Strange Long ParNew GC Pauses (Sample Code Included)

[Kim Barrett]

On May 14, 2015, at 5:52 PM, Jungwoo Ha <jwha at google.com> wrote:
> 
> We have a faster implementation for card table scan, and on our implementation ~20s pause goes down to 1~2s with above-mentioned experiments. It doesn't solve the problem but it alleviates it (and actually shows that card table scan is faster).
> If someone can sponsor this, I will create a patch for the review.

I would be interested in looking at proposals for improving card table scanning.  It’s an area I’ve been doing some thinking about and some experiments, and would be happy to collaborate or help integrate someone else’s improvements.