yeom [Tue, 6 Oct 2009 17:37:16 +0000 (17:37 +0000)]
changes to handle SESE placeholder correctly
yeom [Tue, 6 Oct 2009 01:23:57 +0000 (01:23 +0000)]
bug fix and changes to handle properly the mapping between heap# in the callee graph and heap# in the caller graph.
jjenista [Mon, 5 Oct 2009 23:10:34 +0000 (23:10 +0000)]
a bug that only allows more aliasing than can be possible--benign, leave for now
yeom [Mon, 5 Oct 2009 18:12:39 +0000 (18:12 +0000)]
*** empty log message ***
yeom [Mon, 5 Oct 2009 17:04:46 +0000 (17:04 +0000)]
bug fix
jihoonl [Sat, 3 Oct 2009 07:49:30 +0000 (07:49 +0000)]
removed unnecessary code
jihoonl [Sat, 3 Oct 2009 07:45:04 +0000 (07:45 +0000)]
modifies getTypeObj()
fix in sortPiles()
yeom [Sat, 3 Oct 2009 06:26:09 +0000 (06:26 +0000)]
add analyzing SESE effects. generate read&write set for each live-in variable.
jjenista [Fri, 2 Oct 2009 18:43:43 +0000 (18:43 +0000)]
trying to make disjointness example versus shape even simpler, still get totally-shared result
jjenista [Fri, 2 Oct 2009 18:37:09 +0000 (18:37 +0000)]
add option to graph visualization that supresses reachability subsets, for improved readability. Algorithm untouched.
bdemsky [Fri, 2 Oct 2009 02:53:57 +0000 (02:53 +0000)]
change
bdemsky [Fri, 2 Oct 2009 00:48:14 +0000 (00:48 +0000)]
fix build
jihoonl [Thu, 1 Oct 2009 18:52:56 +0000 (18:52 +0000)]
to run Task.execute method outside of transaction
bdemsky [Thu, 1 Oct 2009 07:13:26 +0000 (07:13 +0000)]
remove stuff from STMSTATS that was breaking things...
bdemsky [Thu, 1 Oct 2009 06:50:36 +0000 (06:50 +0000)]
fix build options
bdemsky [Thu, 1 Oct 2009 06:46:38 +0000 (06:46 +0000)]
add sandbox support
bdemsky [Thu, 1 Oct 2009 06:46:05 +0000 (06:46 +0000)]
beginning of sandbox support...
#define out overheads for STMSTATS
bdemsky [Thu, 1 Oct 2009 06:44:36 +0000 (06:44 +0000)]
sandbox option
bdemsky [Thu, 1 Oct 2009 06:43:18 +0000 (06:43 +0000)]
change
bdemsky [Thu, 1 Oct 2009 05:21:23 +0000 (05:21 +0000)]
changes for sandboxing
yeom [Wed, 30 Sep 2009 21:40:57 +0000 (21:40 +0000)]
changes to be tainted only if the method or its callees created new edge.
yeom [Wed, 30 Sep 2009 17:41:59 +0000 (17:41 +0000)]
changes
yeom [Tue, 29 Sep 2009 21:36:41 +0000 (21:36 +0000)]
Add a running flag 'methodeffects' for new analysis.
yeom [Mon, 28 Sep 2009 23:58:43 +0000 (23:58 +0000)]
bug fix
adash [Mon, 28 Sep 2009 23:04:44 +0000 (23:04 +0000)]
changes to benchmark
yeom [Mon, 28 Sep 2009 20:58:28 +0000 (20:58 +0000)]
initial commit for maintaining reference edges with taint information.
adash [Sat, 26 Sep 2009 18:38:24 +0000 (18:38 +0000)]
changes to the benchmarks
adash [Sat, 26 Sep 2009 18:32:31 +0000 (18:32 +0000)]
new tree microbenchmark
bdemsky [Sat, 26 Sep 2009 09:07:50 +0000 (09:07 +0000)]
fix accounting of bytes received
bdemsky [Sat, 26 Sep 2009 03:00:13 +0000 (03:00 +0000)]
major bug fix
adash [Fri, 25 Sep 2009 22:05:55 +0000 (22:05 +0000)]
bug fix in gcollect prefetch cache flush
other small changes
yeom [Fri, 25 Sep 2009 21:56:30 +0000 (21:56 +0000)]
fixes to handle non-static method properly.
yeom [Fri, 25 Sep 2009 17:29:56 +0000 (17:29 +0000)]
minor changes
yeom [Fri, 25 Sep 2009 17:08:11 +0000 (17:08 +0000)]
move method effects analysis into disjoint analysis.
methodeffects object maintains analysis result for each method context.
