P. A. Wilsey, D. A. Hensgen, N. B. Abu-Ghazaleh, C. E. Slusher, and D. Y. Hollinden. The concurrent execution of non-communicating programs on SIMD processors. Proc. of the 4th Symposium on the Frontiers of Massively Parallel Computation, pages 29--36, October 1992.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....cycle, then an MIMD like program could efficiently execute on the SIMD machine [17] The instruction level approach implements this idea directly. That is, the instructions are interpreted in parallel across all of the processors by control signals emanating from the central control unit [41]. The major constraint of this approach is that the central control unit has to cycle through almost the entire instruction set for each instruction execution because each processor may execute different instructions. Furthermore, this approach must insert proper synchronization to ensure correct ....

....of logic circuits on a SIMD machine [6, 20] The major difference between these works and our approach is that we handle general asynchronous and loosely synchronous problems instead of studying individual problems. The instruction level approach has been studied by a number of researchers [8, 9, 22, 25, 41, 40]. As mentioned above, the major restriction of this approach is that the entire instruction set must be cycled through to execute one instruction step for every processor. A common method to reduce the average number of instructions emanated in each execution cycle is to perform global or s to ....

[Article contains additional citation context not shown here]

P. Wilsey, D. Hensgen, N. Abu-Ghazaleh, C. Slusher, and D. Hollinden. The concurrent execution of non-communicating programs on SIMD processors. In The Fourth Symposium on the Frontiers of Massively Parallel Computation, October 1992.

.... instruction cycle, then an MIMD like program could efficiently execute on the SIMD machine [8] The instruction level approach implements this idea directly: the instructions are interpreted in parallel across all of the processors by control signals emanating from the central control unit [3, 4, 12, 14, 23]. The major constraint of this approach is that the central control unit has to cycle through almost the entire instruction set for each instruction execution because each processor may execute different instructions. A common method to reduce the average number of instructions emanated in each ....

....entire instruction set for each instruction execution because each processor may execute different instructions. A common method to reduce the average number of instructions emanated in each execution cycle is to perform global or s to determine whether the instruction is needed by any processor [4, 23]. It was believed that a program would not diverge quickly and only a few instructions need to be emanated [4] However, later research found that the programs diverged fast and the global or solution did not deliver much benefit [23] It might be necessary to insert barrier synchronizations at ....

[Article contains additional citation context not shown here]

P. Wilsey, D. Hensgen, N. Abu-Ghazaleh, C. Slusher, and D. Hollinden. The concurrent execution of noncommunicating programs on SIMD processors. In The Fourth Symposium on the Frontiers of Massively Parallel Computation, October 1992.

....also looked into loop transformations that help optimize the node level programs when a BLOCK distribution is used [18] 2.1. 5 MIMD Emulators There has been growing interest in determining if SIMD architectures can successfully be used to handle problems that do not fit the data parallel model [22, 37, 60, 66]. One method is to emulate MIMD execution by writing a SIMD program to interpret a MIMD instruction set. Each PE contains a program to be interpreted and its data. The front end then executes an interpreter loop where the instruction decode step iterates over all possible instruction types ....

P. Wilsey, D. Hensgen, N. Abu-Ghazaleh, C. Slusher, and D. Hollinden. The concurrent execution of non-communicating programs on SIMD processors. In Frontiers '92: The 4th Symposium on the Frontiers of Massively Parallel Computation, McLean, VA, October 1992.

.... of instructions chosen from a common instruction set and the control unit interprets that instruction set concurrently (one instruction at a time, across all the PEs that require it) Collins, 1988, Dietz and Cohen, 1992a, Littman and Metcalf, 1988, Nilsson and Tanaka, 1988b, Shu and Wu, 1995, Wilsey et al. 1992 ] Every iteration through the interpreter program allows all the PEs to advance their instruction streams one instruction regardless of the number of PEs. Unfortunately, these concurrent interpretation models are only able to achieve a small fraction of the peak machine throughput. Among ....

....characteristics of an efficient concurrent interpretation environment. Concurrent Interpretation on SIMD machines was developed concurrently (and independently) by several researchers around the same time [ Collins, 1988, Dietz and Cohen, 1992a, Littman and Metcalf, 1988, Nilsson and Tanaka, 1988b, Wilsey et al. 1992 ] Throughout the discussion of concurrent interpretation (this chapter, as well as the following few chapters) the experiments and optimizations developed by our own studies are presented, and references to similar or related optimizations are made when appropriate. 6.1 What is Concurrent ....

[Article contains additional citation context not shown here]

P. A. Wilsey, D. A. Hensgen, N. B. Abu-Ghazaleh, C. E. Slusher, and D. Y. Hollinden. The concurrent execution of non-communicating programs on SIMD processors. Proc. of the 4th Symposium on the Frontiers of Massively Parallel Computation, pages 29--36, October 1992.

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