|
150
|
1 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
|
|
|
2 "http://www.w3.org/TR/html4/strict.dtd">
|
|
|
3 <!-- Material used from: HTML 4.01 specs: http://www.w3.org/TR/html401/ -->
|
|
|
4 <html>
|
|
|
5 <head>
|
|
|
6 <META http-equiv="Content-Type" content="text/html; charset=ISO-8859-1">
|
|
|
7 <title>Polly - Polyhedral optimizations for LLVM</title>
|
|
|
8 <link type="text/css" rel="stylesheet" href="menu.css">
|
|
|
9 <link type="text/css" rel="stylesheet" href="content.css">
|
|
|
10 <script src="video-js/video.js" type="text/javascript" charset="utf-8"></script>
|
|
|
11 <script type="text/javascript">
|
|
|
12 VideoJS.setupAllWhenReady();
|
|
|
13 </script>
|
|
|
14
|
|
|
15 <!-- Include the VideoJS Stylesheet -->
|
|
|
16 <link rel="stylesheet" href="video-js/video-js.css" type="text/css" media="screen" title="Video JS">
|
|
|
17 </head>
|
|
|
18 <body>
|
|
|
19 <div id="box">
|
|
|
20 <!--#include virtual="menu.html.incl"-->
|
|
|
21 <div id="content">
|
|
|
22 <!--*********************************************************************-->
|
|
|
23 <h1>Open Projects</h1>
|
|
|
24 <!--*********************************************************************-->
|
|
|
25
|
|
|
26 LLVM Polly keeps here a list of open projects which each of themselves would
|
|
|
27 be a great contribution to Polly. All of these projects are meant to be self
|
|
|
28 contained and should take a newcomer around 3-4 months of work. The projects
|
|
236
|
29 we propose are all suitable as <a
|
|
150
|
30 href="https://developers.google.com/open-source/gsoc/">Google Summer of
|
|
|
31 Code</a> projects. In case you are interested in a Google Summer of code
|
|
|
32 project make sure to reach out via the Polly <a
|
|
|
33 href="https://groups.google.com/group/polly-dev">mailing list</a> early to
|
|
|
34 discuss your project proposal.
|
|
|
35
|
|
|
36 <h3>Integrate Polly with the LLVM vectorizers</h3>
|
|
|
37 Polly is not only a self-contained optimizer, but also provides a powerful
|
|
|
38 dependence and other program analyses. Currently, these analyses are only used
|
|
|
39 for our own optimizations. However, LLVM passes such as the loop vectorizer
|
|
|
40 would clearly benefit from having direct access to the available Polly
|
|
|
41 analyses. In this project, you would define in collaboration with the LLVM
|
|
|
42 community and considering existing dependence analysis interface a new
|
|
|
43 dependence analysis interface for Polly that allows passes to directly query
|
|
|
44 Polly analysis. Even though this project sounds straightforward at a first
|
|
|
45 glance, sorting out how to actually make this happen with the current and
|
|
|
46 the new pass managers, understanding how and when to invalidate the Polly
|
|
|
47 analysis and if dependence information can be computed on-demand make this
|
|
|
48 still a challenging project. If successful, this project may be a great way
|
|
|
49 to bring features of Polly to standard -O3 optimizations.
|
|
|
50
|
|
|
51 <h3>Register tiling to obtain fast BLAS kernels with Polly</h3>
|
|
|
52 Even though Polly is already able to speep up compute kernels significantly,
|
|
|
53 when comparing to the best BLAS routines we still are at least one order of
|
|
|
54 magnitude off. In this project you will investigate what is needed to close
|
|
|
55 this performance gap. Earlier investigations have shown that register tiling
|
|
|
56 is one important piece towards this goal. In combination with good tile size
|
|
|
57 models and some back-end work, this project is shooting to make common blas
|
|
|
58 operations, but also many non-blas kernels competitive with vendor math
|
|
|
59 libraries and outperforming the code icc/gcc currently generate.
|
|
|
60
|
|
|
61 <h3>Polly support for Julia - First steps</h3>
|
|
|
62 <a href="https://julialang.org/">Julia</a> is a new matlab style programming
|
|
|
63 language that provides C like performance for scientific computing. Even
|
|
|
64 though Julia also translates to LLVM-IR, parsing and optimizing Julia code
|
|
|
65 poses new challenges that currently prevent Polly from optimizing Julia
|
|
|
66 code despite the clear need for optimizations such as loop-tiling for Julia.
|
|
|
67 In this project you will -- starting from first proof-of-concept patches --
|
|
|
68 integrate Polly into Julia and ensure that Julia code can benefit from the
|
|
|
69 same high-level loop optimizations as todays C code already does. If time
|
|
|
70 permits, making Polly's recent bound-check elimination logic work in Julia
|
|
|
71 code would allow the optimization of Julia code, even if save out-of-bound
|
|
|
72 checking is used.
|
|
|
73 <h3>Interactive Polyhedral Web Calculator</h3>
|
|
|
74 At the core of Polly we use the isl math library. isl allows us to describe
|
|
|
75 loop transformations with relatively simple higher level operations while
|
|
|
76 still providing the full expressiveness of integer polyhedra. To understand
|
|
|
77 and describe the transformations we are performing it is often very convenient
|
|
|
78 to quickly script example transformations in a scripting language like python.
|
|
|
79 isl already comes with a python binding generator, with
|
|
|
80 pypyjs there is a python interpreter for the web and with emscriptem isl
|
|
|
81 itself can also be compiled to javascript. In this project you combine all
|
|
|
82 these components to obtain an interactive polyhedral web calculator, that uses
|
|
|
83 latest web technology to nicely illustrate the integer polyhedra you obtain.
|
|
|
84 </div>
|
|
|
85 </div>
|
|
|
86 </body>
|
|
|
87 </html>
|
