IRC.git
11 years agoname pool alloc, useful for seeing its effective reduction of mallocs
jjenista [Thu, 14 Oct 2010 17:49:16 +0000 (17:49 +0000)]
name pool alloc, useful for seeing its effective reduction of mallocs

11 years agocoreprof updates, nothing huge
jjenista [Thu, 14 Oct 2010 17:48:32 +0000 (17:48 +0000)]
coreprof updates, nothing huge

11 years agohash of OIDs should not bit shift
jjenista [Thu, 14 Oct 2010 17:47:51 +0000 (17:47 +0000)]
hash of OIDs should not bit shift

11 years agoFix bug in termination of the multicore gc version w/o tasks
jzhou [Thu, 14 Oct 2010 16:54:08 +0000 (16:54 +0000)]
Fix bug in termination of the multicore gc version w/o tasks

11 years agoAdd the runtime for multicore gc version w/o tasks. Now can start multiple threads...
jzhou [Thu, 14 Oct 2010 01:09:55 +0000 (01:09 +0000)]
Add the runtime for multicore gc version w/o tasks. Now can start multiple threads on Tilera. But still have some problem with the termination schem. Syncronized block is not test yet.

11 years agobug fixes for oooJava, some debug events for coreprof for exploring things not worth...
jjenista [Wed, 13 Oct 2010 23:16:56 +0000 (23:16 +0000)]
bug fixes for oooJava, some debug events for coreprof for exploring things not worth keeping

11 years agojust a convenient script to repeat the official problem size
jjenista [Wed, 13 Oct 2010 23:15:38 +0000 (23:15 +0000)]
just a convenient script to repeat the official problem size

11 years agoadded prefetch when grabbing a record for next one, and fixed a malloc to RUNMALLOC
jjenista [Wed, 13 Oct 2010 23:15:00 +0000 (23:15 +0000)]
added prefetch when grabbing a record for next one, and fixed a malloc to RUNMALLOC

11 years agoupdates for some convenient debug events and a covnert to txt mode
jjenista [Wed, 13 Oct 2010 23:07:35 +0000 (23:07 +0000)]
updates for some convenient debug events and a covnert to txt mode

11 years agochanges...plus a little optimization
bdemsky [Wed, 13 Oct 2010 21:35:18 +0000 (21:35 +0000)]
changes...plus a little optimization

11 years agoa new buildscript option to save teh source code in tmp.c after preprocessor is done...
jjenista [Wed, 13 Oct 2010 18:56:06 +0000 (18:56 +0000)]
a new buildscript option to save teh source code in tmp.c after preprocessor is done iwht it

11 years agochanges
bdemsky [Wed, 13 Oct 2010 03:33:47 +0000 (03:33 +0000)]
changes

11 years agoChanges to accommodate the runtime for multicore gc w/o tasks
jzhou [Wed, 13 Oct 2010 00:17:52 +0000 (00:17 +0000)]
Changes to accommodate the runtime for multicore gc w/o tasks

11 years agomore code towards working...it actually compiles now...doesn't work of course yet
bdemsky [Tue, 12 Oct 2010 23:27:14 +0000 (23:27 +0000)]
more code towards working...it actually compiles now...doesn't work of course yet

11 years agoadd test script for raytracer
yeom [Tue, 12 Oct 2010 23:02:20 +0000 (23:02 +0000)]
add test script for raytracer

11 years agook, this is a wacky commit, purposefully leaving a half-implemented forward Q change...
jjenista [Tue, 12 Oct 2010 22:22:30 +0000 (22:22 +0000)]
ok, this is a wacky commit, purposefully leaving a half-implemented forward Q change in, but benign, because pool allocation by itslef might have fixed ray tracer up very nicely, but gotta check it out first

11 years agomore changes
bdemsky [Tue, 12 Oct 2010 22:06:24 +0000 (22:06 +0000)]
more changes

11 years agoa hack to enable when adding new language features to our compiler
jjenista [Tue, 12 Oct 2010 17:57:17 +0000 (17:57 +0000)]
a hack to enable when adding new language features to our compiler

