From 64d2d3a55e1444d675028e6044dac751c677c561 Mon Sep 17 00:00:00 2001 From: Matthias Maier Date: Fri, 10 May 2019 10:13:13 -0500 Subject: [PATCH] add a short documentation entry --- doc/external-libs/cuda.html | 21 ++++++++++++++++----- 1 file changed, 16 insertions(+), 5 deletions(-) diff --git a/doc/external-libs/cuda.html b/doc/external-libs/cuda.html index 345865b6a9..54605dda39 100644 --- a/doc/external-libs/cuda.html +++ b/doc/external-libs/cuda.html @@ -10,7 +10,7 @@ -

Installing deal.II with CUDA

To compile and run CUDA code, you need a CUDA-enabled GPU, appropriate @@ -45,16 +45,27 @@ -DDEAL_II_CUDA_FLAGS="-arch=sm_60" - -DDEAL_II_CUDA_FLAGS_DEBUG and + -DDEAL_II_CUDA_FLAGS_DEBUG and -DDEAL_II_CUDA_FLAGS_RELEASE are also available if you want a finer control on the CUDA flags. The CUDA compiler and the - CUDA toolkit root directory can be set using - -DDEAL_II_CUDA_COMPILER and + CUDA toolkit root directory can be set using + -DDEAL_II_CUDA_COMPILER and -DDEAL_II_CUDA_TOOLKIT_ROOT_DIR. Finally, the CUDA host compiler is the same as the C++ compiler by default, but can be changed using CUDA flags as well.

+ In order to enable MPI Cuda support (provided that your MPI library + supports it) configure with: +

+
+        -DDEAL_II_WITH_CUDA=ON
+        -DDEAL_II_WITH_MPI=ON
+        -DDEAL_II_MPI_WITH_CUDA_SUPPORT=ON
+

+ This configuration option is currently not autodetected. +

+ - -- 2.39.5