adash [Fri, 25 Sep 2009 08:44:48 +0000 (08:44 +0000)]
changes to makefile
adash [Fri, 25 Sep 2009 08:43:43 +0000 (08:43 +0000)]
new javasingle version
adash [Fri, 25 Sep 2009 08:36:55 +0000 (08:36 +0000)]
changes to copy remote objs read into prefetch cache
print transaction stat counters for remote machines
bdemsky [Fri, 25 Sep 2009 01:45:51 +0000 (01:45 +0000)]
new programs
bdemsky [Fri, 25 Sep 2009 00:23:41 +0000 (00:23 +0000)]
no bug
bdemsky [Fri, 25 Sep 2009 00:14:31 +0000 (00:14 +0000)]
new versionw
jihoonl [Thu, 24 Sep 2009 18:39:36 +0000 (18:39 +0000)]
debug message clear
jihoonl [Thu, 24 Sep 2009 18:33:53 +0000 (18:33 +0000)]
run execute method outside of transaction
jihoonl [Thu, 24 Sep 2009 18:28:37 +0000 (18:28 +0000)]
changes
bdemsky [Thu, 24 Sep 2009 06:25:22 +0000 (06:25 +0000)]
comment out logging
bdemsky [Thu, 24 Sep 2009 06:25:04 +0000 (06:25 +0000)]
changes to support logging
yeom [Thu, 24 Sep 2009 00:22:17 +0000 (00:22 +0000)]
initial commit for method effects analysis.
it captures implementation for last demo.
jjenista [Wed, 23 Sep 2009 23:09:00 +0000 (23:09 +0000)]
new parameter decomposition, can't chain results yet
bdemsky [Wed, 23 Sep 2009 09:36:19 +0000 (09:36 +0000)]
get some more improvements...now at 12x speedup for remote array accesses
bdemsky [Wed, 23 Sep 2009 07:59:07 +0000 (07:59 +0000)]
batch the sends
adash [Wed, 23 Sep 2009 01:20:55 +0000 (01:20 +0000)]
add "count of Prefetch response" into stats
small inline changes that yeild some benefit
A good number to play around is the size of the prefetch
threshold that initiates a garbage collection.
jihoonl [Tue, 22 Sep 2009 21:04:48 +0000 (21:04 +0000)]
saving the working version..
bdemsky [Tue, 22 Sep 2009 20:42:31 +0000 (20:42 +0000)]
add buffered send method
jjenista [Tue, 22 Sep 2009 20:33:27 +0000 (20:33 +0000)]
a common example to compare with shape analyis
bdemsky [Tue, 22 Sep 2009 19:28:35 +0000 (19:28 +0000)]
more bug fixes
bdemsky [Tue, 22 Sep 2009 19:15:11 +0000 (19:15 +0000)]
bug fix
bdemsky [Tue, 22 Sep 2009 11:37:47 +0000 (11:37 +0000)]
almost forgot
bdemsky [Tue, 22 Sep 2009 10:34:04 +0000 (10:34 +0000)]
changes
jihoonl [Tue, 22 Sep 2009 00:19:08 +0000 (00:19 +0000)]
recovery
jihoonl [Tue, 22 Sep 2009 00:14:48 +0000 (00:14 +0000)]
recovery.
Thread.java got a new static start method
jihoonl [Mon, 21 Sep 2009 23:18:38 +0000 (23:18 +0000)]
simple benchmark for recovery
bdemsky [Mon, 21 Sep 2009 22:23:31 +0000 (22:23 +0000)]
change
bdemsky [Mon, 21 Sep 2009 10:16:44 +0000 (10:16 +0000)]
change
jjenista [Fri, 18 Sep 2009 20:01:25 +0000 (20:01 +0000)]
an interface for supporting decomposition of disjointness parameter results
bdemsky [Fri, 18 Sep 2009 04:32:39 +0000 (04:32 +0000)]
fix bug
bdemsky [Fri, 18 Sep 2009 04:32:27 +0000 (04:32 +0000)]
add SESETree
bdemsky [Fri, 18 Sep 2009 03:28:11 +0000 (03:28 +0000)]
only look at things reachable from main
hkhang [Thu, 17 Sep 2009 23:04:22 +0000 (23:04 +0000)]
new dsm web crawler benchmark
jjenista [Thu, 17 Sep 2009 22:08:35 +0000 (22:08 +0000)]
explicit null heap region changes rolled back
bdemsky [Thu, 17 Sep 2009 00:38:10 +0000 (00:38 +0000)]
code to do plots of transaction events..
bdemsky [Thu, 17 Sep 2009 00:04:47 +0000 (00:04 +0000)]
add code to support outputting gnuplot compatible graphs
jjenista [Tue, 15 Sep 2009 22:06:10 +0000 (22:06 +0000)]
allocated regions have class fields initialized to null region, but when I try to add array initialization the same way a completely unrelated assert fails, got to find the problem before null regions are done
bdemsky [Tue, 15 Sep 2009 21:50:20 +0000 (21:50 +0000)]
change
jjenista [Tue, 15 Sep 2009 21:15:46 +0000 (21:15 +0000)]
more null region fixes, but I think method results can still map outgoing edges to the null field
bdemsky [Tue, 15 Sep 2009 19:40:21 +0000 (19:40 +0000)]
simulator now models two locking modes:
lock -- lazy conflict detection
lockcommit -- kill threads as soon as committing thread conflicts with them
jjenista [Tue, 15 Sep 2009 17:57:58 +0000 (17:57 +0000)]
added support for null heap region, doesn't work perfectly because null region gets outgoing edges
bdemsky [Mon, 14 Sep 2009 23:59:22 +0000 (23:59 +0000)]
change
bdemsky [Mon, 14 Sep 2009 23:59:06 +0000 (23:59 +0000)]
add support for alokika's locking scheme to simulator
jjenista [Mon, 14 Sep 2009 22:25:47 +0000 (22:25 +0000)]
fix for x = y.f statements copying by type and ignoring field name
bdemsky [Mon, 14 Sep 2009 08:41:16 +0000 (08:41 +0000)]
fixed some bugs
support more scheduling policies
bdemsky [Mon, 14 Sep 2009 05:12:03 +0000 (05:12 +0000)]
Transaction simulation framework. The idea is to basically give us a way to explore possible benefits of various contention management approaches.
adash [Fri, 11 Sep 2009 22:28:58 +0000 (22:28 +0000)]
added example files for cpu pinning and clock skew detection
adash [Fri, 11 Sep 2009 22:27:44 +0000 (22:27 +0000)]
changes to get clocks ticks, add stmlog option to log results in file
bdemsky [Fri, 11 Sep 2009 05:54:01 +0000 (05:54 +0000)]
fixed
adash [Tue, 8 Sep 2009 23:40:40 +0000 (23:40 +0000)]
bug fix
jjenista [Tue, 8 Sep 2009 22:53:51 +0000 (22:53 +0000)]
obj support in MLP is in good shape, code gets generated but answers wont agree until mem conflict component works
jzhou [Fri, 4 Sep 2009 00:54:38 +0000 (00:54 +0000)]
bug fixes in multicore gc
jjenista [Fri, 4 Sep 2009 00:09:10 +0000 (00:09 +0000)]
Fixed variable dynamic source bookkeeping bug, regression test runs for single and parallelized binaries agree
jzhou [Thu, 3 Sep 2009 00:18:36 +0000 (00:18 +0000)]
bug fixes in multicore gc
bdemsky [Tue, 1 Sep 2009 19:30:48 +0000 (19:30 +0000)]
update email list
adash [Tue, 1 Sep 2009 01:43:50 +0000 (01:43 +0000)]
changes to read clockticks from file
jjenista [Mon, 31 Aug 2009 23:21:49 +0000 (23:21 +0000)]
Variable result table pruning is buggy, remove for now, also adding improved tests. However, for larger test cases inconsistencies appear
adash [Sat, 29 Aug 2009 00:28:28 +0000 (00:28 +0000)]
changes for modifying the microbenchmarks
plot changes
add improved debug statement
measure time lapses using the "timestamp counter" processor register,rdtsc assembly instruction
jzhou [Sat, 29 Aug 2009 00:12:52 +0000 (00:12 +0000)]
bug fixes in multicore gc
jjenista [Fri, 28 Aug 2009 23:11:28 +0000 (23:11 +0000)]
Lots of bug fixes, stable system, parallelized binary computes incorrect answer
jzhou [Fri, 28 Aug 2009 01:35:30 +0000 (01:35 +0000)]
bug fixes in multicore gc
adash [Thu, 27 Aug 2009 23:13:48 +0000 (23:13 +0000)]
check latency of a lock release and lock acquire
jzhou [Thu, 27 Aug 2009 00:26:22 +0000 (00:26 +0000)]
bug fixes in multicore gc
adash [Wed, 26 Aug 2009 01:01:13 +0000 (01:01 +0000)]
estimate thread context switching overhead
jzhou [Wed, 26 Aug 2009 00:10:57 +0000 (00:10 +0000)]
bug fixes in multicore gc
jjenista [Tue, 25 Aug 2009 21:32:41 +0000 (21:32 +0000)]
method call support works for small programs, bigger ones hang