[pypy-commit] extradoc extradoc: Copy and adapt the talk from pycon-italy-2014 (thanks antocuni)
arigo
noreply at buildbot.pypy.org
Sat Jul 19 13:57:30 CEST 2014
Author: Armin Rigo <arigo at tunes.org>
Branch: extradoc
Changeset: r5362:3dd14da26db0
Date: 2014-07-19 13:57 +0200
http://bitbucket.org/pypy/extradoc/changeset/3dd14da26db0/
Log: Copy and adapt the talk from pycon-italy-2014 (thanks antocuni)
diff --git a/talk/pycon-italy-2014/Makefile b/talk/ep2014/status/Makefile
copy from talk/pycon-italy-2014/Makefile
copy to talk/ep2014/status/Makefile
--- a/talk/pycon-italy-2014/Makefile
+++ b/talk/ep2014/status/Makefile
@@ -1,10 +1,10 @@
# you can find rst2beamer.py here:
-# http://codespeak.net/svn/user/antocuni/bin/rst2beamer.py
+# https://bitbucket.org/antocuni/env/raw/default/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
+talk.pdf: talk.rst author.latex stylesheet.latex
python `which rst2beamer.py` --stylesheet=stylesheet.latex --documentoptions=14pt talk.rst talk.latex || exit
#/home/antocuni/.virtualenvs/rst2beamer/bin/python `which rst2beamer.py` --stylesheet=stylesheet.latex --documentoptions=14pt talk.rst talk.latex || exit
sed 's/\\date{}/\\input{author.latex}/' -i talk.latex || exit
diff --git a/talk/pycon-italy-2014/author.latex b/talk/ep2014/status/author.latex
copy from talk/pycon-italy-2014/author.latex
copy to talk/ep2014/status/author.latex
--- a/talk/pycon-italy-2014/author.latex
+++ b/talk/ep2014/status/author.latex
@@ -1,8 +1,10 @@
\definecolor{rrblitbackground}{rgb}{0.0, 0.0, 0.0}
-\title[PyPy Status]{PyPy Status\\\small{(no, PyPy is not dead)}}
-\author[antocuni]
-{Antonio Cuni}
+\title[PyPy Status Talk]{PyPy Status Talk\\\small{(no no, PyPy is not dead)}}
+\author[arigo, rguillebert]
+{Armin Rigo, Romain Guillebert\\
+based on a PyCon Italia talk by Antonio Cuni\\
+\includegraphics[width=80px]{../../img/py-web-new.png}}
-\institute{PyCon Cinque}
-\date{May 24, 2014}
+\institute{EuroPython}
+\date{July 22, 2014}
diff --git a/talk/pycon-italy-2014/beamerdefs.txt b/talk/ep2014/status/beamerdefs.txt
copy from talk/pycon-italy-2014/beamerdefs.txt
copy to talk/ep2014/status/beamerdefs.txt
diff --git a/talk/ep2014/status/speed.png b/talk/ep2014/status/speed.png
new file mode 100644
index 0000000000000000000000000000000000000000..33fe20ac9d81ddbd3ced48f52f9717693dc15518
GIT binary patch
[cut]
diff --git a/talk/pycon-italy-2014/stylesheet.latex b/talk/ep2014/status/stylesheet.latex
copy from talk/pycon-italy-2014/stylesheet.latex
copy to talk/ep2014/status/stylesheet.latex
diff --git a/talk/ep2014/status/talk.pdf b/talk/ep2014/status/talk.pdf
new file mode 100644
index 0000000000000000000000000000000000000000..f9532b263b7bf1aa53ba0d807a875af9991c6a39
GIT binary patch
[cut]
diff --git a/talk/pycon-italy-2014/talk.pdf.info b/talk/ep2014/status/talk.pdf.info
copy from talk/pycon-italy-2014/talk.pdf.info
copy to talk/ep2014/status/talk.pdf.info
diff --git a/talk/pycon-italy-2014/talk.rst b/talk/ep2014/status/talk.rst
copy from talk/pycon-italy-2014/talk.rst
copy to talk/ep2014/status/talk.rst
--- a/talk/pycon-italy-2014/talk.rst
+++ b/talk/ep2014/status/talk.rst
@@ -4,18 +4,6 @@
PyPy Status
================================
-About me
----------
-
-- PyPy core dev
-
-- ``pdb++``, ``fancycompleter``, ...
