Tuesday 11 December 2012

This month's meetings

WorkingGroups/ToolChain/Meetings
<< <  2012 / 12 >  >>
Mon Tue Wed Thu Fri Sat Sun
          1 2
3 4 5 6 7 8 9
10 11 12 13 14 15 16
17 18 19 20 21 22 23
24 25 26 27 28 29 30
31            

Attendees

  • Michael Hope
  • Ramana Radhakrishnan
  • Zhenqiang Chen
  • Christophe Lyon
  • Matthew Gretton-Dann
  • Yvan Roux

Agenda

  • Review action items from last meeting
  • Arndale board
  • 'Popular consumer embedded benchmark' Office
  • FMA
  • Disable peeling
  • AOB

Action Items from this Meeting

  • Matt update meeting template attendees - to include Yvan
  • mgrettondann split LRA blueprint
  • Christophe to update Hot/Cold partitioning bugzilla
  • mgrettondann: benchmark on Hold/Cold partitioning
  • Christophe to look at the loop peeling cost

Action Items from Previous Meeting

  • TODO: mgrettondann split LRA blueprint
  • TODO: Christophe to update Hot/Cold partitioning bugzilla
  • TODO: mgrettondann: benchmark on Hold/Cold partitioning
  • DONE: Michael to log a ticket to improve reporting of benchmarks when the run complete.
  • DONE: Ramana to log SPEC failure into bugzilla. - Fixed upstream
  • INPROGRESS: Christophe to backport bswap16 builtin, except for the testcase which fails in one of our configurations (Thumb1 + hard FP ABI)

Minutes

  • Action Review:
    • Christophe encounter an unexpected failure on the bswap16 backport
  • Arndale board:
    • Insignal have released their kernel
    • Michael will qualify, see if it's stable enough for GCC builds
    • See if its usable for benchmarks
    • Push into lp.net/cbuild/hardware a kernel config and similar
    • Can be used for a hardware pack in the future
    • Else can be used for normal cbuild buliders
  • EEMBC Office
    • Now have
    • Is up on a private branch on linaro-toolchain-benchmarks
    • Pulled in to our standard benchmark run
    • Ghostscript is disabled as it segfaults on run on x86 and ARM
    • Will split Network out of the blueprint and postpone
  • Peeling:
    • Christophe proposed a patch up-stream
    • Had to refine it after up-stream comments
    • On ARM none of vectorizer cost hooks seems to be implemented
    • Ramana: it seems that all the default costs are the best
    • Christophe to look at the loop peeling cost 
  • GCC release process:
    • Yvan had to re-start the process due to a laptop configuration issue.
    • The timing would be good if the build is done before tonight
    • Michael re-spawn the 4.6 armv5 which failed (probably a network issue).

WorkingGroups/ToolChain/Meetings/Archive/2012-12-11 (last modified 2013-08-30 11:49:01)