247 Commits (c42c4a6718599126bd9e7ba7f0407db18618c9e0)

Author	SHA1	Message	Date
Yu Yang	82693e7227	Wait nccl all reduce	7 years ago
Yu Yang	eb0a580e78	Add enforce	7 years ago
Yu Yang	65bc7d17d5	Add mtx to ncclAllReduce	7 years ago
Yu Yang	ba227df941	Expose num_threads	7 years ago
Yu Yang	1533bf12df	Use event and single thread	7 years ago
Yu Yang	95a0d7c7c1	Illegal memory access	7 years ago
Yu Yang	798e6907b4	Change mem order	7 years ago
Yu Yang	1c2b6100b0	Add	7 years ago
Yu Yang	4e43b71377	Add wait log	7 years ago
Yu Yang	dbed123382	Debug	7 years ago
Yu Yang	e53b6aba63	Use no thread	7 years ago
Yu Yang	a8bd7b9809	Add log	7 years ago
Yu Yang	3c9cea597e	Add more log	7 years ago
Yu Yang	f8f1a963d9	Add debug code	7 years ago
Yu Yang	fbbcedda01	Fix bug	7 years ago
Yu Yang	7643c2cbab	Add flag for use event	7 years ago
Yu Yang	ca4b3d2532	Use 12 threads	7 years ago
Yu Yang	f251a58e85	Use base class manage events	7 years ago
Yu Yang	1dd216dc3b	Wait bcast param	7 years ago
Yu Yang	4185dd48e4	Disable multi-thread	7 years ago
Yu Yang	631aa3d10a	Wait all inputs ready	7 years ago
Yu Yang	9b1f4d5d62	After nccl add event	7 years ago
Yu Yang	feb569f8ea	Add log	7 years ago
Yu Yang	260cfe3b86	Stop Wait NCCL Stream	7 years ago
Yu Yang	e025e284c6	Exchange wait op	7 years ago
Yu Yang	3238ce0672	Add wait	7 years ago
Yu Yang	8a9de67e17	Remove wait	7 years ago
Yu Yang	d2cb3790e9	Wait all evernts	7 years ago
Yu Yang	4137bb4eda	Add wait	7 years ago
Yu Yang	3da4159f88	Add run iter	7 years ago
Yu Yang	d3c82c356e	Wait multiple stream	7 years ago
Yu Yang	c18c2f6ab0	Sync all computation streams at the end of run	7 years ago
chengduo	597ba3f3f2	add more times close test (#9215)	7 years ago
Yu Yang	c372ce2885	Add event for computational op	7 years ago
Yu Yang	b94ffacbd7	SetDev	7 years ago
Yu Yang	99f85a9fbc	Set dev	7 years ago
Yu Yang	d26f093f9d	Log	7 years ago
Yu Yang	d55a03d916	Scale loss on place	7 years ago
Yu Yang	932364a275	Sync dev	7 years ago
Yu Yang	dad7bdabd4	Add setDev	7 years ago
Yu Yang	7fd0d24e0c	Add lgo	7 years ago
Yu Yang	bade579826	Wait code	7 years ago
Yu Yang	4a330094f9	Add log	7 years ago
Yu Yang	9824e8f311	Scale loss op use event	7 years ago
Yu Yang	071043c388	Add paddle enforce	7 years ago
Yu Yang	8af57706e2	Only wait same device	7 years ago
Yu Yang	29cc9f308d	SetDev for nccl	7 years ago
Yu Yang	d7badb3ed2	Use event to sync stream	7 years ago
Yu Yang	1f53193a63	Use atomic code	7 years ago
Yu Yang	c7beac1426	Add dummy var	7 years ago
Yu Yang	5fa535b717	Wait all thread done	7 years ago
Xin Pan	898e0ffa21	Merge pull request #9190 from panyx0718/p2p ... Enable P2P memory copy	7 years ago
Yu Yang	7bff02b2ca	Change to pending op	7 years ago
Yu Yang	866f6f1be0	Debug	7 years ago
Yu Yang	a5ba704de0	Counter	7 years ago
Yu Yang	a87ce91c4b	Use mtx	7 years ago
Yu Yang	ea11a0a853	Use volitie	7 years ago
Xin Pan	ce55975bb5	fix	7 years ago
Xin Pan	18ac6947d0	Enable P2P memory copy ... On k40 with 4 devices, time reduces from ~4.0 to ~3.8+, should be more obvious on better hardware	7 years ago
Yu Yang	515e516e77	Add more log	7 years ago
Yu Yang	1f063d0900	Memorder	7 years ago
Yu Yang	b1cb8bbd40	Debug	7 years ago
Yu Yang	b57b880b05	Debug	7 years ago
Yu Yang	f3e983e499	Memory order	7 years ago
Yu Yang	36e0415220	Single Thread	7 years ago
Yu Yang	5957f28b86	Debug	7 years ago
Yu Yang	f52714d391	Debug	7 years ago
Yu Yang	0023c3bcf5	Use atomic bool	7 years ago
Yu Yang	09935ab936	Debug	7 years ago
Yu Yang	f8141d90c8	Debug	7 years ago
Yu Yang	e18a269705	Add debug code	7 years ago
Yu Yang	9cb8f50302	Complete fetch op	7 years ago
Yu Yang	254d7ff4f5	Refactor local_scopes	7 years ago
Yu Yang	b2c7a9b828	Wait by stream	7 years ago
Yu Yang	e8a7e5d1e6	Update	7 years ago
Yu Yang	8f0590e7c5	Add ncclAllReduce	7 years ago
Yu Yang	c15d2c9edc	Update	7 years ago
Yu Yang	d470763f6c	Stash	7 years ago
Yu Yang	9fc0b596a9	Test more	7 years ago
Yu Yang	0ef9edf566	Stash	7 years ago
Liu Yiqun	253ba6672f	Merge branch 'develop' into core_inference_remove_clone	7 years ago
Yu Yang	5e87cd7574	Merge branch 'develop' of https://github.com/PaddlePaddle/Paddle into cpp_parallel_executor	7 years ago
Yu Yang	8c9cd369dc	Polish code style	7 years ago
Yu Yang	6f0dfd89a4	Single GPU ParallelExecutor complete	7 years ago
Xin Pan	21e2c42a46	Merge pull request #9141 from panyx0718/develop ... Fix a program copy regression.	7 years ago
Tao Luo	20be8e7e33	Merge pull request #9104 from ranqiu92/doc_dir ... change the dir of docs	7 years ago
Xin Pan	1ca1e1c384	Fix a program copy regression. ... Single device on se-resnet reduce from 0.56 to 0.50	7 years ago
ranqiu	64775126f3	change the dir of docs	7 years ago
Yu Yang	d84ddcf123	Stash	7 years ago
Yu Yang	193c0a7e43	Handle var hazard	7 years ago
Thuan Nguyen	1e4c504e60	Implement Select OP (#9088) ... * Fix old documentation for channel_recv * Initial design of CSP select * Redesign channel implementation for Select Op * Remove unecessary header * Initial checkin of select op, currently will read all the conditional_op in the cases block and also pull out all channels involved in the select. * Init python select op API * Python select bug fix when checking op creates block * Add case_to_execute as (a) input to select, (b) into the passed inputs into the select op * Add in addition code for select op * Init fibonacci test from python * implement fibonnaci sequence test * update fib unit test * Improve select test cases * Shorten non-pep-8-ed lines * Add methods on channel needed by select op * Fix compile issues, finish implementation, still need to debug code * Fix issue with fibonncci test, it works now! * Change QueueMessage callback to take in an ChannelAction enum, fix select unit test * Fix case attributes * Fix issue with select control flow * Make cases - previously on each selectcase conditional_block - attributes to select * Use class constants for type of channel * Change select op to take in "cases" attribute * return boolean from select callback function to tell Channel if this RECV or SEND should be executed * Improve attributes and inputs comments on select op * Fix issues with python unit test * Assert fibonacci final output * Fix issue when channel name / channel var is null for "default" case in select op * Assert base select test output * Make QueueMessage use shared pointer and modify the order of the callback * Fixing the order in which the callback is called * Move channel utility methods to paddle/fluid/operators/concurrency/channel_util * Create channel_util and move channel util methods * Fix crash when calling select_op * Fix deadlock * Fix issue of channel destructor deadlock * Fix precommit issues * Accidentally checked in changes to beam_search_op, reverting change. * Fix dependency issue in concurrency cmake * add device_context dependency for concurrency target	7 years ago
Yu Yang	35744e7b36	Polish code	7 years ago
Xin Pan	d284cf88e5	Merge pull request #9037 from panyx0718/develop ... Better timeline	7 years ago
Yu Yang	ae88fdefb7	Use thread pool	7 years ago
Abhinav Arora	41894da145	Add changes to channel that are needed for select op (#9084)	7 years ago
Yu Yang	692a0f7425	Better name	7 years ago
Yu Yang	baef1124fb	ParallelExecutor And dependency engine	7 years ago
Yibing Liu	90afbd2856	Move back operator's event to RunImpl()	7 years ago
Xin Pan	4840c49b27	Better timeline	7 years ago
Liu Yiqun	8ecad98578	Add the bool variable to decide whether to have a copy of the program in ExecutorPrepareContext.	7 years ago