Commit 3c22dc58 authored by Nathalie Furmento's avatar Nathalie Furmento
Browse files

website: update tutorials/2014-05-PATC

git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/starpu/website@12901 176f6dd6-97d6-42f4-bd05-d3db9ad07c7a
parent c8d50df1
......@@ -12,7 +12,7 @@
<div class="title">
<h1><a href="../../">StarPU</a></h1>
<h2>StarPU Tutorial - Runtime systems for heterogeneous platform programming - Bordeaux, May 2014</h2>
<h2>StarPU Tutorial - Runtime Systems for Heterogeneous Platform Programming - Bordeaux, May 2014</h2>
</div>
<div class="menu">
......@@ -103,12 +103,12 @@ hostnames for the different node classes, as appropriate.
</div>
<div class="section">
<h2>Session part 1: Task-based programming model</h2>
<h2>Session Part 1: Task-based Programming Model</h2>
<div class="section">
<h3>Application example: vector scaling</h3>
<h3>Application Example: Vector Scaling</h3>
<h4>Making it and running it</h4>
<h4>Making it and Running it</h4>
<p>
A typical <tt>Makefile</tt> for applications using StarPU is then the
......@@ -141,7 +141,7 @@ scheduler using the <a href="files/vector_scal.pbs">given qsub script</a>. It sh
given factor.
</p>
<h4>Computation kernels</h4>
<h4>Computation Kernels</h4>
<p>
Examine the source code, starting from <tt>vector_scal_cpu.c</tt> : this is
the actual computation code, which is wrapped into a <tt>scal_cpu_func</tt>
......@@ -177,7 +177,7 @@ STARPU_NCPUS=0 vector_scal_task_insert
STARPU_NCPUS=0 STARPU_NCUDA=0 vector_scal_task_insert
</pre></tt>
<h4>Main code</h4>
<h4>Main Code</h4>
<p>
Now examine <tt>vector_scal_task_insert.c</tt>: the <tt>cl</tt>
(codelet) structure simply gathers pointers on the functions
......@@ -202,7 +202,7 @@ to main memory.</li>
</div>
<div class="section">
<h3>Data partitioning</h3>
<h3>Data Partitioning</h3>
<p>
In the previous section, we submitted only one task. We here discuss how to
......@@ -295,7 +295,7 @@ programming model, we get multi-GPU support for free!</p>
<!--
<div class="section">
<h3>More advanced examples</h3>
<h3>More Advanced Examples</h3>
<p>
<tt>examples/lu/xlu_implicit.c</tt> is a more involved example: this is a simple
LU decomposition algorithm. The <tt>dw_codelet_facto_v3</tt> is actually the
......@@ -326,14 +326,14 @@ of tasks</p>
</div>
<div class="section">
<h2>Session part 2: Optimizations</h2>
<h2>Session Part 2: Optimizations</h2>
<!--
<p>This is based on StarPU's documentation
<a href="http://runtime.bordeaux.inria.fr/StarPU/starpu.html#Performance-optimization">optimization chapter</a></p>
<div class="section">
<h3>Data management</h3>
<h3>Data Management</h3>
<p>We have explained how StarPU can overlap computation and data transfers
thanks to DMAs. This is however only possible when CUDA has control over the
......@@ -343,7 +343,7 @@ when allocating its buffer, to permit asynchronous DMAs from and to it.</p>
</div>
<div class="section">
<h3>Task submission</h3>
<h3>Task Submission</h3>
<p>To let StarPU reorder tasks, submit data transfers in advance, etc., task
submission should be asynchronous whenever possible. Ideally, the application
......@@ -353,7 +353,7 @@ whole graph of tasks, and wait for termination.</p>
</div>
<div class="section">
<h3>Task scheduling policy</h3>
<h3>Task Scheduling Policy</h3>
<p>By default, StarPU uses the <tt>eager</tt> simple greedy scheduler. This is
because it provides correct load balance even if the application codelets do not
have performance models: it uses a single central queue, from which workers draw
......@@ -395,7 +395,7 @@ less great performance.</p>
</div>
<div class="section">
<h3>Performance model calibration</h3>
<h3>Performance Model Calibration</h3>
<p>Performance prediction is essential for proper scheduling decisions, the
performance models thus have to be calibrated. This is done automatically by
......@@ -435,7 +435,7 @@ performance model thus has to be recalibrated from start. To do so, use
</div>
<div class="section">
<h2>MPI support</h2>
<h2>Sessions Part 3: MPI Support</h2>
<!--
<p>StarPU provides support for MPI communications. Basically, it provides
......@@ -474,7 +474,7 @@ For any questions regarding StarPU, please contact the StarPU developers mailing
</div>
<div class="section">
<h2>More performance optimizations</h2>
<h2>More Performance Optimizations</h2>
<p>
The StarPU
documentation <a href="http://runtime.bordeaux.inria.fr/StarPU/doc/html/PerformanceFeedback.html">performance
......@@ -483,7 +483,7 @@ reading after this tutorial.
</p>
<div class="section">
<h3>FxT tracing support</h3>
<h3>FxT Tracing Support</h3>
<p>
In addition to online profiling, StarPU provides offline profiling tools,
......
Supports Markdown
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment