Slow code due to many AbstractVector::check and AbstractShuffle::checkIndexes checks in C2
john.r.rose at oracle.com
Tue May 26 21:17:02 UTC 2020
On May 26, 2020, at 1:30 PM, Paul Sandoz <paul.sandoz at oracle.com> wrote:
> We “just" need to optimize the shuffle. I would expect if we are using constant shuffles that the bounds checks can be completely elided.
That’s pretty much the same as vectorizing array range checks
for scatter/gather ops. If we do that now, we should be able to
adapt the code. Maybe there’s even a refactoring here, so that
the fiddly business of range checking (a specialty of Java) can
be put in only one place, so we just have to get it right once.
More information about the panama-dev