11 years agobug fix
jjenista [Tue, 12 Oct 2010 17:07:32 +0000 (17:07 +0000)]
bug fix

11 years agotowards getting the compiler to run
bdemsky [Tue, 12 Oct 2010 00:23:59 +0000 (00:23 +0000)]
towards getting the compiler to run

11 years agochanges
bdemsky [Mon, 11 Oct 2010 19:45:48 +0000 (19:45 +0000)]
changes

11 years agotwo methods for generating a reach graph at any desired program point, one is a dummy...
jjenista [Mon, 11 Oct 2010 19:11:16 +0000 (19:11 +0000)]
two methods for generating a reach graph at any desired program point, one is a dummy System method, the other is a dummy FlatNode, turns out dummy FlatNode works fine except that joptimize will optimize it out, could fix that later but its still useful for debugging

11 years agoan example showing empty reach state propagation that we considered for CC submission...
jjenista [Mon, 11 Oct 2010 19:09:04 +0000 (19:09 +0000)]
an example showing empty reach state propagation that we considered for CC submission, and an example of generating reach graphs at a desired program point

11 years agoWe modify benchmarks to run the suite in different ways, just modularized the hefty...
jjenista [Mon, 11 Oct 2010 18:55:27 +0000 (18:55 +0000)]
We modify benchmarks to run the suite in different ways, just modularized the hefty part and only leave source file macros in the local makefiles, should save time and prevent errors when running for paper results

11 years agoAdded rough draft of how a remove from the hashStructure should work. Added method...
stephey [Mon, 11 Oct 2010 09:18:01 +0000 (09:18 +0000)]
Added rough draft of how a remove from the hashStructure should work. Added method signature into the header file.

The remove method needs insight on how to remove read bins. Details are in the comments.

11 years agotest case for rcr
bdemsky [Mon, 11 Oct 2010 07:59:25 +0000 (07:59 +0000)]
test case for rcr

11 years agoclean up printing some more
bdemsky [Mon, 11 Oct 2010 07:58:03 +0000 (07:58 +0000)]
clean up printing some more

11 years agohack OoOJava RCR to not run disjointness twice...we don't need fine-grained conflicts...
bdemsky [Mon, 11 Oct 2010 07:48:19 +0000 (07:48 +0000)]
hack OoOJava RCR to not run disjointness twice...we don't need fine-grained conflicts with RCR...

11 years agoclean up printing formatting a little...make it easier for me to read generated code
bdemsky [Sat, 9 Oct 2010 09:01:04 +0000 (09:01 +0000)]
clean up printing formatting a little...make it easier for me to read generated code

11 years agofix several bugs...
bdemsky [Sat, 9 Oct 2010 08:41:25 +0000 (08:41 +0000)]
fix several bugs...

12 years agochanges.
yeom [Fri, 8 Oct 2010 23:35:43 +0000 (23:35 +0000)]
changes.

12 years agochanges on moldyn.
yeom [Fri, 8 Oct 2010 21:03:37 +0000 (21:03 +0000)]
changes on moldyn.

12 years agoremoving backup version for yong hun
jjenista [Fri, 8 Oct 2010 17:03:48 +0000 (17:03 +0000)]
removing backup version for yong hun

12 years agoMoldyn benchmark.
yeom [Fri, 8 Oct 2010 05:36:13 +0000 (05:36 +0000)]
Moldyn benchmark.

12 years agoFix alias locks. KMeans is now serialized but the execution is stucked somewhere...
jzhou [Fri, 8 Oct 2010 00:56:04 +0000 (00:56 +0000)]
Fix alias locks.  KMeans is now serialized but the execution is stucked somewhere.  Need more investigation and bug fixing

12 years agoRemove debug messages
jzhou [Thu, 7 Oct 2010 22:35:25 +0000 (22:35 +0000)]
Remove debug messages

12 years agoForgot to check in bug fixes of Tracking benchmark
jzhou [Thu, 7 Oct 2010 22:31:56 +0000 (22:31 +0000)]
Forgot to check in bug fixes of Tracking benchmark

12 years agoBug fix for generating shared locks for aliased objects
jzhou [Thu, 7 Oct 2010 22:26:34 +0000 (22:26 +0000)]
Bug fix for generating shared locks for aliased objects

12 years agobug fix
jzhou [Thu, 7 Oct 2010 00:14:43 +0000 (00:14 +0000)]
bug fix

12 years agoupdates
jjenista [Tue, 5 Oct 2010 23:13:22 +0000 (23:13 +0000)]
updates

12 years agotry to get more information about flagged sites.
yeom [Tue, 5 Oct 2010 19:15:39 +0000 (19:15 +0000)]
try to get more information about flagged sites.

12 years agofix bugs to run without pool allocation like the ppopp paper experiments
jjenista [Tue, 5 Oct 2010 18:27:04 +0000 (18:27 +0000)]
fix bugs to run without pool allocation like the ppopp paper experiments

12 years agocpaturing some makefile updates and have OoOJava print number of sites disjoint reach...
jjenista [Tue, 5 Oct 2010 17:07:03 +0000 (17:07 +0000)]
cpaturing some makefile updates and have OoOJava print number of sites disjoint reachability 2nd pass flagged

12 years agoConfig the size of the shared heap for TSP benchmark
jzhou [Mon, 4 Oct 2010 18:10:44 +0000 (18:10 +0000)]
Config the size of the shared heap for TSP benchmark

12 years agoFix bugs in cache adapt verion.
jzhou [Fri, 1 Oct 2010 23:34:35 +0000 (23:34 +0000)]
Fix bugs in cache adapt verion.

12 years agoWork-In-Progress check in. Will not error out if RCR is not enabled HOWEVER it will...
stephey [Fri, 1 Oct 2010 08:29:10 +0000 (08:29 +0000)]
Work-In-Progress check in. Will not error out if RCR is not enabled HOWEVER it will error out if you compile with RCR. The reason is that the RCR output is not yet connected with the C-side and some of the functions it prints out are just place holder Strings for me to visually check if the output is correct.

12 years agoforgot an #endif at the end of the file
stephey [Fri, 1 Oct 2010 08:05:32 +0000 (08:05 +0000)]
forgot an #endif at the end of the file

12 years agoAdded C files from oooJava folder + minor modifications to names to prevent conflicts
stephey [Fri, 1 Oct 2010 03:34:27 +0000 (03:34 +0000)]
Added C files from oooJava folder + minor modifications to names to prevent conflicts

12 years agofor timing disjoint reachability in oooJava benchmarks
jjenista [Fri, 1 Oct 2010 01:52:27 +0000 (01:52 +0000)]
for timing disjoint reachability in oooJava benchmarks

12 years agovarious hacks...
bdemsky [Wed, 29 Sep 2010 22:23:25 +0000 (22:23 +0000)]
various hacks...
clean up cases if we have duplicate...

12 years agoget rid of truncate macro...just stick it into the bit shifting...
bdemsky [Wed, 29 Sep 2010 21:03:54 +0000 (21:03 +0000)]
get rid of truncate macro...just stick it into the bit shifting...

12 years agoClean up the code
jzhou [Wed, 29 Sep 2010 00:15:56 +0000 (00:15 +0000)]
Clean up the code

12 years agoClean up the code: clean up debug msgs
jzhou [Tue, 28 Sep 2010 18:18:33 +0000 (18:18 +0000)]
Clean up the code: clean up debug msgs

12 years agonames
jjenista [Tue, 28 Sep 2010 18:03:51 +0000 (18:03 +0000)]
names

12 years agobuildscript options to enable exaclty which coreprof events are needed, keeps instrum...
jjenista [Mon, 27 Sep 2010 22:03:37 +0000 (22:03 +0000)]
buildscript options to enable exaclty which coreprof events are needed, keeps instrumentation overhead at minimum, will help for new events to imrpove SOR

12 years agobug fix, new events
jjenista [Mon, 27 Sep 2010 22:02:05 +0000 (22:02 +0000)]
bug fix, new events

12 years agoTested and Debugged versions of waitingQueue for the hashStructure.
stephey [Sun, 26 Sep 2010 05:44:16 +0000 (05:44 +0000)]
Tested and Debugged versions of waitingQueue for the hashStructure.

