RFR(S): 7119908: G1: Cache CSet start region for each worker for subsequent reuse
vitalyd at gmail.com
Mon Dec 12 18:37:01 UTC 2011
Yes makes total sense - I completely missed the n_queues arg in the
allocation (that's what I get for looking at this on my phone :)). Sorry
for the noise.
On Dec 12, 2011 1:20 PM, "John Cuthbertson" <john.cuthbertson at oracle.com>
> Hi Vitaly,
> Thanks for looking at the code. The stamp is a pointer because it is
> actually an array of stamps - a separate stamp for each worker. Setting the
> stamp (for a given worker) at the same time as we cache the starting region
> means that we can determine if the starting region has been set (during the
> current pause) without having to clear the cache at the end of the pause.
> If the stamp for given worker has not been set during the current pause
> then we have to assume that the starting region entry for that worker is
> stale and re-calculate the starting region (setting the stamp and updating
> the cache).
> Does this answer your question?
> Let me know if you any others.
> On 12/12/11 06:27, Vitaly Davidovich wrote:
> Hi John,
> Just curious - why did you make the timestamp a uint* instead of uint?
> On Dec 9, 2011 2:20 PM, "John Cuthbertson" <john.cuthbertson at oracle.com>
> Hi Everyone,
> Can I have a couple of volunteers to review the changes for this CR? The
> webrev can be found at:
> As part of the code review comments for 7112743 (G1: Reduce overhead of
> marking closure during evacuation pauses) it was suggested that instead of
> recalculating the starting heap region for each worker thread, we reuse the
> values calculated during RSet scanning. These changes address that review
> comment. In these changes I maintain a cache that is used and updated by
> G1CollectedHeap::start_cset_region_for_worker(). The first time this
> routine is called by a worker thread during an evacuation pause (currently
> during RSet scanning) the cached value for the worker gets set; when the
> routine is called subsequently the region that was cached for the worker is
> returned. I employ a simple stamp mechanism based upon the number of GCs
> that ensures the validity of the regions in the cache and makes clearing
> the cache unnecessary.
> GC Test suite with a small marking threshold (10%) with and without
> verification; the configuration of SPECjbb used to test 7117423;
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the hotspot-gc-dev