Author: Maciej Fijalkowski <fij...@gmail.com> Branch: extradoc Changeset: r4920:82a0c82fdcf9 Date: 2012-11-16 15:52 +0100 http://bitbucket.org/pypy/extradoc/changeset/82a0c82fdcf9/
Log: merge diff too long, truncating to 2000 out of 21927 lines diff --git a/blog/draft/arm-status-update.rst b/blog/draft/arm-status-update.rst --- a/blog/draft/arm-status-update.rst +++ b/blog/draft/arm-status-update.rst @@ -1,12 +1,17 @@ ARM Backend News Update ======================= -Starting with the good news, we finally merged the ``arm-backend-2`` branch -into PyPy's main development line. As described in previous_ posts_ the main -goal of this branch was to add support for ARM processors to PyPY's JIT. As a -general byproduct PyPy should now do a better job supporting non-x86 +As announced in the previous release notes for the PyPy 2.0 beta the ARM JIT +compiler backend branch ``arm-backend-2`` has been merged into the main +development line and is going to be part of the upcomming release. +As a general byproduct PyPy should now do a better job supporting non-x86 architectures such as ARM and the in-progress support for PPC64. +In this post I want to describe a bit more in detail the current state of +PyPy's ARM support. Currently we provide two versions of PyPy for ARM, one +with and one without the JIT compiler. Currently both target the ARM +soft-float ABI. And in this post I'm going to focus on the JIT version. + On ARM, the JIT requires an ARMv7 processor or newer with a VFP unit targeting the ARM application profile. Although this sounds like a strong restriction, most of the ARM processors used in mobile devices and development boards are @@ -30,10 +35,12 @@ as *softfp*, *soft-float* and *hard-float*. The first two use the core registers to pass floating point arguments and do not make any assumptions about a floating point unit. The first uses a software based -float-implementation, while the second can use a floating point unit. The -latter and incompatible one requires a floating point unit and uses the -coprocessor registers to pass floating arguments to calls. A detailed -comparison can be found `here`_. +float-implementation, while the second can use a floating point unit to +actually perform the floating point operations while still copying the +arguments to and from core registers to perform calls. The latter and +incompatible one requires a floating point unit and uses the coprocessor +registers to pass floating arguments to calls. A detailed comparison can be +found `here`_. At the time we started implementing the float support in the ARM backend of the JIT, the soft-float calling conventions were the most commonly supported ones @@ -62,7 +69,7 @@ We also have some hardware to run the subset of the PyPy test-suite relevant to the ARM-JIT backend and to run the tests suite that tests the translated ARM binaries. The nightly tests are run on a Beagleboard-xM_ and an i.MX53_ -versatile board (kindly provided by Michael Foord), both boards run the ARM port `Ubuntu +versatile board (kindly provided by Michael Foord), both boards run the ARM port of `Ubuntu 12.04 Precise Pangolin`_. The current results for the different builders can be seen on the `PyPy buildbot`_. As can be seen there are still some issues to be fixed, but we are getting there. @@ -94,8 +101,9 @@ a regular basis. A fully emulated approach using QEMU might still be worth trying.* * Improve the tools, i.e. integrate with jitviewer_. -Long term on open topics/projects for ARM: +Long term or open topics/projects for ARM: +* Fully support the hard-float calling convention and have nightly builds and tests. * Review of the generated machine code the JIT generates on ARM to see if the instruction selection makes sense for ARM. * Build a version that runs on Android. diff --git a/blog/draft/numpy-status-update-5.rst b/blog/draft/numpy-status-update-5.rst --- a/blog/draft/numpy-status-update-5.rst +++ b/blog/draft/numpy-status-update-5.rst @@ -7,12 +7,13 @@ and there has been quite a bit of progress on the NumPy front in PyPy in the past two months. Things that happened: -* **complex dtype support** - thanks to matti picus, NumPy on PyPy now supports +* **complex dtype support** - thanks to Matti Picus, NumPy on PyPy now supports complex dtype (only complex128 so far, there is work on the other part) * **big refactoring** - probably the biggest issue we did was finishing a big refactoring that disabled some speedups (notably lazy computation of arrays), but lowered the barrier of implementing cool new features. + XXX: explain better why removing a speedup is a good thing, maybe? * **fancy indexing support** - all fancy indexing tricks should now work, including ``a[b]`` where ``b`` is an array of integers. @@ -30,8 +31,8 @@ * **pickling support for numarray** - hasn't started yet, but next on the list -More importantly, we're getting very close to able to import the python part -of the original numpy with only import modifications and running it's tests. +More importantly, we're getting very close to be able to import the python part +of the original numpy with only import modifications, and running its tests. Most tests will fail at this point, however it'll be a good start for another chapter :-) diff --git a/blog/draft/py3k-status-update-7.rst b/blog/draft/py3k-status-update-7.rst --- a/blog/draft/py3k-status-update-7.rst +++ b/blog/draft/py3k-status-update-7.rst @@ -12,38 +12,38 @@ Python standard library tests. We currently pass 160 out of approximately 355 modules of CPython's standard -test suite, fail 144 and skip apprixmately 51. +test suite, fail 144 and skip approximately 51. Some highlights: -o dictviews (the objects returned by dict.keys/values/items) has been greatly +* dictviews (the objects returned by dict.keys/values/items) has been greatly improved, and now they full support set operators -o a lot of tests has been fixed wrt complex numbers (and in particular the -``__complex__`` method) +* a lot of tests has been fixed wrt complex numbers (and in particular the + ``__complex__`` method) -o _csv has been fixed and now it correctly handles unicode instead of bytes +* _csv has been fixed and now it correctly handles unicode instead of bytes -o more parser fixes, py3k list comprehension semantics; now you can no longer +* more parser fixes, py3k list comprehension semantics; now you can no longer access the list comprehension variable after it finishes -o 2to3'd most of our custom lib_pypy +* 2to3'd most of the lib_pypy modules (pypy's custom standard lib + replacements/additions) -o py3-enabled pyrepl: this means that finally readline works at the command +* py3-enabled pyrepl: this means that finally readline works at the command prompt, as well as builtins.input(). ``pdb`` seems to work, as well as fancycompleter_ to get colorful TAB completions :-) -o py3 round +* py3 round -o further tightening/cleanup of the unicode handling (more usage of -surrogateescape, surrogatepass among other things) +* further tightening/cleanup of the unicode handling (more usage of + surrogateescape, surrogatepass among other things) -o as well as keeping up with some big changes happening on the default branch +* as well as keeping up with some big changes happening on the default branch + and of course various other fixes. -Finally, I would like to thank Amaury Forgeot d'Arc for his significant -contributions; among other things, Amaury recently worked on <all kinds of -stuff listed above> - +Finally, we would like to thank Amaury Forgeot d'Arc for his significant +contributions. cheers, Philip&Antonio @@ -52,3 +52,4 @@ .. _`py3k proposal`: http://pypy.org/py3donate.html .. _`py3k branch`: https://bitbucket.org/pypy/pypy/src/py3k .. _`py3k buildbots`: http://buildbot.pypy.org/summary?branch=py3k +.. _`fancycompleter`: http://pypi.python.org/pypi/fancycompleter diff --git a/talk/fscons2012/GIL.fig b/talk/fscons2012/GIL.fig new file mode 100644 --- /dev/null +++ b/talk/fscons2012/GIL.fig @@ -0,0 +1,25 @@ +#FIG 3.2 Produced by xfig version 3.2.5b +Landscape +Center +Metric +A4 +100.00 +Single +-2 +1200 2 +2 2 0 1 0 9 50 -1 20 0.000 0 0 7 0 0 5 + 1125 1350 2475 1350 2475 1800 1125 1800 1125 1350 +2 2 0 1 0 9 50 -1 20 0.000 0 0 -1 0 0 5 + 3600 1350 4725 1350 4725 1800 3600 1800 3600 1350 +2 2 0 1 0 9 50 -1 20 0.000 0 0 -1 0 0 5 + 5850 1350 7200 1350 7200 1800 5850 1800 5850 1350 +2 2 0 1 0 4 50 -1 20 0.000 0 0 -1 0 0 5 + 2475 1800 3600 1800 3600 2250 2475 2250 2475 1800 +2 2 0 1 0 4 50 -1 20 0.000 0 0 -1 0 0 5 + 4725 1800 5850 1800 5850 2250 4725 2250 4725 1800 +2 1 0 1 0 4 51 -1 20 0.000 0 0 7 1 0 2 + 0 0 1.00 60.00 120.00 + 1125 2025 7425 2025 +2 1 0 1 0 4 51 -1 20 0.000 0 0 7 1 0 2 + 0 0 1.00 60.00 120.00 + 1125 1575 7425 1575 diff --git a/talk/fscons2012/GIL.png b/talk/fscons2012/GIL.png new file mode 100644 index 0000000000000000000000000000000000000000..aba489c1cacb9e419d76cff1b59c4a395c27abb1 GIT binary patch [cut] diff --git a/talk/fscons2012/Makefile b/talk/fscons2012/Makefile new file mode 100644 --- /dev/null +++ b/talk/fscons2012/Makefile @@ -0,0 +1,18 @@ +# you can find rst2beamer.py here: +# http://codespeak.net/svn/user/antocuni/bin/rst2beamer.py + +# WARNING: to work, it needs this patch for docutils +# https://sourceforge.net/tracker/?func=detail&atid=422032&aid=1459707&group_id=38414 + +talk.pdf: talk.rst author.latex title.latex stylesheet.latex + rst2beamer.py --stylesheet=stylesheet.latex --documentoptions=14pt talk.rst talk.latex || exit + sed 's/\\date{}/\\input{author.latex}/' -i talk.latex || exit + sed 's/\\maketitle/\\input{title.latex}/' -i talk.latex || exit + sed 's/\\usepackage\[latin1\]{inputenc}/\\usepackage[utf8]{inputenc}/' -i talk.latex || exit + pdflatex talk.latex || exit + +view: talk.pdf + evince talk.pdf & + +xpdf: talk.pdf + xpdf talk.pdf & diff --git a/talk/fscons2012/STM.fig b/talk/fscons2012/STM.fig new file mode 100644 --- /dev/null +++ b/talk/fscons2012/STM.fig @@ -0,0 +1,27 @@ +#FIG 3.2 Produced by xfig version 3.2.5b +Landscape +Center +Metric +A4 +100.00 +Single +-2 +1200 2 +2 1 0 1 0 4 51 -1 20 0.000 0 0 7 1 0 2 + 0 0 1.00 60.00 120.00 + 1125 1575 7425 1575 +2 2 0 1 0 9 50 -1 20 0.000 0 0 7 0 0 5 + 1125 1350 2475 1350 2475 1800 1125 1800 1125 1350 +2 1 0 1 0 4 51 -1 20 0.000 0 0 7 1 0 2 + 0 0 1.00 60.00 120.00 + 1125 2025 7425 2025 +2 2 0 1 0 4 50 -1 20 0.000 0 0 -1 0 0 5 + 1125 1800 2250 1800 2250 2250 1125 2250 1125 1800 +2 2 0 1 0 9 50 -1 20 0.000 0 0 -1 0 0 5 + 2700 1350 3825 1350 3825 1800 2700 1800 2700 1350 +2 2 0 1 0 4 50 -1 20 0.000 0 0 -1 0 0 5 + 2475 1800 3600 1800 3600 2250 2475 2250 2475 1800 +2 2 0 1 0 9 50 -1 20 0.000 0 0 -1 0 0 5 + 4050 1350 5400 1350 5400 1800 4050 1800 4050 1350 +2 2 0 1 0 4 50 -1 20 0.000 0 0 -1 0 0 5 + 3825 1800 4950 1800 4950 2250 3825 2250 3825 1800 diff --git a/talk/fscons2012/STM.png b/talk/fscons2012/STM.png new file mode 100644 index 0000000000000000000000000000000000000000..2de6551346d2db3ad3797f265f2b88a232b387fa GIT binary patch [cut] diff --git a/talk/fscons2012/author.latex b/talk/fscons2012/author.latex new file mode 100644 --- /dev/null +++ b/talk/fscons2012/author.latex @@ -0,0 +1,8 @@ +\definecolor{rrblitbackground}{rgb}{0.0, 0.0, 0.0} + +\title[Software Transactional Memory "for real"]{Software Transactional Memory{ }{ }{ }{ }{ }{ }{ }{ }{ }{ }{ }{ }{ }{ }{ }{ }{ }{ }{ }{ }{ }{ }{ }{ }"for real"} +\author[Armin Rigo] +{Armin Rigo} + +\institute{FSCONS 2012} +\date{November 10, 2012} diff --git a/talk/fscons2012/beamerdefs.txt b/talk/fscons2012/beamerdefs.txt new file mode 100644 --- /dev/null +++ b/talk/fscons2012/beamerdefs.txt @@ -0,0 +1,108 @@ +.. colors +.. =========================== + +.. role:: green +.. role:: red + + +.. general useful commands +.. =========================== + +.. |pause| raw:: latex + + \pause + +.. |small| raw:: latex + + {\small + +.. |end_small| raw:: latex + + } + +.. |scriptsize| raw:: latex + + {\scriptsize + +.. |end_scriptsize| raw:: latex + + } + +.. |strike<| raw:: latex + + \sout{ + +.. closed bracket +.. =========================== + +.. |>| raw:: latex + + } + + +.. example