 10.1 available pre-built. - We no longer support `ADD=` runtime triggers for installing Shiny or similar modules. See below for how to extend images to include custom extensions. - Shiny images, like RStudio images, now manage processes via the [S6](https://github.com/just-containers/s6-overlay) supervisor system allowing containers to manage multiple background processes gracefully.