Backtracking-based load balancing

Abstract

High-productivity languages for parallel computing become more important as parallel environments including multicores become more common. Cilk is such a language. It provides good load balancing for many applications including irregular ones; that is, it keeps all workers busy by creating plenty of "logical" threads and adopting the oldest-first work stealing strategy. This paper proposes a "logical thread"-free framework called Tascell, which achieves a higher performance and supports a wider range of parallel environments including clusters without loss of productivity. A Tascell worker spawns a "real" task only when requested by another idle worker. The worker performs the spawning by temporarily "backtracking" and restoring its oldest task-spawnable state. Our approach eliminates the cost of spawning/managing logical threads. It also promotes the reuse of workspaces and improves the locality of reference since it does not need to prepare a workspace for each concurrently runnable logical thread. Furthermore, Tascell enables elegant and highly-efficient backtrack search algorithms with delayed workspace copying. For instance, our 16-queens problem solver is 1.86 times faster than Cilk on a system with two dual-core processors. Our approach also enables a single program to run in both shared and distributed memory environments with reasonable efficiency and scalability.

An Implementation of Exception Handling with Collateral Task Abortion

“…The Tascell framework [4], [12] consists of a compiler for an extended C language, called the Tascell language, and a runtime system for parallel computations.…”

“…Applications having this property are often implemented using taskparallel languages, such as X10 [1], Cilk [2], Intel Cilk Plus [3], and Tascell [4], that allow tasks to be dynamically spawned such that they are automatically assigned to workers, that is, parallel threads and/or processes, so that a worker has an exclusive set of subtrees as its task set.…”

“…Figure 4 shows pseudo code for a parallel binary search for all solutions using this technique. When a worker notices that no solutions exist in the subtree having the root xnode, it prunes the subtree (lines [4][5][6][7][8]. Then, when another worker notices that the worker is traversing the pruned subtree, it can abort the traversal simply by throwing an exception tagged with xnode, which is caught by the catcher (line 20) established when xnode is visited as node.…”

“…For example, Intel Cilk Plus supports throwing an exception beyond synchronization points, but the child tasks spawned in the try block are not aborted automatically. Therefore, we implemented exception handling features with collateral task abortion as an enhancement of the existing task-parallel language Tascell [4]. That is, we designed an enhanced Tascell language by adding the try-catch and throw constructs to the baseline Tascell, and implemented this enhanced Tascell by modifying the Tascell compiler and the task scheduler.…”

Dynamic Multiple Work Stealing Strategy for Flexible Load Balancing

“…Tascell [13] is a parallel programming framework. It includes a multithreaded library and languages.…”