Jing Wang. Thread optimizations in concurrent object oriented languages. Master's thesis, University of California, Santa Barbara, Computer Science Department, September 1998.  Home/Search   Document Details and Download   Summary   Related Articles

....Therefore, we have to balance our desire to have a large pool (to severely curtail or eliminate the possibility that a thread will be created) with our need to have enough memory for the user application to run. We therefore implemented an adaptive thread pool, as originally described in [42]. The pool tries to maintain a size that matches the application s current needs. It monitors the number of idle threads in the pool across a fixed number of pool operations, and then steps the pool size up or down if needed. Between steps, if a thread is needed and the pool is empty, then a ....

....the child method already finished executing. In Figure 7(c) the calling method is nonblocking between the call and the join. Hence, execution of the child method is postponed; the call is a no op. When the join is reached, the parent then executes the child method synchronously. As shown in [42], implementing thread inlining introduces some overhead into the system, due to the necessary bookkeeping. In fact, if the parent makes only 1 or 2 method calls, the costs of inlining outweigh the savings. However, if the parent thread makes many method calls, then the savings can be substantial. ....

Jing Wang. Thread optimizations in concurrent object oriented languages. Master's thesis, University of California, Santa Barbara, Computer Science Department, September 1998.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC