|
|
@@ -2,7 +2,7 @@
|
|
|
*
|
|
|
* Copyright (C) 2013-2014,2016-2017 CNRS
|
|
|
* Copyright (C) 2013 Inria
|
|
|
- * Copyright (C) 2013-2014,2017 Université de Bordeaux
|
|
|
+ * Copyright (C) 2013-2014,2017-2018 Université de Bordeaux
|
|
|
* Copyright (C) 2013 Corentin Salingue
|
|
|
*
|
|
|
* StarPU is free software; you can redistribute it and/or modify
|
|
|
@@ -90,4 +90,20 @@ All structures are in \ref API_Out_Of_Core.
|
|
|
|
|
|
\snippet disk_compute.c To be included. You should update doxygen if you see this text.
|
|
|
|
|
|
+\section Performances
|
|
|
+
|
|
|
+Scheduling heuristics for Out-of-core are still relatively experimental. The
|
|
|
+tricky part is that you usually have to find a compromise between privileging
|
|
|
+locality (which avoids back and forth with the disk) and privileging the
|
|
|
+critical path, i.e. taking into account priorities to avoid lack of parallelism
|
|
|
+at the end of the task graph.
|
|
|
+
|
|
|
+It is notably better to avoid defining different priorities to tasks with low
|
|
|
+priority, since that will make the scheduler want to schedule them by levels of
|
|
|
+priority, at the depense of locality.
|
|
|
+
|
|
|
+The scheduling algorithms worth trying are thus <code>dmdar</code> and
|
|
|
+<code>lws</code>, which privilege data locality over priorities. There will be
|
|
|
+work on this area in the coming future.
|
|
|
+
|
|
|
*/
|
