Johan Montelius. Exploiting Fine-grain Parallelism in Concurrent Constraint Languages. PhD thesis, SICS Swedish Institute of Computer Science, SICS Box 1263, S-164 28 Kista, Sweden, April 1997. SICS Dissertation Series 25.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....order that records distributors in order of creation does not offer any improvement in a concurrent setting. A static order can improve this, but is difficult to define: all concurrent events must be ordered, including injection into spaces. In addition, a static order is costly to implement [60, 92]. On the other hand, expressiveness is still unsatisfactory, see Section 11.6. As a consequence, a simple but expressive way to employ multiple distributors is to explicitly program the order. Multiple distributors are then executed by a single thread in a well defined order. Since the base ....

....for detecting stability. The main disadvantage of scripting is that it is an inherently sequential technique because it only supports the view for a single space at a time. The Penny system, a parallel implementation of AKL, uses a different solution that provides multiple views at a time [93, 92]. Each variable maintains a list of speculative bindings indexed by the space for which the binding is valid. Montelius shows in [92] that this solution is efficient, since speculative bindings are infrequent. Podelski and Smolka study situated simplification in [105] as a technique for detecting ....

[Article contains additional citation context not shown here]

Johan Montelius. Exploiting Fine-grain Parallelism in Concurrent Constraint Languages. PhD thesis, SICS Swedish Institute of Computer Science, SICS Box 1263, S-164 28 Kista, Sweden, April 1997. SICS Dissertation Series 25.

....Implementing the EAM is an extremely complicated task the only attempt made, the AKL system [25] suffers the drawback of being based on a rather different language and semantics w.r.t. Prolog. Different parallel implementations of AKL have been recently completed with good performance results [29, 28, 33]. 12 suspend suspend suspend suspend Binding of Y in r is determinately promoted. Non determinate promotion is performed. X, Y p(X, Y) q(X, Y) r(Y) X = 2, m(Y) X = 3, n(Y) X = 3, t(Y) X = 3, s(Y) Y=5 n(5) t(5) n(5) s(5) X=3, Y=5 X=3, Y=5 X p(X, 5) q(X, 5) X = 2, m(5) X = 3, ....

....in the presence of or parallelism) compared to goal recomputation. Nevertheless, extending 13 the ACE system so that it becomes a true implementation of the EAM is a topic of our current investigation. Implementation mechanisms for flavors of the EAM have been also proposed in the AKL project [28, 33] and in the work by Costa and Lopes [26] The Extended Dynamic DAP has different termination properties than the Basic Andorra Model. This is because in the Basic Andorra Model the goals are reordered, i.e. determinate goals are placed before non determinate ones. This reordering can completely ....

J. Montelius. Exploiting Fine-grain Parallelism in Concurrent Constraint Languages. PhD thesis, Uppsala University, 1997.

....Prolog may have merits, especially if the aim is to design a language for concurrent programming, this is somewhat different from the aim of this research, which is to exploit parallelism in Prolog programs implicitly. In AKL, or parallelism is achieved by copying, and in implementations of AKL [23], this has been shown to cause or parallel programs to run significantly slower. 5 Conclusion This paper outlined the Fire implementation model for combining and and or parallelism, which may have some advantages over existing models, because it simplifies task switching, which is at the heart ....

J. Montelius. Exploiting Fine-grain Parallelism in Concurrent Constraint Languages. PhD thesis, Uppsala University, 1997.

....the EAM is an extremely complicated task the only attempt made, the AKL system [14] suffers the drawback of being based on a rather different language and semantics w.r.t. Prolog. A parallel implementation of AKL, called Penny, has been recently completed with good performance results [17, 16]. 3.1 EDDAP: A Step Towards EAM Dynamic DAP extended with allowing deterministic consumers to become a producer (Extended Dynamic Dependent And Parallelism (EDDAP) is arguably an implementation of a weak form of the EAM. Allowing a consumer goal to bind the dependent variable if it is doing so ....

Montelius, J. Exploiting Fine-grain Parallelism in Concurrent Constraint Languages. PhD thesis, Uppsala University, 1997.

....domain constraints would be used. This approach allows enumeration of the variable, but is less efficient in both time and space than using true finite domain constraints. Exploiting parallelism is another method of improving constraint propagation. The Penny language [Montelius and Haridi, 1997] [Montelius, 1997] is a parallel version of AKL that does not require the programmer to indicate parallelism explicitly. However, Penny is not source code compatible with AKL, and so using Penny would require retargeting the specification compiler and porting the agent library. We are still investigating the ....

Montelius, J. (1997). Exploiting Fine-grain Parallelism in Concurrent Constraint Languages. PhD thesis, Swedish Institute of Computer Science (SICS) and Uppsala University.

....large portion of the execution time. Finally, in subsection 4.5, we discuss the test case, including considering how traditional programming tools would have fared with the issues raised in subsection 4.3. 4. 1 Penny The Penny system is a parallel implementation of the Agents Kernel Language (AKL) [15, 24], a language for Concurrent Constraint Programming (CCP) CCP views an execution as a set of concurrently executing agents that communicate through a shared constraint store. The language is similar to a data flow language, the difference being that the agents wait for constraints to be ....

.... is the interrupt [13] 2] from 0xf1900000 to 0xf1900ffe is the auxio [14] 1] from 0xf1d00000 to 0xf1d03ffe is the counter [15] 890] from 0xf1d10000 to 0xf1d10ffe is the counter [16] 3807] from 0xfec01000 to 0xfec01ffe is the tty [17] 1] from 0xfec02000 to 0xfec02ffe is the info [18] [ 24] from 0xfec03000 to 0xfec03ffe is the elf loader (gdb simics) io 1 19714: cpu0 ip=0xf0084e54 st data=0x00000210 addr=0x08400000 dma 0x0 19713: cpu0 ip=0xf0084e4c ld data=0x00000200 addr=0x08400000 dma 0x0 19712: cpu0 ip=0xf0084e38 st data=0xfff00000 addr=0x08400004 dma 0x4 19711: cpu0 ....

J. Montelius. Exploiting Fine-grain Parallelism in Concurrent Constraint Languages. PhD thesis, Uppsala Thesis in Computing Science 28 / SICS Dissertation Series 25.

....This would indicate a speedup of 15, in contrast to the actual speedup of 7:7. The study shows how important good cache performance is to a system such as Penny, and demonstrates how an instruction set simulator can relate cache events to program code. 1. 1 The Penny system The Penny system [12] is a implementation of AKL [7] on a shared memory architecture. It will automatically extract parallelism in an AKL program. No user annotations are required, thus relieving the programmer from adding explicit information to control parallelism. The system can utilize both and and ....

J. Montelius. Exploiting fine-grain parallelism in concurrent constraint languages. Uppsala Thesis in Computing Science 28, SICS Dissertation Series 25, Uppsala University, SICS, 1997.

No context found.

) Montelius, J. Exploiting Fine-grain Parallelism in Concurrent Constraint Languages. PhD thesis, Uppsala University, 1997.

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