Robert H. Halstead. Multilisp: A language for parallel symbolic computation. ACM Transactions on Programming Languages and Systems, 7(4):501-539, October 1985.  Home/Search   Document Not in Database   Summary   Related Articles   Check

.... concluded an encoding of a version of ML that simultaneously supports functions, recursion, de nitions, pairs, unit type, sum types, void type, recursive types, parametric polymorphic types, intersection types, suspensions with memoization, mutable references, futures in the style of Multilisp [Hal85], and concurrency in the style of CML [Rep99] We further have a representation of the second author s security protocol speci cation framework MSR [Cer01] and representations of the synchronous and asynchronous calculus. Other targets for case studies in the realm of concurrent and imperative ....

Robert H. Halstead. Multilisp: A language for parallel symbolic computation. ACM Transactions on Programming Languages and Systems, 7(4):501-539, October 1985.

....a task. See Appendix B for a list of events traced. Let us first review the Mul T system briefly. Mul T is a parallel Lisp system that runs on an Encore Multimax multiprocessor. It is an extended version of the T system [18] that supports parallel processing using Multilisp s future construct [10]. Mul T uses a modified version of T s ORBIT compiler [14] to generate native code for the Multimax s NS32332 processors. Mul T (like Multilisp) is an extended version of Scheme [1] a lexically scoped dialect of Lisp. Mul T s execution environment contains the same sorts of data types and ....

Robert H. Halstead. Multilisp: A Language for Parallel Symbolic Computation. ACM Transactions on Programming Languages and Systems, 7(4):501-539, October 1985.

....pure, sequential programming language features. We were pleasantly surprised that our destination passing style encoding may be extended to include mutable references and common concurrency primitives as well. Futures. We now come to the rst parallel construct: futures in the style of MultiLisp [Hal85], adapted to ML. There is no new type, since a future can be of any type. A destination D can now serve as a value (called promise D) If a promise is ever needed it needs to be available from then on, which is why we have a new family deliver V D which delivers value V to destination D and will ....

Robert H. Halstead. Multilisp: A language for parallel symbolic computation. ACM Transactions on Programming Languages and Systems, 7(4):501-539, October 1985.

....pure, sequential programming language features. We were pleasantly surprised that our destination passing style encoding may be extended to include mutable references and common concurrency primitives as well. Futures. We now come to the first parallel construct: futures in the style of MultiLisp [Hal85], adapted to ML. There is no new type, since a future can be of any type. A destination D can now serve as a value (called promise D) If a promise is ever needed it needs to be available from then on, which is why we have a new family deliver V D which delivers value V to destination D and will ....

Robert H. Halstead. Multilisp: A language for parallel symbolic computation. ACM Transactions on Programming Languages and Systems, 7(4):501--539, October 1985.

.... concluded an encoding of a version of ML that simultaneously supports functions, recursion, definitions, pairs, unit type, sum types, void type, recursive types, parametric polymorphic types, intersection types, suspensions with memoization, mutable references, futures in the style of Multilisp [Hal85], and concurrency in the style of CML [Rep99] We further have a representation of the second author s security protocol specification framework MSR [Cer01] and representations of the synchronous and asynchronous # calculus. Other targets for case studies in the realm of concurrent and imperative ....

Robert H. Halstead. Multilisp: A language for parallel symbolic computation. ACM Transactions on Programming Languages and Systems, 7(4):501--539, October 1985.

....wait time distribution. We consider three types of synchronization: producer consumer, barrier, and mutual exclusion. Producer consumer synchronization is performed between one producer and one or more con sumers of the data produced Examples of this type of synchronization include futures [13] and I structures [5] Barrier synchronization ensures that all threads participating in a barrier have reached a point in a program before proceeding. Mutual exclusion synchronization is used to provide exclusive access to data structures and critical sections of code. As motivation for the ....

....they allow multiple non locking readers. Semaphores A semaphore is implemented as a one element L structure. semaphore p and semaphore v are easily implemented using L structure reads and writes. Semaphores are used to implement mutual exclusion. Futures Futures specify parallelism in Multilisp [13] and synchronization on the return values of the threads. It is a form of producer consumer synchronization. The future object is simply a memory word that initially holds the queue of waiting consumers and eventually holds the value of the future when resolved. Barriers Barriers ensure that all ....

