id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
work_zmfcmcryxjg5rb3adtopcdqnza	Bérenger Bramas	Increasing the degree of parallelism using speculative execution in task-based runtime systems	2019	25	.pdf	application/pdf	11109	1610	70	Task-based programming models have demonstrated their efficiency in the development of scientific applications on modern high-performance platforms. parallelism when it is uncertain if some tasks will modify data, and we formalize a new methodology to enable speculative execution in a graph of tasks. Increasing the degree of parallelism using speculative execution in task-based runtime systems. objectives are to get a generic pattern/method to use speculation in task-based RS's and to • Introduce SPETABARU, a new task-based RS capable of speculation. Here, task B may or may not write data, but to ensure a correct execution, we must use the data, then the speculation has failed no matter the result of the other uncertain tasks. writes data, then we will have to compute task i+1 and speculate over it with duplicate of 26 // Insert t as a speculative task using data duplicates, 26 // Insert t as a speculative task using data duplicates,	./cache/work_zmfcmcryxjg5rb3adtopcdqnza.pdf	./txt/work_zmfcmcryxjg5rb3adtopcdqnza.txt