12 years agoChanges
stephey [Sun, 26 Sep 2010 05:39:52 +0000 (05:39 +0000)]
Changes

12 years agoremoved locking for waitingQueue add since it's not really needed.
stephey [Sun, 26 Sep 2010 05:22:59 +0000 (05:22 +0000)]
removed locking for waitingQueue add since it's not really needed.

12 years agoWaitingQueue Correctness tests for RCR. FAKE mlp_lock.* and RuntimeConflictResolver...
stephey [Sun, 26 Sep 2010 05:08:24 +0000 (05:08 +0000)]
WaitingQueue Correctness tests for RCR. FAKE mlp_lock.* and RuntimeConflictResolver.* are included so that the waitingQueue can be tested without compiling/including the rest of the project

12 years agofound a bug, wasn't returning pointer at end of poolcreate--pool allocation doesn...
jjenista [Fri, 24 Sep 2010 23:44:38 +0000 (23:44 +0000)]
found a bug, wasn't returning pointer at end of poolcreate--pool allocation doesn't seem to improve SOR speedup

12 years agoClean up the code: remove commented code
jzhou [Fri, 24 Sep 2010 22:59:23 +0000 (22:59 +0000)]
Clean up the code: remove commented code

12 years agothere is definitely an uninitialized mem bug, temp fix is to zero allocated task...
jjenista [Fri, 24 Sep 2010 19:48:00 +0000 (19:48 +0000)]
there is definitely an uninitialized mem bug, temp fix is to zero allocated task records, however even with workaround pool allocation crashes SOR

12 years agosome outstanding mods to coreprof parsing
jjenista [Fri, 24 Sep 2010 19:30:05 +0000 (19:30 +0000)]
some outstanding mods to coreprof parsing

12 years agoworking on task mem pool, there is a crash bug, use -ooodebug-disable-task-mem-pool...
jjenista [Thu, 23 Sep 2010 23:54:19 +0000 (23:54 +0000)]
working on task mem pool, there is a crash bug, use -ooodebug-disable-task-mem-pool if you want to run OoOJava programs for the time being

12 years agowhile working on memory pool for task records, fixed bug where mem Q hashtable entrie...
jjenista [Thu, 23 Sep 2010 19:17:12 +0000 (19:17 +0000)]
while working on memory pool for task records, fixed bug where mem Q hashtable entries had to wait for pointer value but used the wrong object id (oid)

12 years agouse buildscript option to give benchmarks enough heap memory
jjenista [Thu, 23 Sep 2010 19:07:50 +0000 (19:07 +0000)]
use buildscript option to give benchmarks enough heap memory

12 years agosupport for optimizations later, leaf tasks need less data
jjenista [Thu, 23 Sep 2010 19:06:54 +0000 (19:06 +0000)]
support for optimizations later, leaf tasks need less data

12 years agobug fix in gc cache adaption
jzhou [Thu, 23 Sep 2010 00:25:13 +0000 (00:25 +0000)]
bug fix in gc cache adaption

12 years agobug fix
jzhou [Wed, 22 Sep 2010 18:55:18 +0000 (18:55 +0000)]
bug fix

12 years agoAdded getter method to get all effects (needed for RuntimeConflictResolver.java)
stephey [Wed, 22 Sep 2010 11:55:29 +0000 (11:55 +0000)]
Added getter method to get all effects (needed for RuntimeConflictResolver.java)

12 years agoThis version prints out correct C code in terms of where things should be but is...
stephey [Wed, 22 Sep 2010 11:49:24 +0000 (11:49 +0000)]
This version prints out correct C code in terms of where things should be but is only tested on the simplest case. Will try more cases later.

Note: BuildCode.java was also modified to make this work but is not committed since it's only half changed (no stallsite code).

12 years agofix bugs
bdemsky [Wed, 22 Sep 2010 02:29:13 +0000 (02:29 +0000)]
fix bugs

12 years agooptimiz
bdemsky [Wed, 22 Sep 2010 02:26:03 +0000 (02:26 +0000)]
optimiz

12 years agoturn off coreprof by default
jjenista [Tue, 21 Sep 2010 22:13:53 +0000 (22:13 +0000)]
turn off coreprof by default

12 years agofix performance bug
jzhou [Tue, 21 Sep 2010 22:08:15 +0000 (22:08 +0000)]
fix performance bug

12 years agomy changes
jzhou [Tue, 21 Sep 2010 21:46:12 +0000 (21:46 +0000)]
my changes

12 years agoan appropriate test to simulate SOR for new pool allocation
jjenista [Tue, 21 Sep 2010 21:30:47 +0000 (21:30 +0000)]
an appropriate test to simulate SOR for new pool allocation

12 years agodisable coreprof by default
jjenista [Tue, 21 Sep 2010 21:29:11 +0000 (21:29 +0000)]
disable coreprof by default

12 years agohelp user by showing human-readable task identifier
jjenista [Tue, 21 Sep 2010 21:28:39 +0000 (21:28 +0000)]
help user by showing human-readable task identifier

12 years agochanges
bdemsky [Tue, 21 Sep 2010 21:23:12 +0000 (21:23 +0000)]
changes

12 years agoBug fix in gc cache adaption version
jzhou [Tue, 21 Sep 2010 03:59:30 +0000 (03:59 +0000)]
Bug fix in gc cache adaption version

12 years agoRemove the boundary of cache access sampling. Add output of the changed cache policy...
jzhou [Mon, 20 Sep 2010 00:37:27 +0000 (00:37 +0000)]
Remove the boundary of cache access sampling. Add output of the changed cache policy of the shared pages. Also fix some bugs.

12 years agoInitial version of cache adaption for mutator
jzhou [Sat, 18 Sep 2010 20:08:58 +0000 (20:08 +0000)]
Initial version of cache adaption for mutator

12 years agomodularize core prof correctly, starting mem pool strat for task records
jjenista [Thu, 16 Sep 2010 21:54:42 +0000 (21:54 +0000)]
modularize core prof correctly, starting mem pool strat for task records

12 years agomini test to mimic SOR for mem pool testing
jjenista [Thu, 16 Sep 2010 21:53:40 +0000 (21:53 +0000)]
mini test to mimic SOR for mem pool testing

12 years agotemporarily disabling this coreprof event, it appears to have a significant impact...
jjenista [Thu, 16 Sep 2010 18:11:14 +0000 (18:11 +0000)]
temporarily disabling this coreprof event, it appears to have a significant impact compared to system overhead, only put back in when we really need it

12 years agoMake the statistic data for cache adaption irrelavent with the timer counter. Use...
jzhou [Thu, 16 Sep 2010 00:04:23 +0000 (00:04 +0000)]
Make the statistic data for cache adaption irrelavent with the timer counter. Use smaller page for cache adaption version. Also make the cache policy decision modular

12 years agoAdd code for change the cache policy for the mutator. In each round of gc, decide...
jzhou [Wed, 15 Sep 2010 22:38:32 +0000 (22:38 +0000)]
Add code for change the cache policy for the mutator.  In each round of gc, decide the new policy for each page according to the gathered statistic data and adapt the new policies right before finishing gc.

