== Progress == * 64-bits ops in Neon: upstream accepted the patch for gcc-4.9-stage1 * smin-umin: several benchmarks runs necessary to confirm that the generic patch I suspected actually caused the regression. However, it also brings improvements in other benches (office-type). * looked at libasan: very little configuration seems necessary, but running a sample program under qemu leads to asan runtime complains. * internal tasks

== Next == * smin-umin: run benchmarks on trunk to confirm the regression is also present. * vectorizer cost model: handle Richard's feedback. * get more codecs sample codes. * libasan: understand error messages

Christophe.