Tamarin:WeeklyUpdates: Difference between revisions

From MozillaWiki
Jump to navigation Jump to search
Line 10: Line 10:
* join irc.mozilla.org #tamarin for attendence taking and questions
* join irc.mozilla.org #tamarin for attendence taking and questions


= 18th March 2008 =
= 25th March 2008 =


== Attendees ==
== Attendees ==


* Adobe: Edwin, Jennifer, Tom, Scott, Brent, Erik, Rick, Jeff
* Adobe: Edwin, Jennifer, Scott, Erik, Rick, Dan
* Mozilla: Graydon
* Mozilla: Graydon, Benjamin, Mark, Chris
* Intel: Moh, Shengnan, Marsha
* Intel: Moh, Shengnan, Marsha, Karthik, Mingqui Sun


== Action Items ==
== Action Items ==
* '''ACTION:''' moh will look at GCbench test and suggest some optimizations.
 
* '''ACTION:''' tom will see if he can find additional GC/memset tests to recommend to moh and post to test directory.  maybe GCheap?
* '''ACTION:''' Edwin/Scott/or Jennifer to send out call for action items each Monday- these will be posted on the wiki
* '''ACTION:''' graydon will investigate Google "Summer of Code" agenda
* '''ACTION:''' Benjamin needs tom to review bug 432692
* '''ACTION:''' Graydon: edwin wants to raise the priority of fixing string implementation -- The TT string optimization bug is https://bugzilla.mozilla.org/show_bug.cgi?id=411163
* '''ACTION:''' Graydon: Per "callgrind" (call-graph profiler derived from "valgrind") - as hotspots are found, log bugs for them with profile data
 
 
== Previous Action Items ==
* '''ACTION:''' moh will look at GCbench test and suggest some optimizations. -- looked at memset and memcopy; optimizations are already in visual studio.  will look at other platforms.
* '''ACTION:''' tom will see if he can find additional GC/memset tests to recommend to moh and post to test directory.  maybe GCheap?  
* '''ACTION:''' graydon will investigate Google "Summer of Code" agenda -- we're welcome to propose items to present.
 


== Updates ==
== Updates ==
''''Edwin''''
* working on trace trees
** main strategy is "pruning" so trees don't get so "bushy"
* how do people feel about meeting twice a month, instead of every week?
** moh suggests it would be better to keep weekly meeting on the calendars
** edwin, scott, or jennifer will send out a call for agenda items on mondays
*** if there are hot topics, we'll meet. if not, we won't.
''''Benjamin''''
* ACTION: needs tom to review bug 432692
'''Moh (Vtune)'''
'''Moh (Vtune)'''
* hot loops investigation:
* vtune patch for central has been checked in
** testsinglebyte function in stringobject.cpp creates a loop that is causing slow-down
 
** shingha has submitted patch for this, which needs to be approved.  looks like performance has improved by 50% already.
** instructions are posted on tamarin wiki
* rick asks: what are you seeing with memset?
 
** moh can have intel look at places where memset appears to be used a lot.
** per tom, flash player zeroes memory during initialization.  if we can improve zeroing time, it improves pause times.
** '''ACTION:''' moh will look at GCbench test and suggest some optimizations.
** '''ACTION:''' tom will see if he can find additional GC/memset tests to recommend to moh and post to test directory.  maybe GCheap?
** moh saw "fastzero" in tamarin-tracing but hasn't had a chance to investigate yet. 
** great places for memory clearing optimizations:
*** GCheap when zeroing a block
*** GCalloc when zeroing a non-RC object
**** freeitem in GCalloc.cpp: RC objects need to be zeroed manually.  non-RC objects are zeroed automatically by memset. 
* moh asks: on tamarin-central, testsinglebyte slow-down does not appear.  why?
** UTF-8 is used in tamarin-tracing, which causes VM to search variable-length strings.
   
   
'''Jeff'''
 
* still working on ECMAScript 4 spec.
''''Karthik''''
* chris peyer is automating the spidermonkey test running ESC on tamarin-central.
 
** 96% pass rate initially, so looking good!
* been trying to profile sunspider benchmark
** lars is fixing the biggest bugs
 
** eval references would have to be removed to run in AS3, so we're not doing that.
** string-unpack provided an opportunity for optimization -- 4-5% improvement
** chris will post automated tests to buildbot when done.
 
** other benchmarks seem a bit too small for profling (run for less than 1 second)
 
* kathik will send patch to edwin, rick, or tom for review
 
   
   
'''Tom'''
 
* noticed that some mozilla folks are participating in google's "summer of code" -- would be cool to get some tamarin stuff on the agenda.
'''Mark'''
* '''ACTION:''' graydon will investigate
 
* had been preparing for a conference, where discussed the future of mozilla and tamarin.
 