block +.. =========================== + +.. |example<| raw:: latex + + \begin{exampleblock}{ + + +.. |end_example| raw:: latex + + \end{exampleblock} + + + +.. alert block +.. =========================== + +.. |alert<| raw:: latex + + \begin{alertblock}{ + + +.. |end_alert| raw:: latex + + \end{alertblock} + + + +.. columns +.. =========================== + +.. |column1| raw:: latex + + \begin{columns} + \begin{column}{0.45\textwidth} + +.. |column2| raw:: latex + + \end{column} + \begin{column}{0.45\textwidth} + + +.. |end_columns| raw:: latex + + \end{column} + \end{columns} + + + +.. |snake| image:: ../../img/py-web-new.png + :scale: 15% + + + +.. nested blocks +.. =========================== + +.. |nested| raw:: latex + + \begin{columns} + \begin{column}{0.85\textwidth} + +.. |end_nested| raw:: latex + + \end{column} + \end{columns} diff --git a/talk/fscons2012/stylesheet.latex b/talk/fscons2012/stylesheet.latex new file mode 100644 --- /dev/null +++ b/talk/fscons2012/stylesheet.latex @@ -0,0 +1,11 @@ +\usetheme{Boadilla} +\usecolortheme{whale} +\setbeamercovered{transparent} +\setbeamertemplate{navigation symbols}{} + +\definecolor{darkgreen}{rgb}{0, 0.5, 0.0} +\newcommand{\docutilsrolegreen}[1]{\color{darkgreen}#1\normalcolor} +\newcommand{\docutilsrolered}[1]{\color{red}#1\normalcolor} + +\newcommand{\green}[1]{\color{darkgreen}#1\normalcolor} +\newcommand{\red}[1]{\color{red}#1\normalcolor} diff --git a/talk/fscons2012/summary.txt b/talk/fscons2012/summary.txt new file mode 100644 --- /dev/null +++ b/talk/fscons2012/summary.txt @@ -0,0 +1,36 @@ + +- multicore machines, manycore machines + +- focus on cases where using more cores is harder than making more processes + + +- using locks +- using stm/htm + +- HTM: Haswell +- STM: read/write barriers, picture + + +- the deeper problem is: you have to use threads +- messy + +- parallel with GC: common programming languages are either fully GC'ed or + not at all + +- what do you get if you run *everything* in STM/HTM? + +- longer transactions, corresponding to larger parts of the program + +- the thread model becomes implicit + +- demo + + +- GIL/STM pictures +- can pretend it is one-core + +- always gives correct results, but maybe too many conflicts + +- "the right side" of the problem, in my opinion: debugging tools etc. + (same as: GC is "the right side": no crashes, but you need occasionally + to understand memory-not-freed problems) diff --git a/talk/fscons2012/talk.latex b/talk/fscons2012/talk.latex new file mode 100644 --- /dev/null +++ b/talk/fscons2012/talk.latex @@ -0,0 +1,722 @@ +\documentclass[14pt]{beamer} +\usepackage{hyperref} +\usepackage{fancyvrb} + +\makeatletter +\def\PY@reset{\let\PY@it=\relax \let\PY@bf=\relax% + \let\PY@ul=\relax \let\PY@tc=\relax% + \let\PY@bc=\relax \let\PY@ff=\relax} +\def\PY@tok#1{\csname PY@tok@#1\endcsname} +\def\PY@toks#1+{\ifx\relax#1\empty\else% + \PY@tok{#1}\expandafter\PY@toks\fi} +\def\PY@do#1{\PY@bc{\PY@tc{\PY@ul{% + \PY@it{\PY@bf{\PY@ff{#1}}}}}}} +\def\PY#1#2{\PY@reset\PY@toks#1+\relax+\PY@do{#2}} + +\def\PY@tok@gd{\def\PY@bc##1{\fcolorbox[rgb]{0.80,0.00,0.00}{1.00,0.80,0.80}{##1}}} +\def\PY@tok@gu{\let\PY@bf=\textbf\def\PY@tc##1{\textcolor[rgb]{0.00,0.20,0.00}{##1}}} +\def\PY@tok@gt{\def\PY@tc##1{\textcolor[rgb]{0.60,0.80,0.40}{##1}}} +\def\PY@tok@gs{\let\PY@bf=\textbf} +\def\PY@tok@gr{\def\PY@tc##1{\textcolor[rgb]{1.00,0.00,0.00}{##1}}} +\def\PY@tok@cm{\let\PY@it=\textit\def\PY@tc##1{\textcolor[rgb]{0.00,0.60,1.00}{##1}}} +\def\PY@tok@vg{\def\PY@tc##1{\textcolor[rgb]{0.00,0.20,0.20}{##1}}} +\def\PY@tok@m{\def\PY@tc##1{\textcolor[rgb]{1.00,0.40,0.00}{##1}}} +\def\PY@tok@mh{\def\PY@tc##1{\textcolor[rgb]{1.00,0.40,0.00}{##1}}} +\def\PY@tok@cs{\let\PY@bf=\textbf\let\PY@it=\textit\def\PY@tc##1{\textcolor[rgb]{0.00,0.60,1.00}{##1}}} +\def\PY@tok@ge{\let\PY@it=\textit} +\def\PY@tok@vc{\def\PY@tc##1{\textcolor[rgb]{0.00,0.20,0.20}{##1}}} +\def\PY@tok@il{\def\PY@tc##1{\textcolor[rgb]{1.00,0.40,0.00}{##1}}} +\def\PY@tok@go{\def\PY@tc##1{\textcolor[rgb]{0.67,0.67,0.67}{##1}}} +\def\PY@tok@cp{\def\PY@tc##1{\textcolor[rgb]{0.00,0.60,0.60}{##1}}} +\def\PY@tok@gi{\def\PY@bc##1{\fcolorbox[rgb]{0.00,0.80,0.00}{0.80,1.00,0.80}{##1}}} +\def\PY@tok@gh{\let\PY@bf=\textbf\def\PY@tc##1{\textcolor[rgb]{0.00,0.20,0.00}{##1}}} +\def\PY@tok@ni{\let\PY@bf=\textbf\def\PY@tc##1{\textcolor[rgb]{0.60,0.60,0.60}{##1}}} +\def\PY@tok@nl{\def\PY@tc##1{\textcolor[rgb]{0.60,0.60,1.00}{##1}}} +\def\PY@tok@nn{\let\PY@bf=\textbf\def\PY@tc##1{\textcolor[rgb]{0.00,0.80,1.00}{##1}}} +\def\PY@tok@no{\def\PY@tc##1{\textcolor[rgb]{0.20,0.40,0.00}{##1}}} +\def\PY@tok@na{\def\PY@tc##1{\textcolor[rgb]{0.20,0.00,0.60}{##1}}} +\def\PY@tok@nb{\def\PY@tc##1{\textcolor[rgb]{0.20,0.40,0.40}{##1}}} +\def\PY@tok@nc{\let\PY@bf=\textbf\def\PY@tc##1{\textcolor[rgb]{0.00,0.67,0.53}{##1}}} +\def\PY@tok@nd{\def\PY@tc##1{\textcolor[rgb]{0.60,0.60,1.00}{##1}}} +\def\PY@tok@ne{\let\PY@bf=\textbf\def\PY@tc##1{\textcolor[rgb]{0.80,0.00,0.00}{##1}}} +\def\PY@tok@nf{\def\PY@tc##1{\textcolor[rgb]{0.80,0.00,1.00}{##1}}} +\def\PY@tok@si{\def\PY@tc##1{\textcolor[rgb]{0.67,0.00,0.00}{##1}}} +\def\PY@tok@s2{\def\PY@tc##1{\textcolor[rgb]{0.80,0.20,0.00}{##1}}} +\def\PY@tok@vi{\def\PY@tc##1{\textcolor[rgb]{0.00,0.20,0.20}{##1}}} +\def\PY@tok@nt{\let\PY@bf=\textbf\def\PY@tc##1{\textcolor[rgb]{0.20,0.00,0.60}{##1}}} +\def\PY@tok@nv{\def\PY@tc##1{\textcolor[rgb]{0.00,0.20,0.20}{##1}}} +\def\PY@tok@s1{\def\PY@tc##1{\textcolor[rgb]{0.80,0.20,0.00}{##1}}} +\def\PY@tok@gp{\let\PY@bf=\textbf\def\PY@tc##1{\textcolor[rgb]{0.00,0.00,0.60}{##1}}} +\def\PY@tok@sh{\def\PY@tc##1{\textcolor[rgb]{0.80,0.20,0.00}{##1}}} +\def\PY@tok@ow{\let\PY@bf=\textbf\def\PY@tc##1{\textcolor[rgb]{0.00,0.00,0.00}{##1}}} +\def\PY@tok@sx{\def\PY@tc##1{\textcolor[rgb]{0.80,0.20,0.00}{##1}}} +\def\PY@tok@bp{\def\PY@tc##1{\textcolor[rgb]{0.20,0.40,0.40}{##1}}} +\def\PY@tok@c1{\let\PY@it=\textit\def\PY@tc##1{\textcolor[rgb]{0.00,0.60,1.00}{##1}}} +\def\PY@tok@kc{\let\PY@bf=\textbf\def\PY@tc##1{\textcolor[rgb]{0.00,0.40,0.60}{##1}}} +\def\PY@tok@c{\let