Author: Armin Rigo <[email protected]> Branch: extradoc Changeset: r5361:49a0972a00d9 Date: 2014-07-17 15:44 +0200 http://bitbucket.org/pypy/extradoc/changeset/49a0972a00d9/
Log: Tweaks diff --git a/talk/ep2014/stm/talk.html b/talk/ep2014/stm/talk.html --- a/talk/ep2014/stm/talk.html +++ b/talk/ep2014/stm/talk.html @@ -383,6 +383,18 @@ <div class="slide" id="part-1-intro-and-current-status"> <h1>Part 1 - Intro and Current Status</h1> </div> +<div class="slide" id="introduction"> +<h1>Introduction</h1> +<ul class="simple"> +<li>PyPy-STM: Software Transactional Memory</li> +<li>On-going research project:<ul> +<li>by Remi Meier and myself</li> +<li>helped by crowdfunding, thanks to all donors</li> +</ul> +</li> +<li>Started as a EuroPython 2011 lightning talk</li> +</ul> +</div> <div class="slide" id="why-is-there-a-gil"> <h1>Why is there a GIL?</h1> <ul class="simple"> @@ -495,10 +507,8 @@ <ul class="simple"> <li>application-level locks still needed...</li> <li>but <em>can be very coarse:</em><ul> -<li>the idea is to make sure, internally, that one transaction -covers the whole time during which the lock was acquired</li> <li>even two big transactions can hopefully run in parallel</li> -<li>even if they both acquire and release the <em>same</em> lock</li> +<li>even if they both <em>acquire and release the same lock</em></li> </ul> </li> </ul> @@ -511,6 +521,7 @@ <div class="slide" id="demo-1"> <h1>Demo 1</h1> <ul class="simple"> +<li>"Twisted apps made parallel out of the box"</li> <li>Bottle web server</li> </ul> </div> @@ -534,6 +545,7 @@ <ul class="simple"> <li>current status:<ul> <li>basics work</li> +<li>best case 25-40% overhead (much better than originally planned)</li> <li>tons of things to improve</li> <li>tons of things to improve</li> <li>tons of things to improve</li> @@ -561,6 +573,10 @@ <li>as long as they do multiple things that are "often independent"</li> </ul> </li> +<li>Keep locks coarse-grained:<ul> +<li>need to track and fix issues in case of systematic conflicts</li> +</ul> +</li> </ul> </div> <div class="slide" id="part-2-under-the-hood"> diff --git a/talk/ep2014/stm/talk.rst b/talk/ep2014/stm/talk.rst --- a/talk/ep2014/stm/talk.rst +++ b/talk/ep2014/stm/talk.rst @@ -17,6 +17,19 @@ --------------------------------- +Introduction +------------ + +* PyPy-STM: Software Transactional Memory + +* On-going research project: + + - by Remi Meier and myself + - helped by crowdfunding, thanks to all donors + +* Started as a EuroPython 2011 lightning talk + + Why is there a GIL? ------------------- @@ -132,12 +145,9 @@ * but *can be very coarse:* - - the idea is to make sure, internally, that one transaction - covers the whole time during which the lock was acquired - - even two big transactions can hopefully run in parallel - - even if they both acquire and release the *same* lock + - even if they both *acquire and release the same lock* Big Point @@ -149,6 +159,8 @@ Demo 1 ------ +* "Twisted apps made parallel out of the box" + * Bottle web server @@ -173,6 +185,7 @@ * current status: - basics work + - best case 25-40% overhead (much better than originally planned) - tons of things to improve - tons of things to improve - tons of things to improve @@ -201,6 +214,9 @@ - as long as they do multiple things that are "often independent" +* Keep locks coarse-grained: + + - need to track and fix issues in case of systematic conflicts Part 2 - Under The Hood _______________________________________________ pypy-commit mailing list [email protected] https://mail.python.org/mailman/listinfo/pypy-commit
