[an error occurred while processing this directive]

<h1>General Purpose Computation on GPUs<br>
Spring Semester 2014<br>
</h1>
<hr size=3 noshade>

<p class="normal">
<strong>Kick-off meeting</strong>: (tentative) Mon. April 7th,
                                   16:00&mdash;17:00, LE 305<br>
<strong>No</strong> weekly lecture.<br>
<strong>Instructors</strong>:
<a href="http://hpc.uni-due.de/fogal.html">Thomas Fogal</a>,
<a href="http://hpc.uni-due.de/krueger.html">Jens Kr&uuml;ger</a><br>
<strong>Office</strong>: LE 305<br>
<strong>Phone</strong>: 203 379 1314<br>
<strong>E-Mail</strong>: thomas.fogal at uni-due punkt de<br>
<strong>Office hours</strong>: just stop by (Tom's almost always around); or by
appointment (send an email).
</p>

<h2>News (newest first)</h2>
<ol>
  <li>[17.04.2014] As posted on moodle, milestone 1 is due <em>May 16th</em>!
  <li>[24.01.2014] Course announced!
</ol>

<h2>Course Details</h2>

<p class="normal"><strong>NOTE:</strong> This course will be given in
<i>English</i>!</p>

<p class="normal"> <strong>Course Description</strong>:
<blockquote>
Project-focused course on using accelerators to increase program
performance.
</blockquote>
</p>

<p class="normal">There are no explicit prerequisites, but you should be
comfortable with systems programming.  A basic background in
operating systems would be useful.
You may find
<a
href="http://www.lsf.uni-due.de/lsf/rds?state=verpublish&status=init&vmfile=no&moduleCall=webInfo&publishConfFile=webInfo&publishSubDir=veranstaltung&veranstaltung.veranstid=211311">Scientific
Visualization</a> to be useful.
Please talk
to the instructor if you have any doubts about your readiness.
</p>

<p class="normal">
We may include or additionally offer a crash course in C, if there is
enough interest.
</p>

<p class="normal"><strong>If you plan to take the course</strong>,
please register
on <a href="https://moodle2.uni-due.de/course/view.php?id=3678">the
courses' moodle page</a>. The registration code is just 'gpgpu' (all
lowercase).</p>

<h2>GPGPU</h2>

<p class="normal">
<em>G</em>eneral <em>P</em>urpose computation on <em>G</em>raphics
<em>P</em>rocessing <em>U</em>nits refers to using a GPU to accelerate
the computation of a non-graphical task.  A GPU exposes a highly
<a href="https://en.wikipedia.org/wiki/Data_parallelism">data
parallel</a> model, which allows one to specify what operations to
perform on data at a very fine-grained level.  Where and how these
operations get executed is abstracted out by the runtime.  In some
ways, programming a GPU is how one would program a CPU if that CPU had
5000 cores: it no longer makes sense for us to manually dole out tasks
to this many processing elements; instead, we let a program transform
program fragments&mdash;'kernels', in GPU parlance&mdash;to do the
appropriate work assignment.
</p>

<p class="normal">
We should note that using the term "GPU" is a slight anachronism.
After people started realizing these architectures were excellent ways
to achieve good performance, companies started producing data parallel
hardware that was not actually a GPU.  NVIDIA's <a
href="http://www.nvidia.com/object/tesla-supercomputing-solutions.html">Tesla</a> and
Intel's <a
href="http://www.intel.com/content/www/us/en/architecture-and-technology/many-integrated-core/intel-many-integrated-core-architecture.html">Many
Integrated Core (MIC)</a> cards are examples of this broadening of the
architecture.  This course will stick to GPUs purely for practical reasons, but
the ideas will carry over into other parallel environments.
</p>

<h2>Assignments and Grading</h2>

<p class="normal">
As this is a Forschungsprojekt, the entirety of the course grade will
be based on your project.  The final project will be due sometime
after the end of lectures, and there will be approximately 3 required
milestones throughout the semester.
</p>

<p class="normal">
<strong>WARNING:</strong> All groups are expected to do their own
work on the programming assignments (and exams for that matter). No
cross-group collaboration is allowed. A general rule to follow is that
you may discuss the programs with other groups at the concept level
but never at the coding level. If you are at all unclear about this
general rule, do not discuss your programs with other students at all.
</p>

<h2>Reading Materials</h2>

<p class="normal">
There are no required textbooks for this course.  However, you may find
<a
href="http://tacc-web.austin.utexas.edu/staff/home/veijkhout/public_html/istc/istc.html">Viktor
Eijkhout's HPC book</a> useful for general information on parallel scientific
computing.
</p>

<p class="normal">
Other resources you may find useful are the documentation for:
<ul>
  <li><a
href="http://www.gnu.org/software/make/manual/html_node/index.html">Make</a>
  <li><a href="http://www.gnu.org/software/gdb/documentation/">GNU Debugger</a>
  <li><a href="http://gcc.gnu.org/onlinedocs/">GNU Compiler</a>
  <li><a href="http://docs.nvidia.com/cuda/cuda-c-programming-guide/">CUDA
  programming guide</a>
  <li><a href="http://www.khronos.org/opencl/">OpenCL specifications</a>
  <li>&quot;Programming Massively Parallel Processors: A Hands On Approach&quot;
  by David Kirk and Wen-mei Hwu
  <li><a
href="http://developer.amd.com/tools-and-sdks/heterogeneous-computing/amd-accelerated-parallel-processing-app-sdk/introductory-tutorial-to-opencl/">Introduction
  to OpenCL</a>
</ul>
</p>

<h2>Computer Accounts</h2>

<p class="normal">
We will provide remote access to a machine which you can run your
code on.  You may also, of course, write your code locally, but your
software <strong>must</strong> run on the provided platform.<br>
Details on how to access the machine are forthcoming.
</p>

[an error occurred while processing this directive]