\PY@it=\textit\def\PY@tc##1{\textcolor[rgb]{0.00,0.60,1.00}{##1}}} +\def\PY@tok@mf{\def\PY@tc##1{\textcolor[rgb]{1.00,0.40,0.00}{##1}}} +\def\PY@tok@err{\def\PY@tc##1{\textcolor[rgb]{0.67,0.00,0.00}{##1}}\def\PY@bc##1{\colorbox[rgb]{1.00,0.67,0.67}{##1}}} +\def\PY@tok@kd{\let\PY@bf=\textbf\def\PY@tc##1{\textcolor[rgb]{0.00,0.40,0.60}{##1}}} +\def\PY@tok@ss{\def\PY@tc##1{\textcolor[rgb]{1.00,0.80,0.20}{##1}}} +\def\PY@tok@sr{\def\PY@tc##1{\textcolor[rgb]{0.20,0.67,0.67}{##1}}} +\def\PY@tok@mo{\def\PY@tc##1{\textcolor[rgb]{1.00,0.40,0.00}{##1}}} +\def\PY@tok@mi{\def\PY@tc##1{\textcolor[rgb]{1.00,0.40,0.00}{##1}}} +\def\PY@tok@kn{\let\PY@bf=\textbf\def\PY@tc##1{\textcolor[rgb]{0.00,0.40,0.60}{##1}}} +\def\PY@tok@o{\def\PY@tc##1{\textcolor[rgb]{0.33,0.33,0.33}{##1}}} +\def\PY@tok@kr{\let\PY@bf=\textbf\def\PY@tc##1{\textcolor[rgb]{0.00,0.40,0.60}{##1}}} +\def\PY@tok@s{\def\PY@tc##1{\textcolor[rgb]{0.80,0.20,0.00}{##1}}} +\def\PY@tok@kp{\def\PY@tc##1{\textcolor[rgb]{0.00,0.40,0.60}{##1}}} +\def\PY@tok@w{\def\PY@tc##1{\textcolor[rgb]{0.73,0.73,0.73}{##1}}} +\def\PY@tok@kt{\let\PY@bf=\textbf\def\PY@tc##1{\textcolor[rgb]{0.00,0.47,0.53}{##1}}} +\def\PY@tok@sc{\def\PY@tc##1{\textcolor[rgb]{0.80,0.20,0.00}{##1}}} +\def\PY@tok@sb{\def\PY@tc##1{\textcolor[rgb]{0.80,0.20,0.00}{##1}}} +\def\PY@tok@k{\let\PY@bf=\textbf\def\PY@tc##1{\textcolor[rgb]{0.00,0.40,0.60}{##1}}} +\def\PY@tok@se{\let\PY@bf=\textbf\def\PY@tc##1{\textcolor[rgb]{0.80,0.20,0.00}{##1}}} +\def\PY@tok@sd{\let\PY@it=\textit\def\PY@tc##1{\textcolor[rgb]{0.80,0.20,0.00}{##1}}} + +\def\PYZbs{\char`\\} +\def\PYZus{\char`\_} +\def\PYZob{\char`\{} +\def\PYZcb{\char`\}} +\def\PYZca{\char`\^} +\def\PYZsh{\char`\#} +\def\PYZpc{\char`\%} +\def\PYZdl{\char`\$} +\def\PYZti{\char`\~} +% for compatibility with earlier versions +\def\PYZat{@} +\def\PYZlb{[} +\def\PYZrb{]} +\makeatother + + +\definecolor{rrblitbackground}{rgb}{0.55, 0.3, 0.1} + +\newenvironment{rtbliteral}{ + +\begin{ttfamily} + +\color{rrblitbackground} + +}{ + +\end{ttfamily} + +} + +% generated by Docutils <http://docutils.sourceforge.net/> +\usepackage{fixltx2e} % LaTeX patches, \textsubscript +\usepackage{cmap} % fix search and cut-and-paste in Acrobat +\usepackage{ifthen} +\usepackage[T1]{fontenc} +\usepackage[utf8]{inputenc} +\usepackage{graphicx} + +%%% Custom LaTeX preamble +% PDF Standard Fonts +\usepackage{mathptmx} % Times +\usepackage[scaled=.90]{helvet} +\usepackage{courier} + +%%% User specified packages and stylesheets +\input{stylesheet.latex} + +%%% Fallback definitions for Docutils-specific commands + +% hyperlinks: +\ifthenelse{\isundefined{\hypersetup}}{ + \usepackage[colorlinks=true,linkcolor=blue,urlcolor=blue]{hyperref} + \urlstyle{same} % normal text font (alternatives: tt, rm, sf) +}{} +\hypersetup{ + pdftitle={Software Transactional Memory ``for real''}, +} + +%%% Title Data +\title{\phantomsection% + Software Transactional Memory ``for real''% + \label{software-transactional-memory-for-real}} +\author{} +\input{author.latex} + +%%% Body +\begin{document} +\input{title.latex} + +% colors + +% =========================== + +% general useful commands + +% =========================== + +% closed bracket + +% =========================== + +% example block + +% =========================== + +% alert block + +% =========================== + +% columns + +% =========================== + +% nested blocks + +% =========================== +\begin{frame} +\frametitle{Introduction} + +% +\begin{itemize} + +\item This talk is about programming multi- or many-core machines + +\end{itemize} +\end{frame} +\begin{frame} +\frametitle{About myself} + +% +\begin{itemize} + +\item Armin Rigo + +\item ``Language implementation guy'' + +\end{itemize} + +\pause +% +\begin{itemize} + +\item PyPy project +% +\begin{itemize} + +\item Python in Python + +\item includes a Just-in-Time Compiler ``Generator'' for Python +and any other dynamic language + +\end{itemize} + +\end{itemize} +\end{frame} +\begin{frame} +\frametitle{Motivation} + +% +\begin{itemize} + +\item A single-core program is getting exponentially slower than a multi-core one + +\end{itemize} + +\pause +% +\begin{itemize} + +\item Using several processes exchanging data +% +\begin{itemize} + +\item works fine in some cases + +\item but becomes a large mess in others + +\end{itemize} + +\end{itemize} + +\pause +% +\begin{itemize} + +\item Using several threads +% +\begin{itemize} + +\item this talk! + +\end{itemize} + +\end{itemize} +\end{frame} +\begin{frame} +\frametitle{Common solution} + +% +\begin{itemize} + +\item Organize your program in multiple threads + +\item Add synchronization when accessing shared, non-read-only data + +\end{itemize} +\end{frame} +\begin{frame} +\frametitle{Synchronization with locks} + +% +\begin{itemize} + +\item Carefully place locks around every access to shared data + +\end{itemize} + +\pause +% +\begin{itemize} + +\item How do you know if you missed a place? +% +\begin{itemize} + +\item hard to catch by writing tests + +\item instead you get obscure rare run-time crashes + +\end{itemize} + +\end{itemize} + +\pause +% +\begin{itemize} + +\item Issues when scaling to a large program +% +\begin{itemize} + +\item order of acquisition + +\item deadlocks + +\end{itemize} + +\end{itemize} +\end{frame} +\begin{frame}[containsverbatim,fragile] +\frametitle{Synchronization with TM} + +% +\begin{itemize} + +\item TM = Transactional Memory + +\end{itemize} + +\pause + +\begin{Verbatim}[commandchars=\\\{\}] +---------------- -------------------- +Locks Transactional Memory +---------------- -------------------- + +mylock.acquire(); atomic \PYZob{} +x = list1.pop(); x = list1.pop(); +list2.append(x); list2.append(x); +mylock.release(); \PYZcb{} +\end{Verbatim} + +\end{frame} +\begin{frame} +\frametitle{Locks versus TM} + +% +\begin{itemize} + +\item Locks + +\end{itemize} + +\noindent\makebox[\textwidth][c]{\includegraphics[scale=0.700000]{withlock.png}} +% +\begin{itemize} + +\item TM + +\end{itemize} + +\noindent\makebox[\textwidth][c]{\includegraphics[scale=0.700000]{withstm0.png}} +\end{frame} +\begin{frame} +\frametitle{Locks versus TM} + +% +\begin{itemize} + +\item Locks + +\end{itemize} + +\noindent\makebox[\textwidth][c]{\includegraphics[scale=0.700000]{withlock.png}} +% +\begin{itemize} + +\item TM in case of conflict + +\end{itemize} + +\noindent\makebox[\textwidth][c]{\includegraphics[scale=0.700000]{withstm.png}} +\end{frame} +\begin{frame} +\frametitle{Synchronization with TM} + +% +\begin{itemize} + +\item ``Optimistic'' approach: +% +\begin{itemize} + +\item no lock to protect shared data in memory + +\item instead, track all memory accesses + +\item detect actual conflicts + +\item if conflict, restart the whole ``transaction'' + +\end{itemize} + +\end{itemize} + +\pause +% +\begin{itemize} + +\item Easier to use +% +\begin{itemize} + +\item no need to name locks + +\item no deadlocks + +\item ``composability'' + +\end{itemize} + +\end{itemize} +\end{frame} +\begin{frame} +\frametitle{HTM versus STM} + +% +\begin{itemize} + +\item HTM = Hardware Transactional Memory +% +\begin{itemize} + +\item Intel Haswell CPU, 2013 + +\item and others + +\end{itemize} + +\end{itemize} + +\pause +% +\begin{itemize} + +\item STM = Software Transactional Memory +% +\begin{itemize} + +\item various approaches + +\item large overhead (2x-10x), but getting faster + +\item experimental in PyPy: read/write barriers, as with GC + +\end{itemize} + +\end{itemize} +\end{frame} +\begin{frame} +\frametitle{The catch} + + +\pause +% +\begin{itemize} + +\item You Still Have To Use Threads + +\end{itemize} + +\pause +% +\begin{itemize} + +\item Threads are hard to debug, non-reproductible + +\end{itemize} + +\pause +% +\begin{itemize} + +\item Threads are Messy + +\end{itemize} +\end{frame} +\begin{frame} +\frametitle{Issue with threads} + +% +\begin{itemize} + +\item TM does not solve this problem: + +\item How do you know if you missed a place to put \texttt{atomic} around? +% +\begin{itemize} + +\item hard to catch by writing tests + +\item instead you get obscure rare run-time crashes + +\end{itemize} + +\end{itemize} + +\pause +% +\begin{itemize} + +\item What if we put \texttt{atomic} everywhere? + +\end{itemize} +\end{frame} +\begin{frame} +\frametitle{Analogy with Garbage Collection} + +% +\begin{itemize} + +\item Explicit Memory Management: +% +\begin{itemize} + +\item messy, hard to debug rare leaks or corruptions + +\end{itemize} + +\end{itemize} + +\pause +% +\begin{itemize} + +\item Automatic GC solves it +% +\begin{itemize} + +\item common languages either have a GC or not + +\item if they have a GC, it controls almost \emph{all} objects + +\item not just a small part of them + +\end{itemize} + +\end{itemize} +\end{frame} +\begin{frame} +\frametitle{Proposed solution} + +% +\begin{itemize} + +\item Put \texttt{atomic} everywhere... + +\item in other words, Run Everything with TM + +\end{itemize} +\end{frame} +\begin{frame} +\frametitle{Proposed solution} + +% +\begin{itemize} + +\item Really needs TM. With locks, you'd get this: + +\end{itemize} + +\noindent\makebox[\textwidth][c]{\includegraphics[scale=0.700000]{GIL.png}} +% +\begin{itemize} + +\item With TM you can get this: + +\end{itemize} + +\noindent\makebox[\textwidth][c]{\includegraphics[scale=0.700000]{STM.png}} +\end{frame} +\begin{frame} +\frametitle{In a few words} + +% +\begin{itemize} + +\item Longer transactions + +\item Corresponding to larger parts of the program + +\item The underlying multi-threaded model becomes implicit + +\end{itemize} +\end{frame} +\begin{frame} +\frametitle{Typical example} + +% +\begin{itemize} + +\item You want to run \texttt{f1()} and \texttt{f2()} and \texttt{f3()} + +\end{itemize} + +\pause +% +\begin{itemize} + +\item Assume they are ``mostly independent'' +% +\begin{itemize} + +\item i.e. we expect that we can run them in parallel + +\item but we cannot prove it, we just hope that in the common case we can + +\end{itemize} + +\end{itemize} + +\pause +% +\begin{itemize} + +\item In case of conflicts, we don't want random behavior +% +\begin{itemize} + +\item i.e. we don't want thread-like non-determinism and crashes + +\end{itemize} + +\end{itemize} +\end{frame} +\begin{frame} +\frametitle{Pooling and atomic statements} + +% +\begin{itemize} + +\item Solution: use a library that creates a pool of threads + +\item Each thread picks a function from the list and runs it +with \texttt{atomic} + +\end{itemize} +\end{frame} +\begin{frame} +\frametitle{Results} + +% +\begin{itemize} + +\item The behavior is ``as if'' we had run \texttt{f1()}, \texttt{f2()} +and \texttt{f3()} sequentially + +\item The programmer chooses if he wants this fixed order, +or if any order is fine + +\item Threads are hidden from the programmer + +\end{itemize} +\end{frame} +\begin{frame} +\frametitle{More generally} + +% +\begin{itemize} + +\item This was an example only + +\item \textbf{TM gives various new ways to hide threads under a nice interface} + +\end{itemize} +\end{frame} +\begin{frame} +\frametitle{Not the Ultimate Solution} + +% +\begin{itemize} + +\item Much easier for the programmer to get reproducible results + +\item But maybe too many conflicts + +\end{itemize} + +\pause +% +\begin{itemize} + +\item ``The right side'' of the problem +% +\begin{itemize} + +\item start with a working program, and improve performance + +\item as opposed to: with locks, start with a fast program, and debug crashes + +\item we will need new debugging tools + +\end{itemize} + +\end{itemize} +\end{frame} +\begin{frame} +\frametitle{PyPy-STM} + +% +\begin{itemize} + +\item PyPy-STM: a version of PyPy with Software Transactional Memory +% +\begin{itemize} + +\item in-progress, but basically working + +\item solves the ``GIL issue'' but more importantly adds \texttt{atomic} + +\end{itemize} + +\item \url{http://pypy.org/} + +\end{itemize} + +\pause +% +\begin{itemize} + +\item Thank you! + +\end{itemize} +\end{frame} + +\end{document} diff --git a/talk/fscons2012/talk.pdf b/talk/fscons2012/talk.pdf new file mode 100644 index 