** lots of excitement!
 
** interest in ironmonkey
 
* will be catching up with compiler updates soon
 
   
   
'''Graydon'''
 
* further investigation of mini-tracing
''''Graydon''''
** running acceptance tests and will post patch this afternoon if all goes well
 
** mini-tracing is about 1.5 times faster than doing nothing at allsuperwords is a little slower than mini-tracing but a little faster than doing nothing at all.
* has identified some more hotspots
** need to define how much code is needed to generate how much speed-up. is it worth it?
 
** would be great to have more diverse benchmarks, as it's hard to tell what's noise when performance results are so close.
* submitted patch 424891 this week, which was more effective than all mini-tracing work so farwow!
*** flex apps? 
 
*** a version of flash that loads tamarin-tracing dynamically?
* ACTION: edwin wants to raise the priority of fixing string implementation -- The TT string optimization bug is https://bugzilla.mozilla.org/show_bug.cgi?id=411163
*** if we can get ESC running, we can find testing content from the internet (gmail, etc)
 
*** these are all a lot of work.
* steven is on vacation but interested in tracing                                                                                                                   
*** run ESC on tamarin-tracing -- easier than the aforementioned ideas. 
 
*** JSbench tests
* ACTION: Per"callgrind" (call-graph profiler derived from "valgrind") - as hotspots are found, log bugs for them with profile data
*** set loop threshold higher


= Older meetings =
= Older meetings =

Revision as of 15:54, 26 March 2008

These updates concern Tamarin and related projects only.

Meeting Details

  • 2:00pm Pacific Time (21:00 UTC) on Tuesdays
    • (5PM Eastern US, 11PM Oslo, 6AM (Wed) Seoul, 7AM (Wed) Melbourne)
  • Location: Tel: 866 705 2554 (us), 913 227 1201 (int’l)
    • Passcode: 9140087
  • Duration: 60 minutes
  • join irc.mozilla.org #tamarin for attendence taking and questions

25th March 2008

Attendees

  • Adobe: Edwin, Jennifer, Scott, Erik, Rick, Dan
  • Mozilla: Graydon, Benjamin, Mark, Chris
  • Intel: Moh, Shengnan, Marsha, Karthik, Mingqui Sun

Action Items

  • ACTION: Edwin/Scott/or Jennifer to send out call for action items each Monday- these will be posted on the wiki
  • ACTION: Benjamin needs tom to review bug 432692
  • ACTION: Graydon: edwin wants to raise the priority of fixing string implementation -- The TT string optimization bug is https://bugzilla.mozilla.org/show_bug.cgi?id=411163
  • ACTION: Graydon: Per "callgrind" (call-graph profiler derived from "valgrind") - as hotspots are found, log bugs for them with profile data


Previous Action Items

  • ACTION: moh will look at GCbench test and suggest some optimizations. -- looked at memset and memcopy; optimizations are already in visual studio. will look at other platforms.
  • ACTION: tom will see if he can find additional GC/memset tests to recommend to moh and post to test directory. maybe GCheap?
  • ACTION: graydon will investigate Google "Summer of Code" agenda -- we're welcome to propose items to present.


Updates

'Edwin'

  • working on trace trees
    • main strategy is "pruning" so trees don't get so "bushy"
  • how do people feel about meeting twice a month, instead of every week?
    • moh suggests it would be better to keep weekly meeting on the calendars
    • edwin, scott, or jennifer will send out a call for agenda items on mondays
      • if there are hot topics, we'll meet. if not, we won't.


'Benjamin'

  • ACTION: needs tom to review bug 432692


Moh (Vtune)

  • vtune patch for central has been checked in
    • instructions are posted on tamarin wiki


'Karthik'

  • been trying to profile sunspider benchmark
    • string-unpack provided an opportunity for optimization -- 4-5% improvement
    • other benchmarks seem a bit too small for profling (run for less than 1 second)
  • kathik will send patch to edwin, rick, or tom for review


Mark

  • had been preparing for a conference, where discussed the future of mozilla and tamarin.
    • lots of excitement!
    • interest in ironmonkey
  • will be catching up with compiler updates soon


'Graydon'

  • has identified some more hotspots
  • submitted patch 424891 this week, which was more effective than all mini-tracing work so far. wow!
  • steven is on vacation but interested in tracing
  • ACTION: Per"callgrind" (call-graph profiler derived from "valgrind") - as hotspots are found, log bugs for them with profile data

Older meetings