nathan
|
dd35477346
|
Added more direct iteration
|
2015-05-09 15:06:37 +02:00 |
|
nathan
|
aaf8e0ddf4
|
Created backup of MpmcTransferArrayQueue, created/implemented 'multi' version, code polish
|
2015-05-09 14:38:06 +02:00 |
|
nathan
|
3aa5ca59f6
|
Cleaned up tests, added Thread interruption back, code polish
|
2015-05-09 00:09:48 +02:00 |
|
nathan
|
7fa59650a8
|
WIP, code polish. discovered rare test hang
|
2015-05-08 17:32:19 +02:00 |
|
nathan
|
b967733a63
|
Added jctools library, cleaned up. Slow performance for 4x4??
|
2015-05-08 13:32:56 +02:00 |
|
nathan
|
1a591165a8
|
WIP - pre-lib for jctools
|
2015-05-06 20:47:02 +02:00 |
|
nathan
|
f96a2f4c19
|
Improved simpleq, now beats LTQ in concurrent tests
|
2015-05-04 19:03:50 +02:00 |
|
nathan
|
2705daa073
|
much better results. almost same as LTQ, but with no heap objects
|
2015-05-03 22:43:35 +02:00 |
|
nathan
|
8c01959080
|
WIP, crap multiprocessor performance with simplequeue abstraction, 160k/s
|
2015-05-03 15:55:26 +02:00 |
|
nathan
|
2e1feacc0e
|
Cleanup old mpmc code
|
2015-05-03 00:19:29 +02:00 |
|
nathan
|
fca3d02652
|
Tweaked queue to simpler logic. added park/unpark. 6.3m ops/sec
|
2015-05-03 00:08:54 +02:00 |
|
nathan
|
ac4e396f5d
|
renamed queue
|
2015-05-01 00:40:28 +02:00 |
|
nathan
|
4276f881e8
|
WIP - got mpmc queue w/ parks + fixed-deadlocking. @ 4.6m ops/sec
|
2015-04-22 22:47:23 +02:00 |
|
nathan
|
8b23c992e1
|
back to custom MPMC queue. wip locking
|
2015-04-21 22:50:39 +02:00 |
|
nathan
|
b4e012c87c
|
WIP, not liking results
|
2015-04-21 19:47:34 +02:00 |
|
nathan
|
5be16c6345
|
WIP linkedarraylist performance
|
2015-04-21 16:30:19 +02:00 |
|
nathan
|
2c87177b31
|
WIP LinkedArrayList queue
|
2015-04-21 15:44:15 +02:00 |
|
nathan
|
eacca670de
|
working, but slow. Using MPMC array queue
|
2015-04-20 21:25:59 +02:00 |
|
nathan
|
42230706c7
|
WIP - got queue to no longer deadlock
|
2015-04-19 14:25:12 +02:00 |
|
nathan
|
2e6ba71eed
|
WIP, node stucture is finish, 20M op/s
|
2015-04-13 00:41:32 +02:00 |
|
nathan
|
196aa2466a
|
WIP. node perf 70m
|
2015-04-12 23:21:09 +02:00 |
|
nathan
|
72cc6d15d4
|
WIP - 64Mop/s w/ node
|
2015-04-10 13:10:10 +02:00 |
|
nathan
|
f6bf52c57f
|
Added JCTools MPMC queue, added more benchmark tests, WIP simpleQ
|
2015-04-09 23:08:12 +02:00 |
|
nathan
|
9acffac11b
|
wip exchanger. wip simpleq
|
2015-03-01 12:29:35 +01:00 |
|
nathan
|
3c021f383c
|
WIP disruptor. WIP simple queue
|
2015-03-01 02:23:41 +01:00 |
|
nathan
|
e257aced31
|
polish
|
2015-02-28 01:30:51 +01:00 |
|
nathan
|
7d410e78be
|
Code cleanup, restored sub/unsub unit test to original version
|
2015-02-28 01:13:11 +01:00 |
|
nathan
|
45614e8bdd
|
WIP cleanup, unit tests
|
2015-02-27 16:32:05 +01:00 |
|
nathan
|
dd31c45dc3
|
Moved project to dorkbox name. WIP multi-messages, and vararg/vararg-superclass
|
2015-02-27 00:01:38 +01:00 |
|
nathan
|
01f87da371
|
Added StrongConcurrentSetV8, WIP - var arg. Added boolean to skip checking for subtypes/varargs. Code polish
|
2015-02-26 16:15:48 +01:00 |
|
nathan
|
b8f06b7c33
|
Changed to single access for sub/unsub (now passes unit tests). Changed to ConcurrentHashMapV8, 174/87 ns/op
|
2015-02-24 14:26:32 +01:00 |
|
nathan
|
4eef758e89
|
changed striping to num_threads. 176/100 ns/op
|
2015-02-22 23:20:07 +01:00 |
|
nathan
|
b42eb86531
|
Small amount of code polish. 186/93 ns/op
|
2015-02-22 23:10:15 +01:00 |
|
nathan
|
0171a5558c
|
WIP - now have superclass path enabled. @ 192 ns/op
|
2015-02-22 02:39:08 +01:00 |
|
nathan
|
3315f079d5
|
Tuned concurrency
|
2015-02-20 20:21:00 +01:00 |
|
nathan
|
d0584d52b0
|
Cuncurrent sub/pub. 169ish ns/op VS 256 ns/op
|
2015-02-20 19:54:53 +01:00 |
|
nathan
|
2be96d7ff6
|
Now have 300ms publication for both 2/4million handled events
|
2015-02-20 15:43:39 +01:00 |
|
nathan
|
758a1c837a
|
Changed superClassSubs to use fast iterator
|
2015-02-20 12:17:07 +01:00 |
|
nathan
|
13cee46f6c
|
WIP - more even dist of performance (smaller up/down spikes). Using faster collections. SUB/UNSUB is write-locked
|
2015-02-20 01:46:03 +01:00 |
|
nathan
|
86d04c899c
|
WIP - getting better performance, using non-blocking. around 300ms for 2million, 550 for 4million
|
2015-02-19 18:39:05 +01:00 |
|
nathan
|
120054afd4
|
WIP - playing with concurrent maps + ability to have non-blocking for updates
|
2015-02-19 17:52:48 +01:00 |
|
nathan
|
8b69d53263
|
WIP, cleaning code. Getting multi-arg versions working
|
2015-02-17 22:43:48 +01:00 |
|
nathan
|
40d8bb70bd
|
NOT MPMC, LTQ - fastest so far via using non-copy interation for subscriptions
|
2015-02-17 07:59:45 +01:00 |
|
nathan
|
67c68dbc14
|
WIP - best performance for low cores
|
2015-02-17 00:26:46 +01:00 |
|
nathan
|
a91b3af27f
|
it's slower, not sure if correct direction
|
2015-02-16 23:37:41 +01:00 |
|
nathan
|
d1338ad46c
|
Cleaned/removed disruptor
|
2015-02-13 17:59:13 +01:00 |
|
nathan
|
9b0e0c2bb6
|
wip, wrapping up
|
2015-02-13 17:19:10 +01:00 |
|
nathan
|
139f4c3069
|
using disruptor + LTQ
|
2015-02-13 11:02:43 +01:00 |
|
nathan
|
3418b08529
|
WIP. have LinkedTransferQueue working, so far is the best
|
2015-02-12 21:51:10 +01:00 |
|
nathan
|
220e522ff0
|
Changed reflection to use (much faster) ReflectASM
|
2015-02-08 14:51:30 +01:00 |
|