0000000000000000000000000000000000000000..1a6257d12b0ddced11a617033ddb00dbafc7612d GIT binary patch [cut] diff --git a/talk/fscons2012/talk.rst b/talk/fscons2012/talk.rst new file mode 100644 --- /dev/null +++ b/talk/fscons2012/talk.rst @@ -0,0 +1,340 @@ +.. include:: beamerdefs.txt + +============================================ +Software Transactional Memory "for real" +============================================ + + +Introduction +------------------ + +* This talk is about programming multi- or many-core machines + + +About myself +------------------ + +* Armin Rigo + +* "Language implementation guy" + +|pause| + +* PyPy project + + - Python in Python + + - includes a Just-in-Time Compiler "Generator" for Python + and any other dynamic language + + +Motivation +---------------------- + +* A single-core program is getting exponentially slower than a multi-core one + +|pause| + +* Using several processes exchanging data + + - works fine in some cases + + - but becomes a large mess in others + +|pause| + +* Using several threads + + - this talk! + + +Common solution +---------------------- + +* Organize your program in multiple threads + +* Add synchronization when accessing shared, non-read-only data + + +Synchronization with locks +-------------------------- + +* Carefully place locks around every access to shared data + +|pause| + +* How do you know if you missed a place? + + - hard to catch by writing tests + + - instead you get obscure rare run-time crashes + +|pause| + +* Issues when scaling to a large program + + - order of acquisition + + - deadlocks + + +Synchronization with TM +----------------------- + +* TM = Transactional Memory + +|pause| + +.. sourcecode:: plain + + ---------------- -------------------- + Locks Transactional Memory + ---------------- -------------------- + + mylock.acquire(); atomic { + x = list1.pop(); x = list1.pop(); + list2.append(x); list2.append(x); + mylock.release(); } + + +Locks versus TM +--------------- + +* Locks + +.. image:: withlock.png + :scale: 70% + :align: center + +* TM + +.. image:: withstm0.png + :scale: 70% + :align: center + + +Locks versus TM +--------------- + +* Locks + +.. image:: withlock.png + :scale: 70% + :align: center + +* TM in case of conflict + +.. image:: withstm.png + :scale: 70% + :align: center + + +Synchronization with TM +----------------------- + +* "Optimistic" approach: + + - no lock to protect shared data in memory + + - instead, track all memory accesses + + - detect actual conflicts + + - if conflict, restart the whole "transaction" + +|pause| + +* Easier to use + + - no need to name locks + + - no deadlocks + + - "composability" + + +HTM versus STM +-------------- + +* HTM = Hardware Transactional Memory + + - Intel Haswell CPU, 2013 + + - and others + +|pause| + +* STM = Software Transactional Memory + + - various approaches + + - large overhead (2x-10x), but getting faster + + - experimental in PyPy: read/write barriers, as with GC + + +The catch +--------- + +|pause| + +* You Still Have To Use Threads + +|pause| + +* Threads are hard to debug, non-reproductible + +|pause| + +* Threads are Messy + + +Issue with threads +------------------ + +* TM does not solve this problem: + +* How do you know if you missed a place to put ``atomic`` around? + + - hard to catch by writing tests + + - instead you get obscure rare run-time crashes + +|pause| + +* What if we put ``atomic`` everywhere? + + +Analogy with Garbage Collection +------------------------------- + +* Explicit Memory Management: + + - messy, hard to debug rare leaks or corruptions + +|pause| + +* Automatic GC solves it + + - common languages either have a GC or not + + - if they have a GC, it controls almost *all* objects + + - not just a small part of them + + +Proposed solution +----------------- + +* Put ``atomic`` everywhere... + +* in other words, Run Everything with TM + + +Proposed solution +----------------- + +* Really needs TM. With locks, you'd get this: + +.. image:: GIL.png + :scale: 70% + :align: center + +* With TM you can get this: + +.. image:: STM.png + :scale: 70% + :align: center + + +In a few words +-------------- + +* Longer transactions + +* Corresponding to larger parts of the program + +* The underlying multi-threaded model becomes implicit + + +Typical example +--------------- + +* You want to run ``f1()`` and ``f2()`` and ``f3()`` + +|pause| + +* Assume they are "mostly independent" + + - i.e. we expect that we can run them in parallel + + - but we cannot prove it, we just hope that in the common case we can + +|pause| + +* In case of conflicts, we don't want random behavior + + - i.e. we don't want thread-like non-determinism and crashes + + +Pooling and atomic statements +----------------------------- + +* Solution: use a library that creates a pool of threads + +* Each thread picks a function from the list and runs it + with ``atomic`` + + +Results +------- + +* The behavior is "as if" we had run ``f1()``, ``f2()`` + and ``f3()`` sequentially + +* The programmer chooses if he wants this fixed order, + or if any order is fine + +* Threads are hidden from the programmer + + +More generally +-------------- + +* This was an example only + +* **TM gives various new ways to hide threads under a nice interface** + + +Not the Ultimate Solution +------------------------- + +* Much easier for the programmer to get reproducible results + +* But maybe too many conflicts + +|pause| + +* "The right side" of the problem + + - start with a working program, and improve performance + + - as opposed to: with locks, start with a fast program, and debug crashes + + - we will need new debugging tools + + +PyPy-STM +-------- + +* PyPy-STM: a version of PyPy with Software Transactional Memory + + - in-progress, but basically working + + - solves the "GIL issue" but more importantly adds ``atomic`` + +* http://pypy.org/ + +|pause| + +* Thank you! diff --git a/talk/fscons2012/title.latex b/talk/fscons2012/title.latex new file mode 100644 --- /dev/null +++ b/talk/fscons2012/title.latex @@ -0,0 +1,5 @@ +\begin{titlepage} +\begin{figure}[h] +\includegraphics[width=60px]{../../logo/pypy_small128.png} +\end{figure} +\end{titlepage} diff --git a/talk/fscons2012/withlock.fig b/talk/fscons2012/withlock.fig new file mode 100644 --- /dev/null +++ b/talk/fscons2012/withlock.fig @@ -0,0 +1,27 @@ +#FIG 3.2 Produced by xfig version 3.2.5b +Landscape +Center +Metric +A4 +100.00 +Single +-2 +1200 2 +2 2 0 1 0 9 50 -1 20 0.000 0 0 -1 0 0 5 + 3600 1350 4725 1350 4725 1800 3600 1800 3600 1350 +2 2 0 1 0 4 50 -1 20 0.000 0 0 -1 0 0 5 + 4725 1800 5850 1800 5850 2250 4725 2250 4725 1800 +2 1 0 1 0 4 51 -1 20 0.000 0 0 7 1 0 2 + 0 0 1.00 60.00 120.00 + 630 2025 8010 2025 +2 1 0 1 0 4 51 -1 20 0.000 0 0 7 1 0 2 + 0 0 1.00 60.00 120.00 + 630 1575 8010 1575 +2 2 0 1 0 6 50 -1 20 0.000 0 0 7 0 0 5 + 1125 1350 3600 1350 3600 1800 1125 1800 1125 1350 +2 2 0 1 0 6 50 -1 20 0.000 0 0 -1 0 0 5 + 4725 1350 7200 1350 7200 1800 4725 1800 4725 1350 +2 2 0 1 0 6 50 -1 20 0.000 0 0 -1 0 0 5 + 5850 1800 7515 1800 7515 2250 5850 2250 5850 1800 +2 2 0 1 0 6 50 -1 20 0.000 0 0 -1 0 0 5 + 1125 1800 3735 1800 3735 2295 1125 2295 1125 1800 diff --git a/talk/fscons2012/withlock.png b/talk/fscons2012/withlock.png new file mode 100644 index 0000000000000000000000000000000000000000..005e6f0d2dc8168971fa293603656d515687de5e GIT binary patch [cut] diff --git a/talk/fscons2012/withstm.fig b/talk/fscons2012/withstm.fig new file mode 100644 --- /dev/null +++ b/talk/fscons2012/withstm.fig @@ -0,0 +1,29 @@ +#FIG 3.2 Produced by xfig version 3.2.5b +Landscape +Center +Metric +A4 +100.00 +Single +-2 +1200 2 +2 2 0 1 0 6 50 -1 20 0.000 0 0 -1 0 0 5 + 4725 1350 7200 1350 7200 1800 4725 1800 4725 1350 +2 1 0 1 0 4 51 -1 20 0.000 0 0 7 1 0 2 + 0 0 1.00 60.00 120.00 + 630 1575 8010 1575 +2 1 0 1 0 4 51 -1 20 0.000 0 0 7 1 0 2 + 0 0 1.00 60.00 120.00 + 630 2025 8010 2025 +2 2 0 1 0 6 50 -1 20 0.000 0 0 7 0 0 5 + 1125 1350 3600 1350 3600 1800 1125 1800 1125 1350 +2 2 0 1 0 9 50 -1 20 0.000 0 0 -1 0 0 5 + 3600 1350 4725 1350 4725 1800 3600 1800 3600 1350 +2 2 0 1 0 18 50 -1 20 0.000 0 0 -1 0 0 5 + 3735 1800 4860 1800 4860 2295 3735 2295 3735 1800 +2 2 0 1 0 4 50 -1 20 0.000 0 0 -1 0 0 5 + 4275 1800 5400 1800 5400 2295 4275 2295 4275 1800 +2 2 0 1 0 6 50 -1 20 0.000 0 0 -1 0 0 5 + 5400 1800 7065 1800 7065 2295 5400 2295 5400 1800 +2 2 0 1 0 6 50 -1 20 0.000 0 0 -1 0 0 5 + 1125 1800 3735 1800 3735 2295 1125 2295 1125 1800 diff --git a/talk/fscons2012/withstm.png b/talk/fscons2012/withstm.png new file mode 100644 index 0000000000000000000000000000000000000000..f266dceadc96324510e1919d661d81245d24a74c GIT binary patch [cut] diff --git a/talk/fscons2012/withstm0.fig b/talk/fscons2012/withstm0.fig new file mode 100644 --- /dev/null +++ b/talk/fscons2012/withstm0.fig @@ -0,0 +1,27 @@ +#FIG 3.2 Produced by xfig version 3.2.5b +Landscape +Center +Metric +A4 +100.00 +Single +-2 +1200 2 +2 2 0 1 0 6 50 -1 20 0.000 0 0 -1 0 0 5 + 4725 1350 7200 1350 7200 1800 4725 1800 4725 1350 +2 1 0 1 0 4 51 -1 20 0.000 0 0 7 1 0 2 + 0 0 1.00 60.00 120.00 + 630 1575 8010 1575 +2 1 0 1 0 4 51 -1 20 0.000 0 0 7 1 0 2 + 0 0 1.00 60.00 120.00 + 630 2025 8010 2025 +2 2 0 1 0 6 50 -1 20 0.000 0 0 7 0 0 5 + 1125 1350 3600 1350 3600 1800 1125 1800 1125 1350 +2 2 0 1 0 9 50 -1 20 0.000 0 0 -1 0 0 5 + 3600 1350 4725 1350 4725 1800 3600 1800 3600 1350 +2 2 0 1 0 6 50 -1 20 0.000 0 0 -1 0 0 5 + 4860 1800 6525 1800 6525 2295 4860 2295 4860 1800 +2 2 0 1 0 4 50 -1 20 0.000 0 0 -1 0 0 5 + 3735 1800 4860 1800 4860 2295 3735 2295 3735 1800 +2 2 0 1 0 6 50 -1 20 0.000 0 0 -1 0 0 5 + 1125 1800 3735 1800 3735 2295 1125 2295 1125 1800 diff --git a/talk/fscons2012/withstm0.png b/talk/fscons2012/withstm0.png new file mode 100644 index 0000000000000000000000000000000000000000..19fddb01f1ec2fd774299bde2d3995ca59b18eec GIT binary patch [cut] diff --git a/talk/vmil2012/Makefile b/talk/vmil2012/Makefile --- a/talk/vmil2012/Makefile +++ b/talk/vmil2012/Makefile @@ -1,11 +1,23 @@ -jit-guards.pdf: paper.tex paper.bib zotero.bib figures/log.tex figures/example.tex figures/benchmarks_table.tex figures/backend_table.tex figures/ops_count_table.tex figures/loop_bridge.pdf figures/guard_table.tex figures/resume_data_table.tex figures/failing_guards_table.tex +jit-guards.pdf: paper.tex paper.bib zotero.bib figures/log.tex figures/example.tex figures/benchmarks_table.tex figures/backend_table.tex figures/ops_count_table.tex figures/resume_data.pdf figures/loop_bridge.pdf figures/guard_table.tex figures/resume_data_table.tex figures/failing_guards_table.tex pdflatex paper bibtex paper pdflatex paper pdflatex paper mv paper.pdf jit-guards.pdf +print: paper.tex paper.bib zotero.bib figures/log.tex figures/example.tex figures/benchmarks_table.tex figures/backend_table.tex figures/ops_count_table.tex figures/resume_data.eps figures/loop_bridge.eps figures/guard_table.tex figures/resume_data_table.tex figures/failing_guards_table.tex + latex paper + bibtex paper + latex paper + latex paper + dvips -t letter paper + sed -i '' -e "s/paper.dvi/The Efficient Handling of Guards in the Design of RPython's Tracing JIT/g" paper.ps + ps2pdf -sPAPERSIZE=letter paper.ps + mv paper.pdf vmil01-schneider.pdf + mv paper.ps vmil01-schneider.ps + @echo "Or run Adobe Distiller" + UNAME := $(shell "uname") view: jit-guards.pdf ifeq ($(UNAME), Linux) @@ -48,4 +60,6 @@ clean: rm -f *.aux *.bbl *.blg *.log *.tdo rm -f *.pdf + rm -f *.dvi + rm -f *.ps rm -f figures/*table.tex figures/*table.aux diff --git a/talk/vmil2012/figures/loop_bridge.eps b/talk/vmil2012/figures/loop_bridge.eps new file mode 100644 --- /dev/null +++ b/talk/vmil2012/figures/loop_bridge.eps @@ -0,0 +1,15289 @@ +%!PS-Adobe-3.0 EPSF-3.0 +%%HiResBoundingBox: 0.000000 0.000000 407.000000 544.000000 +%APL_DSC_Encoding: UTF8 +%APLProducer: (Version 10.8.2 (Build 12C60) Quartz PS Context) +%%Title: (Unknown) +%%Creator: (Unknown) +%%CreationDate: (Unknown) +%%For: (Unknown) +%%DocumentData: Clean7Bit +%%LanguageLevel: 2 +%%Pages: 1 +%%BoundingBox: 0 0 407 544 +%%EndComments +%%BeginProlog +%%BeginFile: cg-pdf.ps +%%Copyright: