From 1834e6c6fd3b56f983a9736516e04d37fb8a957c Mon Sep 17 00:00:00 2001 From: Benson Muite Date: Thu, 6 Jan 2022 16:13:24 +0300 Subject: [PATCH 1/2] Create 06-parallel-profiling.md --- _episodes/06-parallel-profiling.md | 22 ++++++++++++++++++++++ 1 file changed, 22 insertions(+) create mode 100644 _episodes/06-parallel-profiling.md diff --git a/_episodes/06-parallel-profiling.md b/_episodes/06-parallel-profiling.md new file mode 100644 index 0000000..25443d8 --- /dev/null +++ b/_episodes/06-parallel-profiling.md @@ -0,0 +1,22 @@ +--- +title: "Profiling parallel programs" +teaching: 45 +exercises: 10 +questions: +- "How do I profile a program using the message passing interface (MPI)?" +objectives: +- "Profile a program to examine computation and communication bottlenecks" +keypoints: +- "Profiling can help explain how to speed up code." +- "Parallel profiling tools are similar to serial ones, but typicaly add + communication and computation imbalance information" +--- + +## Parallel Profiling and Tracing Tools + +There are a number of portable open source parallel profiling and tracing tools +including +- [HPCToolkit](http://hpctoolkit.org/) +- [Extrae](https://tools.bsc.es/extrae) +- [Scalasca](http://www.scalasca.org/) +- [TAU](https://www.cs.uoregon.edu/research/tau/home.php) From 632b1367ea8c447227e640c213e838a37f794480 Mon Sep 17 00:00:00 2001 From: Benson Muite Date: Thu, 6 Jan 2022 16:15:11 +0300 Subject: [PATCH 2/2] Rename section Using MPI should be one of the main focal points for this lesson. --- _episodes/{05-bonus-mpi-for-pi.md => 05-mpi-for-pi.md} | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) rename _episodes/{05-bonus-mpi-for-pi.md => 05-mpi-for-pi.md} (99%) diff --git a/_episodes/05-bonus-mpi-for-pi.md b/_episodes/05-mpi-for-pi.md similarity index 99% rename from _episodes/05-bonus-mpi-for-pi.md rename to _episodes/05-mpi-for-pi.md index 7b07839..a45a55e 100644 --- a/_episodes/05-bonus-mpi-for-pi.md +++ b/_episodes/05-mpi-for-pi.md @@ -1,5 +1,5 @@ --- -title: "Bonus session: Distributing computations among computers" +title: "Using MPI to distribute computations among computers" teaching: 45 exercises: 10 questions: