D.W. Jones. \An empirical comparison of priority{queue and event{set implementations". Comm. ACM, 29(4):300-311, 1986.  Home/Search   Document Not in Database   Summary   ACM   TOC   Related Articles   Check

.... January 23, 2001 1 Introduction Binary search trees such as red black trees, AVL trees, and splay trees have been developed for the representation of dictionaries ( 5] for example) Suitably modified versions of these data structures result in the efficient implementation of single ended [3] and double ended priority queues. The number of applications of binary search trees that permit fast search, insert, and delete capability is sufficiently large that a red black tree implementation is provided in the C Standard Templates Library as well as in Java s java.utl package (the class ....

....these n searches is measured. Note that some of these n searches will be unsuccessful (i.e. searches for keys that are not in the dictionary) Finally, the n pairs in the dictionary are removed in some random order, and the time for all n deletions is measured. 2. Hold Model] In the hold model [3], as can be seen in Figure 14(b) we start with a dictionary that has n pairs and perform an intermixed sequence of insert and delete operations. This sequence Srand (a) Base lin Dd (c) Stack Figure 14: Four test models (I: Insert, S: Search, D: Delete, rand: sequence, dec: decreasing ....

D. Jones, "An empirical comparison of priority-queue and event-set implementation," Communications of the Association for Computing Machinery, 29(4), pp. 300-311, 1986.

....event of a neuron can be easily found at the top of the neuron s local event queue. Note that this change of queuing model does not increase the order of scheduling cost. The complete binary tree (heap tree) algorithm, which is a most popular and empirically e#cient algorithm for a priority queue [21], needs a cost of O( log n) for insertion and retrieval of an entry, where n is the entries in the queue. Suppose a neural network has N neurons and each neuron has # pending events. In the single queue model, the insertion retrieval cost is O( log #N ) On the other hand, in the object queue ....

D. W. Jones. An empirical comparison of priority queue and event set implementations. Communications of the ACM, 29:300--311, 1986. (cited in page 82)

....cant speedups can be achieved by using layouts optimized for the memory hierarchy see e.g. the paper by Chatterjee et al. 8] and the references it contains. LaMarca and Ladner consider the question in connection with heaps [16] Among other things, they repeat an experiment performed by Jones [15] ten years earlier, and demonstrate that due to the increased gaps in access time between levels in the memory hierarchy, the d ary heap has increased competitiveness relative to the pointer based priority queues. For search trees, B trees are the standard way to implement trees optimized for the ....

D. W. Jones. An Empirical Comparison of Priority-Queue and Event-Set Implementations. Communications of the ACM, 29(4):300-311, 1986.

....of a neuron can be easily found at the top of the neuron s local event queue. Note that this change of queuing model does not increase the order of scheduling cost. The complete binary tree (heap tree) algorithm, which is a most popular and empirically efficient algorithm for a priority queue [11], needs a cost of O(logn) for insertion and retrieval of an entry, where n is the entries in the queue. Suppose a neural network has N neurons and each neuron has n pending events. In the single queue model, the insertion retrieval cost is O(lognN) On the other hand, in the object queue model, we ....

D. W. Jones. An empirical comparison of priority queue and event set implementations. Communications of the ACM, 29:300--311, 1986.

....speedups can be achieved by using layouts optimized for the memory hierarchy see e.g. the paper by Chatterjee et al. 8] and the references it contains. LaMarca and Ladner consider the question in connection with heaps [16] Among other things, they repeat an experiment performed by Jones [15] ten years earlier, and demonstrate that due to the increased gaps in access time between levels in the memory hierarchy, the d ary heap has increased competitiveness relative to the pointer based priority queues. For search trees, B trees are the standard way to implement trees optimized for the ....

D. W. Jones. An Empirical Comparison of Priority-Queue and Event-Set Implementations. Communications of the ACM, 29(4):300--311, 1986.

....and is at least equally well tuned. We have chosen implicit binary heaps and aligned 4 ary heaps. In a recent study [15] these two algorithms outperform the pointer based data structures splay tree and skew heap by more than a factor two although the latter two performed best in an older study [12]. Not least because we need the same code for the insertion buffer, binary heaps were coded perhaps even more carefully than the remaining components binary heaps are the only part of the code for which we took care that the assembler code contains no unnecessary memory accesses, redundant ....

D. Jones. An empirical comparison of priority-queue and event set implementations. Communications of the ACM, 29(4):300--311, 1986.

....menu of allowable values are easy to implement, e.g. using combinational priority encoder circuits. However, for priority queues with many thousand entries and with values drawn from a large set of allowable numbers, heap or calendar queue data structures must be used. Other heap like structures [Jones86] are interesting in software but are not adaptable to high speed hardware implementation. 70 79 60 69 50 59 40 49 30 39 20 29 10 19 0 9 L4 L3 L2 L1 (b) a) 90 99 80 89 mod 100 L1 L2 L3 L4 104 37 32 99 55 99 125 55 32 57 56 37 81 30 125 104 30 56 30 55 57 81 99 125 32 57 ....

D. Jones: "An Empirical Comparison of Priority-Queue and Event-Set Implementations ", Commun. of the ACM, vol. 29, no. 4, Apr. 1986, pp. 300-311.

....menu of allowable values are easy to implement, e.g. using combinational priority encoder circuits. However, for priority queues with many thousand entries and with values drawn from a large set of allowable numbers, heap or calendar queue data structures must be used. Other heap like structures [7] are interesting in software but are not adaptable to high speed hardware implementation. L3 L2 L1 L4 L1 L2 L3 L4 125 55 32 57 56 37 81 30 125 30 55 32 57 56 37 81 104 99 99 104 Figure 1: Heap priority queue Fig. 1 illustrates a heap. It is a binary tree (top) physically stored in ....

D. Jones: "An empirical comparison of priority-queue and event-set implementations", Commun. of the ACM, vol. 29, no. 4, Apr. 1986, pp. 300-311.

....speedups can be achieved by using layouts optimized for the memory hierarchy see e.g. the paper by Chatterjee et al. 8] and the references it contains. LaMarca and Ladner consider the question in connection with heaps [16] Among other things, they repeat an experiment performed by Jones [15] ten years earlier, and demonstrate that due to the increased gaps in access time between levels in the memory hierarchy, the d ary heap has increased competitiveness relative to the pointer based priority queues. For search trees, B trees are the standard way to implement trees optimized for the ....

D. W. Jones. An Empirical Comparison of PriorityQueue and Event-Set Implementations. Communications of the ACM, 29(4):300--311, 1986.

....and used in this chapter [55, 54] 4.3 Governor Heap Among current scheduling algorithms, the smallest timestamp first method can reduce rollbacks to a reasonable level. The most commonly used data structure to support the method is either a priority queue based on a splay tree data structure [48, 100] or the Calendar queue [13] More detail in performance comparison can be found in [85] In either case, the events for different logical processes (gates flip flops) are mixed together as long as those processes are assigned to the same processor. Note that for a typical simulation, a processor ....

D. W. Jones, "An Empirical Comparison of Priority-Queue and Event-Set Implementations, " Communications of the ACM, Vol. 29, No. 4, pp. 300-311, April 1986.

....menu of allowable values are easy to implement, e.g. using combinational priority encoder circuits. However, for priority queues with many thousand entries and with values drawn from a large set of allowable numbers, heap or calendar queue data structures must be used. Other heap like structures [Jones86] are interesting in software but are not adaptable to high speed hardware implementation. Figure 2 (a) illustrates a heap. It is a binary tree (top) physically stored in a linear array (bottom) Non empty entries are pushed all the way up and left. The entry in each 1 in arbitrary units; there ....

D. Jones: "An Empirical Comparison of Priority-Queue and Event-Set Implementations ", Commun. of the ACM, vol. 29, no. 4, Apr. 1986, pp. 300-311.

....adding information to the tree representation or 7 using a more intricate algorithm in which an item to be deleted is merely marked as deleted, the actual deletion is carried out during subsequent Min operations. Leftist trees were eliminated from this analysis because it has been shown by Jones [6] that they are consistently about 30 percent slower than implicit heaps. 2.4 B Trees B Trees, introduced by Bayer and McCreight [7] are a generalization of the binary search tree. They maintain the linear left right order condition on the node that binary search trees possess, while maintaining ....

....than leftist or AVL trees but are complex to code. Binomial queues were left out from the experiments here because Brown [11] has shown that heaps are slightly faster than binomial queues on the average and considerably faster in the worst case. These results were reopened and questioned in [6], where it was claimed that binomial queues are nearly optimal. 2.8 Pagodas Pagodas [12] like leftist trees, are based on heap ordered binary trees. However, unlike leftist trees, in a Pagoda the primary pointers lead from the leaves of the tree towards the root. A secondary pointer in each ....

[Article contains additional citation context not shown here]

Jones, D.W., An Empirical Comparison of Priority-Queue and Event-Set Implementations. Communications of the ACM, 1986. 29(4): p. 300-311.

....kept a low level of experimentation active, but did not attempt to provide methodological pointers. More recent and comprehensive work began with Bentley s many contributions in his Programming Pearls (starting in 1983 [Ben83] then with Jones comparison of data structures for priority queues [Jon86] and Stasko and Vitter s combination of analytical and experimental work in the study of pairing heaps [SV87] An early experimental study on a large scale was that of Moret and Shapiro on sorting algorithms [MS91a] Chapter 8) itself inspired by the work of Knuth in his Volume III [Knu98] ....

D. W. Jones, An empirical comparison of priority queues and event-set implementations, Commun. ACM 29 (1986), 300--311.

....of the operations. For this we use locks and the interface to the locks has to provide both blocking and non blocking locking functions. We also assume that the processes can pass messages asynchronously. To compare di#erent priority queues both theoretically and practically the hold model [10] has been used. In this model a priority queue of size N is created and a hold operation is performed a number of times. The hold operations is a sequence of min; deleteMin; and insert operations, hence N is not changed. The priority of the newly inserted element is t 0 d for some value d. In ....

Douglas W. Jones. An empirical comparison of priority-queue and event-set implementations. Communications of the ACM, 29(4):300--311, April 1986.

.... [19] attempts at characterizing the e#ects of caching on the behavior of implementations [1, 11, 16 18, 27, 29] ever more e#cient implementation of network flow algorithms [7, 8, 13] and the characterization of the behavior of everyday algorithms and data structures such as priority queues [15, 32], shortest paths [6] minimum spanning trees [22] and sorting [21] More references can be found in [20] as well as by going to the web site for the ACM Journal of Experimental Algorithmics at www.jea.acm.org. High Performance Algorithm Engineering High Performance Algorithm Engineering focuses ....

Jones, D.W., "An empirical comparison of priority queues and event-set implementations, " Commun. ACM 29 (1986), 300--311.

....problems (e.g. TSP) For many applications, priority queue operations (especially deleting the minimum or maximum key) are the dominant factor in the performance of the application. Priority queues have been extensively analyzed and numerous implementations and empirical studies exist; see, e.g. [AHU74, Meh84, Jon86, SV87, LL96]. This note presents an experimental evaluation of one of the priority queues recently introduced in [MVY94] the (word based) radix tree. This data structure is qualitatively different than traditional priority queues in that its time per operation does not increase as the number of keys grows ....

D. W. Jones. An empirical comparison of priority queue and event set implementations.

....for a new data structure based on calendar queues [1] as used in sequential simulation. It is obtained that these new data structure, called Time Warp calendar queue or TWCQ , performs substantially better than the normally used DLL. 2 Time Warp Calendar Queues Besides other data structures [3], calendar queues were developed to implement the set of pending events in sequential simulation. Unlike most other, tree based data structures, calendar queues showed an O(1) average case behavior. A calendar queue is organized like a desk calendar with one page for each day of the year. Each ....

....latency) as possible) All measurements were repeated thirty times with different random seeds. Any two tests for DLLs and TWCQs were performed with the same random seeds. All measured operations were performed 1000 times. We chose the same priority distributions and queue sizes as in [3, 1]. 0.01 0.1 1 10 1 10 100 1000 10000 Time(ms) Queue Size DLL (Exponential) DLL (Uniform) DLL (Biased) DLL (Bimodal) DLL (Triangular) TWCQ (Exponential) TWCQ (Uniform) TWCQ (Biased) TWCQ (Bimodal) TWCQ (Triangular) Figure 1: Hold Times Figure 1 shows that the good performance of ....

D. W. Jones. An empirical comparison of priorityqueue and event-set implementations. Commun. ACM, 29(4):300--311, Apr. 1986.

No context found.

D.W. Jones. \An empirical comparison of priority{queue and event{set implementations". Comm. ACM, 29(4):300-311, 1986.

No context found.

Douglas W. Jones. An empirical comparison of priority-queue and event-set implementations. Comm. of the ACM, 29(4):300--311, 1986.

No context found.

D.W. Jones. "An empirical comparison of priority--queue and event--set implementations". Comm. ACM, 29(4):300--311, 1986.

No context found.

D. W. Jones, An empirical comparison of priority-queue and event-set implementations, Comm. ACM 29 (1986), 300#311.

No context found.

D. W. Jones, `An empirical comparison of priority-queue and event-set implementations', Communications of the ACM, 29(4), 300--311 (1986).

No context found.

D. Jones. An empirical comparison of priority queues and event-set implementations. Communications of the ACM, 29:300-311, 1986.

No context found.

Jones, D. W., "An Empirical Comparison of Priority Queue and Event Set Implementations," Communications of the ACM, 29 (1986), 300-311.

No context found.

Jones, D. W. An empirical comparison of priority-queue and event-set implementations. Commun. ACM 29, 4 (April 1986).

First 50 documents  Next 50

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