12 years agothis micro benchmark does not accurately imitate OOOjava, just checking in current...
jjenista [Wed, 15 Sep 2010 21:16:05 +0000 (21:16 +0000)]
this micro benchmark does not accurately imitate OOOjava, just checking in current code for possible reuse later

12 years agoa self-contained test of the mempool strategy, doesnt seem promising yet
jjenista [Sun, 12 Sep 2010 21:23:31 +0000 (21:23 +0000)]
a self-contained test of the mempool strategy, doesnt seem promising yet

12 years agoWork In Progress Commit.
stephey [Sat, 11 Sep 2010 01:37:47 +0000 (01:37 +0000)]
Work In Progress Commit.
Added logic that will print out when traverser needs to look in waitingQueue/HashStructure. Separated enumeration of heaproots and printing of C methods (for the very end). Added TEMP functions that will create an array of HashStructures to simulate threadlocal variables (so that it's easier to debug)

Next Steps: Modify and Complete C-side. Test/Debug C-side. Connected both sides. Test/Debug both sides together.

Next Next Step: Work with Jim/YongHun to integrate RCR into main project.

12 years agochange parser to match runtime changes, still some kooky issue where child events...
jjenista [Fri, 10 Sep 2010 19:44:57 +0000 (19:44 +0000)]
change parser to match runtime changes, still some kooky issue where child events appear to take infinity perfent of parent sometimes, strange

12 years agotask execute, dispatch, reture and work grabbing events debugged
jjenista [Fri, 10 Sep 2010 19:44:11 +0000 (19:44 +0000)]
task execute, dispatch, reture and work grabbing events debugged

12 years agoWork In Progress Commit.
stephey [Fri, 10 Sep 2010 10:01:11 +0000 (10:01 +0000)]
Work In Progress Commit.
Added supporting data structures and logic for keeping track of what needs to be entered into a waitingQueue upon 'stopping' the traverser. Need to write code that spits out the correct C-code with waitingQueue stuff (halfway done)

Next steps: Do the above, separate the enumeration of heaproots and printing of C-methods until the very end, and modify/write C-side stuff to correctly respond to the generated calls.

12 years agoAdded logic for determining weakly connected groups of Heaproots and changed traverse...
stephey [Thu, 9 Sep 2010 01:38:32 +0000 (01:38 +0000)]
Added logic for determining weakly connected groups of Heaproots and changed traverser invocation/generation method header to support resuming traversal from predefined points (incomplete, still need to modify addChecker to take advantage of the new header).

12 years agoChanged parsing of coreprof data to keep parent/child event relationships, breaks...
jjenista [Wed, 8 Sep 2010 22:46:11 +0000 (22:46 +0000)]
Changed parsing of coreprof data to keep parent/child event relationships, breaks parent into percentage spent in each child

12 years agoJim told me to change the int hashCodeSpecific() function to return id.hashCode(...
stephey [Wed, 8 Sep 2010 22:39:34 +0000 (22:39 +0000)]
Jim told me to change the int hashCodeSpecific() function to return id.hashCode() instead of id itself.

12 years agoChanged coreprof macros to match coreprof.h, altered output to print self-time as...
jjenista [Tue, 7 Sep 2010 19:47:04 +0000 (19:47 +0000)]
Changed coreprof macros to match coreprof.h, altered output to print self-time as a percentage of total time.

12 years agoSeveral changes:
jjenista [Tue, 7 Sep 2010 19:40:31 +0000 (19:40 +0000)]
Several changes:

-Can control space allocated for coreprof events with a buildscript option.

-Attempted to exit ooojava worker threads gracefully with pthread_cleanup_XXX
and pthread_cancel in asynchronous mode.  Cleanup doesn't trigger from asynch
cancel, so this doesn't do the trick.  The reason this is nice is that COREPROF_EXIT
is never called by worker threads, so the timestamp of the MAIN duration is being
fudged by using the last timestamp of a sub event.  If we wanted to add more
interesting functionality to COREPROF_EXIT we can't use the workaround and we'll
have to revisit this--left the commented pthread code as a starting point for later.

-Adjusted event IDs and the baseshift so same number of bits are used, but
it is easier to look at the coreprof output file in a hex editor to visually
inspect the raw data, for debugging.