Robert H. Halstead. Multilisp: A Language for Parallel Symbolic Computation. ACM Transactions on Programming Languages and Systems, 7(4):501 539, October 1985.

....Rapid context switching makes the use of coherence protocols that guarantee sequential consistency feasible because the processor can switch to a different thread while the acknowledgments to outstanding memory transactions are awaited. Rapid trap handling also allows efficient handling of Futures [20] and full empty bit [2] traps. Useful execution : Overhead r ] idle dme Figure 1: Hiding network latency by multithreading the processor. 3 A Multithreaded Processor Performance Model A model for a multithreaded processor must represent the tradeoff between increased processor utilization ....

Robert H. Halstead. Multilisp: A Language for Parallel Symbolic Computation. ACM Transactions on Programming Languages and Systems, 7(4):501-539, October 1985.

....wait time distribution. We consider three types of synchronization: producer consumer, barrier, and mutual exclusion. Producer consumer synchronization is performed between one producer and one or more consumers of the data produced. 2 Examples of this type of synchronization include futures [13] and I structures [5] Barrier synchronization ensures that all threads participating in a barrier have reached a point in a program before proceeding. Mutual exclusion synchronization is used to provide exclusive access to data structures and critical sections of code. As motivation for the use ....

....they allow multiple non locking readers. Semaphores A semaphore is implemented as a one element L structure. semaphore p and semaphore v are easily implemented using L structure reads and writes. Semaphores are used to implement mutual exclusion. Futures Futures specify parallelism in Multilisp [13] and synchronization on the return values of the threads. It is a form of producer consumer synchronization. The future object is simply a memory word that initially holds the queue of waiting consumers and eventually holds the value of the future when resolved. Barriers Barriers ensure that all ....

Robert H. Halstead. Multilisp: A Language for Parallel Symbolic Computation. ACM Transactions on Programming Languages and Systems, 7(4):501--539, October 1985.

....an atomic fetch store instruction, and a set of instructions that manipulate full empty bits[53, 28] associated with each memory word. These instructions are directly supported by the Alewife CMMU, and are cache coherent. ffl Alewife supports automatic detection of unresolved futures [22] by using the least significant bit of a data word as a tag bit, and by trapping on misaligned addresses in ORBIT Compiler Sparcle Simulator CMMU Simulator C parser Parallel C program Mul T program T intermediate form Run Time System and Library code Sparcle Object Code Memory ....

....to implement synchronization protocols and the tradeoffs that arise. In research on waiting algorithms, we consider additional waiting mechanisms that are made possible through multithreading. We also consider the performance of waiting algorithms for several synchronization types, such as futures [22] and I structures [6] that are implemented with tagging and full empty bits. 2.2.4 Run Time System Assumptions The scheduling policy and the run time overhead of thread management have a significant impact on the performance of waiting algorithms. In Alewife s run time system, thread scheduling ....

[Article contains additional citation context not shown here]

Robert H. Halstead. Multilisp: A Language for Parallel Symbolic Computation. ACM Transactions on Programming Languages and Systems, 7(4):501--539, October 1985.

....system performance (although the pipeline design is complicated by the need to handle pipeline dependencies) In APRIL, thread scheduling is done in software, and unlimited virtual dynamic threads are supported. APRIL supports full empty bit synchronization, and provides tag support for futures [9]. In this paper the terms process, thread, context, and task are used equivalently. By taking a systems level design approach that considers not only the processor, but also the compiler and run time system, we were able to migrate several noncritical operations into the software system, greatly ....

Robert H. Halstead. Multilisp: A Language for Parallel Symbolic Computation. ACM Transactions on Programming Languages and Systems, 7(4):501--539, October 1985.

No context found.

Robert H. Halstead. Multilisp: A language for parallel symbolic computation. ACM Transactions on Programming Languages and Systems, 7(4):501-539, October 1985.

No context found.

Robert H. Halstead. Multilisp: A language for parallel symbolic computation. ACM Transactions on Programming Languages and Systems, 7(4):501--539, October 1985.

No context found.

R. H. Halstead. Multilisp: A language for parallel symbolic computation. ACM Transactions on Programming Languages and Systems, 7(4):501--539, Oct. 1985.

No context found.

Robert H. Halstead. Multilisp: A language for parallel symbolic computation. ACM Transactions on Programming Languages and Systems, 7(4):501--539, October 1985.

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