Basilio B. Fraguela, Ramón Doallo, and Emilio L. Zapata. "Modeling Set Associative Caches Behavior for Irregular Computations". In SIGMETRICS '98/PERFORMANCE '98: Proceedings of the 1998. Home/Search Document Details and Download
Summary Related Articles Check |
This paper is cited in the following contexts:
Modelling Data Locality for the Sparse Matrix-Vector.. - Heras, Cabaleiro, Rivera (2001) (Correct)
....need to take care of the details of the irregular computations. For the case of the sparse matrix by a dense vector product (SpM Theta V ) as an example of irregular code, the locality in the accesses measured through the number of cache misses is characterized through analytical modelling in [20,7]. Unlike the work described above, we develop a more general model because we model the locality (spatial plus temporal) in the execution of the code as defined in Sect. 1, we do not consider the movement of data between levels or the reuse. For this reason our model takes into account only the ....
B. B. Fraguela, R. Doallo, and E. L. Zapata. Modeling set associative caches behavior for irregular computations. ACM Performance Evaluation Review (Proc. SIGMETRICS/PERFORMANCE98), 26(1):192--201, June 1998.
A Compiler Tool to Predict Memory Hierarchy.. - Fraguela, Doallo, .. (2004) (3 citations) Self-citation (Fraguela Doallo Zapata) (Correct)
No context found.
B.B. Fraguela, R. Doallo, E.L. Zapata, Modeling set associative caches behavior for irregular computations, ACM Performance Evaluation Review (Proc. SIGMETRICS/PERFORMANCE'98) 26 (1) (1998) 192--201.
Memory Hierarchy Performance Prediction for Blocked.. - Fraguela, Doallo, Zapata (1999) (1 citation) Self-citation (Fraguela Doallo Zapata) (Correct)
....that fit hardware improvements of current high performance microprocessors. As an example, an optimized version of the sparse matrix dense matrix product described in [10] is modeled. In this work we apply the probabilistic model for K way associative caches with LRU replacement introduced in [11], where only simple algebra kernels such as sparse matrix vector product and sparse matrix transposition were considered. Optimum block sizes for the memory hierarchy of an arbitrary system can be predicted by means of this model. A uniform distribution of the non zero elements is considered, but ....
....received K Gamma i lines of this program vector during the execution of a given portion of code. The exception is SV 0 , which is the ratio of sets that have received K or more lines. Different expressions have been developed to calculate the area vectors associated to typical access patterns [11]. Next, we describe those ones that arise in our particular code in Figure 1. 3.1 Access Patterns Modeled One of the most common patterns is the sequential access to n consecutive words, whose area vector S s (n) is S s (K Gammablc) n) 1 Gamma (l Gamma blc) S s (K Gammablc Gamma1) n) l ....
B.B. Fraguela, R. Doallo and E.L. Zapata, Modeling Set Associative Caches Behavior for Irregular Computations, in Proc. ACM Sigmetrics/Performance Joint International Conf. on Measurement and Modeling of Computer Systems, Madison, Wisconsin, June 1998, 192--201.
Automatic Analytical Modeling for the Estimation of Cache.. - Fraguela, Doallo, Zapata (1999) (12 citations) Self-citation (Fraguela Doallo Zapata) (Correct)
....the accesses to V during a given period of the program execution. The element in the i th position of the vector stands for the ratio of sets that have received K Gamma i lines of this structure. Only SV0 has a different meaning, as it is the ratio of sets that have received K or more lines. In [7] and [6] we show different expressions that can be used to calculate the vectors associated with typical access patterns. Two kinds of interference area vectors are considered, depending on the source of the interference. When the interference is generated by the accesses to the data structure ....
....least one reuse loop for the references to matrix A or the structures referenced and the order in which these references are performed (mainly those of A) is similar in loops j(k Gamma 1) and jk. Otherwise precision can be seriously affected, requiring an analysis similar to the one presented in [7] for the sparse matrix transposition. In this work the hit probability in the reuse is calculated for each line of one vector, studying the accesses that take place between its last access in the preceding loop and its first access in the analyzed loop. For the references inside loop j0, F j (R; ....
[Article contains additional citation context not shown here]
B. B. Fraguela, R. Doallo, and E. L. Zapata. Modeling set associative caches behavior for irregular computations. ACM Performance Evaluation Review (Proc. SIGMETRICS/PERFORMANCE'98), 26(1):192--201, June 1998.
Cache Misses Prediction for High Performance Sparse.. - Fraguela, Doallo, Zapata (1998) (1 citation) Self-citation (Fraguela Doallo Zapata) (Correct)
....to more complex algorithms, such as the one mentioned above. This fact can make possible further improvements for these codes by making easier the study of the effect of these techniques on the memory hierarchy. An extension of the proposed model for K way associative caches has been introduced in [3]. The remainder of the paper is organized as follows: Section 2 describes the algorithm to be modeled. Basic model aparameters and concepts are introduced in Sect. 3 together with a brief explanation, due to space limitations, of the modeling process. In Sect. 4 the model is validated and used to ....
Fraguela, B.B, Doallo, R., Zapata, E.L.: Modeling Set Associative Caches Behavior for Irregular Computations. To appear in Proc. ACM Sigmetrics/Performance Joint Int'l. Conf. on Measurement and Modeling of Computer Systems, Madison, Wisconsin, (1998).
Set Associative Cache Behavior Optimization - Doallo, Fraguela, Zapata (1999) Self-citation (Fraguela Doallo Zapata) (Correct)
....K Gamma i lines of the structure. The first position, SV0 , has a different meaning: it is the ratio of sets that have received K or more lines. Different expressions have been devel oped that estimate the area vectors as a function of the access pattern of the corresponding data structure [2]. The formulae and algorithms associated to the sequential access and the access to a number of regions of consecutive memory words which have a constant distance between the start of two such regions (access with a constant stride) may be found in [3] A mechanism for adding the area vectors ....
....in order to optimize modeling times. The whole regions of the vectors and matrices accessed during the execution of the outer non reuse loop including a reference to the studied data structure are used to compute the interference area vectors. More accurate approaches have been already developed [2]. 4 Validation and Applications A simple automatic analyzer based in our modeling strategy has been built in order to validate it. Function calls are used to describe the loops, data structures and references to the analyzer, just as the semantic stage of a code analyzer could do. Its predictions ....
[Article contains additional citation context not shown here]
Fraguela, B.B., Doallo, R., Zapata, E.L.: Modeling set associative caches behavior for irregular computations. ACM Performance Evaluation Review (Proc. SIGMETRICS /PERFORMANCE'98) 26(1) (1998) 192--201
Analysis and Evaluation of The Synchronized - Pipelined Parallelism Model (2006) (Correct)
No context found.
Basilio B. Fraguela, Ramón Doallo, and Emilio L. Zapata. "Modeling Set Associative Caches Behavior for Irregular Computations". In SIGMETRICS '98/PERFORMANCE '98: Proceedings of the 1998.
Efficient and Accurate Analytical Modeling of Whole-Program Data .. - Xue, Vera (2003) (Correct)
No context found.
B. B. Fraguela, R. Doallo, and E. L. Zapata. Modeling set associative caches behavior for irregular computations. ACM Performance Evaluation Review, 26(1):192--201, June 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