Open Hub
Open Hub

 Settings |  Report Duplicate

0
I Use This!
Inactive

Commits : Listings

Analyzed about 21 hours ago. based on code collected about 21 hours ago.
Sep 29, 2024 — Sep 29, 2025
Commit Message Contributor Files Modified Lines Added Lines Removed Code Location Date
Added state tracking number of mark cycles initiated and completed in each RROF full cycle.
pnkfelix
More... almost 15 years ago
Fixed summarization scheduling for F3 > 1.
pnkfelix
More... almost 15 years ago
A bit more instrumentation and internal state tracking to better understand summarization scheduling.
pnkfelix
More... almost 15 years ago
Exposes bug in summarization policy, perhaps due to change in coverage calcuation.
pnkfelix
More... almost 15 years ago
An alternate computation of summarization coverage. I do not fully trust it yet as 5nboyer:5 exposes problems with it.
pnkfelix
More... almost 15 years ago
Investigating summarization scheduling failure when running perm 200 with following parameter settings:
pnkfelix
More... almost 15 years ago
Make summarization scans able to kill remset entries when it sees that an object no longer has region-crossing references into collectable (ie non-infamous, non-static) regions.
pnkfelix
More... almost 15 years ago
Turning off synchronization of mark cycles and full cycles in order to get benefits of mark refinement more often.
pnkfelix
More... almost 15 years ago
Experiment to reduce worst-case pause times: only do mark steps (smircy_step) on minor collections.
pnkfelix
More... almost 15 years ago
Bug fix in format string for message.
pnkfelix
More... almost 15 years ago
Bug fix: this state was uninitialized at summ_matrix creation time, and that was (in some rare cases) leading to global inconsistencies.
pnkfelix
More... almost 15 years ago
Took out an assertion was premature, and turned it into a (disabled) check and note about the whole affair.
pnkfelix
More... almost 15 years ago
Fix for missing values that I refer to in an assertion below.
pnkfelix
More... almost 15 years ago
Added some checks that seem obvious but I am worried that other assertions related to this bit have been failing intermittently.
pnkfelix
More... almost 15 years ago
Cut down on iterations of queue and pqueue benchmarks as this is "just" a smoke test.
pnkfelix
More... almost 15 years ago
Ensure that the granulariy of construction does not exceed the upper bound implied by a (1/(F_1 * F_2))*N/R budget and max F_3 passes.
pnkfelix
More... almost 15 years ago
Extended summ_matrix constructor to pass along F_1 F_2 and F_3 parameters directly. I had avoided doing this because I had thought it would be simpler to use the derivative values that had "more meaningful" names, but at this point it is easier in some cases to have direct access to what the original values were (e.g. I just found a useful derived parameter that is simply the product of the three F_i).
pnkfelix
More... almost 15 years ago
1. Fixed a couple places where I mixed up the F1:F2::budget:coverage correspondence.
pnkfelix
More... almost 15 years ago
The benchmark files are named queue3 and pueue3, not queue and pueue.
pnkfelix
More... almost 15 years ago
I have run into assertion failures recently, so I made an explicit check with an error message for the two cases.
pnkfelix
More... almost 15 years ago
This is the code for (at least some) remset filtering for infamous regions as well as remset rebuilding for advertised regions.
pnkfelix
More... almost 15 years ago
Some fixes so that summ_cols for pop regions have their word counts zeroed before ever being reclassified as filled regions.
pnkfelix
More... almost 15 years ago
Oops I had forgotten to implement this method.
pnkfelix
More... almost 15 years ago
Refactored code since I already had this helper method handy.
pnkfelix
More... almost 15 years ago
Split hasbeen state into two separate groups to accomodate a handoff between summ_matrix and smircy when reconstructing a particular region's remembered set. (Note that such reconstruction has not been implemented and is not yet necessary since pointers into infamous regions still are not filtered from the remembered set.)
pnkfelix
More... almost 15 years ago
Added supported for -infamy option.
pnkfelix
More... almost 15 years ago
Refactoring incoming words update and added predicated shift to infamous region group.
pnkfelix
More... almost 15 years ago
Removed signature for unused and unimplemented function.
pnkfelix
More... almost 15 years ago
This is approximate implementation of infamy shifts (both in and out). But it is for all practical purposes off (on purpose); the default infamy factor is unreachably large (I think) for now and I have not added command line option to override it yet.
pnkfelix
More... almost 15 years ago
Added queue benchmarks to smoke tests.
pnkfelix
More... almost 15 years ago