Copyright 2000-2004 Apple Computer Incorporated. +%%Copyright: All Rights Reserved. +currentpacking true setpacking +/cg_md 141 dict def +cg_md begin +/L3? languagelevel 3 ge def +/bd{bind def}bind def +/ld{load def}bd +/xs{exch store}bd +/xd{exch def}bd +/cmmtx matrix def +mark +/sc/setcolor +/scs/setcolorspace +/dr/defineresource +/fr/findresource +/T/true +/F/false +/d/setdash +/w/setlinewidth +/J/setlinecap +/j/setlinejoin +/M/setmiterlimit +/i/setflat +/rc/rectclip +/rf/rectfill +/rs/rectstroke +/f/fill +/f*/eofill +/sf/selectfont +/s/show +/xS/xshow +/yS/yshow +/xyS/xyshow +/S/stroke +/m/moveto +/l/lineto +/c/curveto +/h/closepath +/n/newpath +/q/gsave +/Q/grestore +counttomark 2 idiv +{ld}repeat pop +/SC{ + /ColorSpace fr scs +}bd +/sopr /setoverprint where{pop/setoverprint}{/pop}ifelse ld +/soprm /setoverprintmode where{pop/setoverprintmode}{/pop}ifelse ld +/cgmtx matrix def +/sdmtx{cgmtx currentmatrix pop}bd +/CM {cgmtx setmatrix}bd +/cm {cmmtx astore CM concat}bd +/W{clip newpath}bd +/W*{eoclip newpath}bd +statusdict begin product end dup (HP) anchorsearch{ + pop pop pop + true +}{ + pop + (hp) anchorsearch{ + pop pop true + }{ + pop false + }ifelse +}ifelse +{ + { + { + pop pop + (0)dup 0 4 -1 roll put + F charpath + }cshow + } +}{ + {F charpath} +}ifelse +/cply exch bd +/cps {cply stroke}bd +/pgsave 0 def +/bp{/pgsave save store}bd +/ep{pgsave restore showpage}def +/re{4 2 roll m 1 index 0 rlineto 0 exch rlineto neg 0 rlineto h}bd +/scrdict 10 dict def +/scrmtx matrix def +/patarray 0 def +/createpat{patarray 3 1 roll put}bd +/makepat{ +scrmtx astore pop +gsave +initgraphics +CM +patarray exch get +scrmtx +makepattern +grestore +setpattern +}bd +/cg_BeginEPSF{ + userdict save/cg_b4_Inc_state exch put + userdict/cg_endepsf/cg_EndEPSF load put + count userdict/cg_op_count 3 -1 roll put + countdictstack dup array dictstack userdict/cg_dict_array 3 -1 roll put + 3 sub{end}repeat + /showpage {} def + 0 setgray 0 setlinecap 1 setlinewidth 0 setlinejoin + 10 setmiterlimit [] 0 setdash newpath + false setstrokeadjust false setoverprint +}bd +/cg_EndEPSF{ + countdictstack 3 sub { end } repeat + cg_dict_array 3 1 index length 3 sub getinterval + {begin}forall + count userdict/cg_op_count get sub{pop}repeat + userdict/cg_b4_Inc_state get restore + F setpacking +}bd +/cg_biproc{currentfile/RunLengthDecode filter}bd +/cg_aiproc{currentfile/ASCII85Decode filter/RunLengthDecode filter}bd +/ImageDataSource 0 def +L3?{ + /cg_mibiproc{pop pop/ImageDataSource{cg_biproc}def}bd + /cg_miaiproc{pop pop/ImageDataSource{cg_aiproc}def}bd +}{ + /ImageBandMask 0 def + /ImageBandData 0 def + /cg_mibiproc{ + string/ImageBandMask xs + string/ImageBandData xs + /ImageDataSource{[currentfile/RunLengthDecode filter dup ImageBandMask/readstring cvx + /pop cvx dup ImageBandData/readstring cvx/pop cvx]cvx bind}bd + }bd + /cg_miaiproc{ + string/ImageBandMask xs + string/ImageBandData xs + /ImageDataSource{[currentfile/ASCII85Decode filter/RunLengthDecode filter + dup ImageBandMask/readstring cvx + /pop cvx dup ImageBandData/readstring cvx/pop cvx]cvx bind}bd + }bd +}ifelse +/imsave 0 def +/BI{save/imsave xd mark}bd +/EI{imsave restore}bd +/ID{ +counttomark 2 idiv +dup 2 add +dict begin +{def} repeat +pop +/ImageType 1 def +/ImageMatrix[Width 0 0 Height neg 0 Height]def +currentdict dup/ImageMask known{ImageMask}{F}ifelse exch +L3?{ + dup/MaskedImage known + { + pop + << + /ImageType 3 + /InterleaveType 2 + /DataDict currentdict + /MaskDict + << /ImageType 1 + /Width Width + /Height Height + /ImageMatrix ImageMatrix + /BitsPerComponent 1 + /Decode [0 1] + currentdict/Interpolate known + {/Interpolate Interpolate}if + >> + >> + }if +}if +exch +{imagemask}{image}ifelse +end +}bd +/cguidfix{statusdict begin mark version end +{cvr}stopped{cleartomark 0}{exch pop}ifelse +2012 lt{dup findfont dup length dict begin +{1 index/FID ne 2 index/UniqueID ne and +{def} {pop pop} ifelse}forall +currentdict end definefont pop +}{pop}ifelse +}bd +/t_array 0 def +/t_i 0 def +/t_c 1 string def +/x_proc{ + exch t_array t_i get add exch moveto + /t_i t_i 1 add store +}bd +/y_proc{ + t_array t_i get add moveto + /t_i t_i 1 add store +}bd +/xy_proc{ + + t_array t_i 2 copy 1 add get 3 1 roll get + 4 -1 roll add 3 1 roll add moveto + /t_i t_i 2 add store +}bd +/sop 0 def +/cp_proc/x_proc ld +/base_charpath +{ + /t_array xs + /t_i 0 def + { + t_c 0 3 -1 roll put + currentpoint + t_c cply sop + cp_proc + }forall + /t_array 0 def +}bd +/sop/stroke ld +/nop{}def +/xsp/base_charpath ld +/ysp{/cp_proc/y_proc ld base_charpath/cp_proc/x_proc ld}bd +/xysp{/cp_proc/xy_proc ld base_charpath/cp_proc/x_proc ld}bd +/xmp{/sop/nop ld /cp_proc/x_proc ld base_charpath/sop/stroke ld}bd +/ymp{/sop/nop ld /cp_proc/y_proc ld base_charpath/sop/stroke ld}bd +/xymp{/sop/nop ld /cp_proc/xy_proc ld base_charpath/sop/stroke ld}bd +/refnt{ +findfont dup length dict copy dup +/Encoding 4 -1 roll put +definefont pop +}bd +/renmfont{ +findfont dup length dict copy definefont pop +}bd +L3? dup dup{save exch}if +/Range 0 def +/DataSource 0 def +/val 0 def +/nRange 0 def +/mulRange 0 def +/d0 0 def +/r0 0 def +/di 0 def +/ri 0 def +/a0 0 def +/a1 0 def +/r1 0 def +/r2 0 def +/dx 0 def +/Nsteps 0 def +/sh3tp 0 def +/ymax 0 def +/ymin 0 def +/xmax 0 def +/xmin 0 def +/setupFunEval +{ + begin + /nRange Range length 2 idiv store + /mulRange + + [ + 0 1 nRange 1 sub + { + 2 mul/nDim2 xd + Range nDim2 get + Range nDim2 1 add get + 1 index sub + + 255 div + exch + }for + ]store + end +}bd +/FunEval +{ + begin + + nRange mul /val xd + + 0 1 nRange 1 sub + { + dup 2 mul/nDim2 xd + val + add DataSource exch get + mulRange nDim2 get mul + mulRange nDim2 1 add get + add + }for + end +}bd +/max +{ + 2 copy lt + {exch pop}{pop}ifelse +}bd +/sh2 +{ + /Coords load aload pop + 3 index 3 index translate + + 3 -1 roll sub + 3 1 roll exch + sub + 2 copy + dup mul exch dup mul add sqrt + dup + scale _______________________________________________ pypy-commit mailing list pypy-commit@python.org http://mail.python.org/mailman/listinfo/pypy-commit
