I. Martel, D. Ortega, E. Ayguade, and M. Valero. Increasing effective IPC by exploiting distant parallelism. In Proc. Int'l conf. on Supercomputing, pages 348--355, 1999. Home/Search Document Details and Download
Summary Related Articles Check |
This paper is cited in the following contexts:
A Feasibility Study of Hierarchical Multithreading - Zahran, Franklin (2002) (Correct)
....the ongoing quest for faster execution of programs. There is broad consensus that barring the use of radically novel technologies such as quantum computing and biological computing, the key to further progress in this quest is to do parallel processing at different granularities. Many studies [7][9][13] have confirmed that a lot of parallelism exists at different granularities. The commodity microprocessor industry has been traditionally looking to fine grained or instruction level parallelism (ILP) for improving performance, by means of sophisticated microarchitectural techniques and ....
I. Martel, D. Ortega, E. Ayguade, and M. Valero, "Increasing Effective IPC by exploiting Distant Parallelism," in Proc. Int'l Conf. on Supercomputing, pp. 348--355, 1999.
Increasing Effective IPC by Exploiting Distant.. - Martel, Ortega.. (1999) (2 citations) Self-citation (Martel Ortega Ayguad'e Valero) (Correct)
....(statically in the original source code and or dynamically when they are executed) Later in the next section we focus on the parallelisation of some non numerical applications from SPEC. Additional results for some numerical SPEC applications can be found in the extended version of this paper [10]. 2.1 Numerical applications Existing compiler techniques for finding parallelism in numerical applications refer primarily to loops. Different techniques have been proposed to analyse and transform codes in order to make loops totally parallel. Although other levels of parallelism can exist in ....
....the level of different matrices, uxw has parallelism at the level of loops. Similarly, the routines invoked to implement the time stepping scheme also have parallelism at the level of loops. Figure 1 summarises the parallelism structure of this application. In the extended version of this paper [10] we show the possible speed ups for this application when combining different levels of parallelism. The highest speed up achieved is 391.05 when all levels of parallelism are opened. Nevertheless, most of this speed up (309.78) is achieved when two levels are opened, the highest one at the ....
I. Martel, D. Ortega, E. Ayguad'e, and M. Valero. Increasing effective ipc by exploiting distant parallelism. Technical Report UPC-DAC-1998-59, Departmento de Arquitectura de Computadores, Universidad Polit'ecnica de Catalu~na--Barcelona, December 1998.
Quantifying the Benefits of SPECint Distant.. - Ortega, Martel.. Self-citation (Martel Ortega Ayguade Valero) (Correct)
....In this paper we will show that non numerical applications have inherent parallelism, and that reasonable performance gains can be expected from exploiting it. During the analysis of different non numerical applications, we have found that they posses lots of semantic thread level parallelism [11], i.e. zones of code representing different computations which not necessarily must be done in a sequential order. However, semantic parallelism is difficult to find automatically, for many times the programmer or even the compiler may introduce dependencies among these parallel zones in the ....
....we are going to explain briefly the amount of code parallelised, the average number of threads in each of the zones parallelised and the theoretical speed ups obtainable without considering any problems in the simulations. For a deeper explanation on the parallelisations, please refer to [11]. compress95 has been divided in two different benchmarks, that comprise the compression and the decompression phase. Both of them were tested with a normalised data input, that covers one cycle of compression, the amount of data between two cleanings of the hash table used to compress. This ....
I. Martel, D. Ortega, E. Ayguade, and M. Valero. Increasing effective ipc by exploiting distant parallelism. International Conference on Supercomputing, June 1999.
Dynamic Thread Resizing for Speculative Multithreaded Processors - Zahran, Franklin (Correct)
No context found.
I. Martel, D. Ortega, E. Ayguade, and M. Valero. Increasing effective IPC by exploiting distant parallelism. In Proc. Int'l conf. on Supercomputing, pages 348--355, 1999.
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