== Progress ==

* Buildbots - Setting up Chromebook with a buildbot - http://llvm.org/viewvc/llvm-project?view=revision&revision=174421 - http://lab.llvm.org:8011/builders/clang-native-arm-cortex-a15 - 1h25min builds (3x faster than Pandas)

* LAVA - Trying LLVM builds on Arndale - GCC segfaults: - "The bug is not reproducible, so it is likely a hardware or OS problem."

* Distributed Panda - Setting up the boards, getting distcc on them - Speedup is squat (1.9%) because of too many parallel linking - Need to add LD to a group with less resources on Ninja

* Vectorization - Reviewing some patches, discussing implementation details - Vectorization vs. O3 speedup: 2.5x ~ 3.0x (on both ARM and x86_64) - Working on global structure alias analysis

* LLVM - Campaigning for integrates-as to be turned on by defaultfor ARM - ARM seems interested in fixing the eventual bugs - Adding 'armv7l' triple to default to 'cortex-a8' - http://llvm.org/viewvc/llvm-project?view=revision&revision=174466 - Campaining for buildbots to use CMAke/Ninja

* EuroLLVM - Followup meeting, securing funding

== Plan ==

* Change buildbots to not "make clean", try cmake * Try to reduce parallel linking flood, try one more panda * Continue with global structure alias analysis * Try to setup a test-suite buildbot