-
-- Consultant, trainer
-
-- http://antocuni.eu
-
-
PyPy is not dead
----------------
@@ -114,6 +102,8 @@
* ~7x faster than standard PHP
+ * comparable speed as HHVM
+
* http://hippyvm.com/
@@ -121,18 +111,27 @@
Fundraising campaign
---------------------
-- py3k: 50'852 $ of 105'000 $ (48.4%)
+- py3k: 52'000 $ of 105'000 $ (50%)
-- numpy: 48'121 $ of 60'000 $ (80.2%)
+- numpy: 48'000 $ of 60'000 $ (80%)
- STM, 1st call: 25'000 $
-- STM, 2nd call: 2'097 $ of 80'000 $ (2.6%)
-
- * more on STM later
+- STM, 2nd call: 3'000 $ of 80'000 $ (4%)
- thank to all donors!
+Commercial support
+------------------
+
+- We offer commercial support for PyPy
+
+- Consultancy and training
+
+- Performance issues for open- or closed-source programs, porting,
+ improving support in parts of the Python or non-Python interpreters,
+ etc.
+
Current status
---------------
@@ -155,11 +154,11 @@
- numpy: in-progress (more later)
-Speed: 6.3x faster than CPython
+Speed: 6.5x faster than CPython
--------------------------------
.. image:: speed.png
- :scale: 47%
+ :scale: 44%
ARM
@@ -252,92 +251,19 @@
* by Armin Rigo and Remi Meier
-STM semantics
--------------
-
-- N threads
-
-- Each thread split into atomic blocks
-
-- Sequential execution in some arbitrary order
-
-- In practice:
-
-- Parallel execution, conflicts solved by STM
-
-
-Unit of execution (1)
----------------------
-
-- Atomic blocks == 1 Python bytecode
-
-- Threads are executed in arbitrary order, but bytecodes are atomic
-
-- ==> Same semantics as GIL
-
-- "and this will solve the GIL problem" (A. Rigo, EuroPython 2011 lighting talk)
-
-Unit of execution (2)
+Current status for STM
----------------------
-- Larger atomic blocks
-
-- ``with atomic:``
-
-- Much easier to use than explicit locks
-
-- Can be hidden by libraries to provide even higher level paradigms
-
- * e.g.: Twisted apps made parallel out of the box
-
-Race conditions
----------------
-
-- They don't magically disappear
-
-- With explicit locks
-
- * ==> BOOM
-
- * you fix bugs by preventing race conditions
-
-- With atomic blocks
-
- * ==> Rollaback
-
- * Performance penalty
-
- * You optimize by preventing race conditions
-
-- Fast&broken vs. Slower&correct
-
-
-Implementation
----------------
-
-- Conflicts detection, commit and rollaback is costly
-
-- Original goal (2011): 2x-5x slower than PyPy without STM
-
- * But parallelizable!
-
-|pause|
-
-- Current goal (2014): 25% slower than PyPy without STM
-
-- Yes, that's 10x less overhead than original goal
-
-- mmap black magic
-
-Current status
----------------
-
- Preliminary versions of pypy-jit-stm available
-- The JIT overhead is still a bit too high
+- The overhead is still a bit too high and hard to precict
- Lots of polishing needed
+- More fundamentally, how to best use it is still unknown
+
+- See talk tomorrow
+
Contacts, Q&A
--------------
@@ -346,13 +272,4 @@
- http://morepypy.blogspot.com/
-- twitter: @antocuni
-
-- Available for consultancy & training:
-
- * http://antocuni.eu
-
- * info at antocuni.eu
-
- Any question?
-
More information about the pypy-commit
mailing list