Hi.
I've been looking at the 4.1 projects page. On the project about hot and cold
partitioning, wouldn't it be possible to optimize cold partitions for size where
function inlining criteria will be the code size and there will be no loop
unrolling, etc.?
Or in other words, can we toggle optimize_size for basic blocks and have a sort of non-uniform optimization throughout the translation unit?
Cheers for 4.0!
Stelios
