Rajive Bagrodia, Stephen Docy, and Andy Kahn. "Parallel Simulation of Parallel File Systems and I/O Programs." In Proceedings of 1997 Supercomputing Conference, San Jose, CA, 1997. To appear.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....in a multi threaded environment without changing the programming interface. Most of gave up this design due to insufficient address space and software incompatibility [27] previous MPI research is focused on distributed memory machines or workstation clusters, e.g. 9] The MPISIM project [6, 7] has used multi threading to simulate MPI execution on distributed memory machines as we will discuss in Section 3.1. Thread safety is addressed in [3, 26, 29] However, their concern is how multiple threads can be invoked in each MPI node, but not how to execute each MPI node as a thread. These ....

R. Bagrodia, S. Docy, and A. Kahn. Parallel Simulation of Parallel File Systems and I/O Programs. In Proc. of SuperComputing '97.

....the behavior of the application on a particular configuration of the machine, and cannot be used when the machine configuration is changed. An alternative approach to using traces or analytical models is to use simulators that run application programs. A variety of such simulators are available [3, 10, 13, 14]. In general, application source or binary codes must be augmented or the application is required to use the simulator API so that events can be passed to the simulator so that simulated time can progress. In order to increase the efficiency of simulation, most simulators use the direct execution ....

....9177 5032 136 498 4096 18028 16388 18274 24562 142 1481 8192 36035 77641 36437 65137 148 4873 Table 3: Execution times for loosely coupled simulation. Both estimated and execution times are in seconds. models for less accurate but fast simulation, or use parallel machines to run the simulators [3, 13]. Our work differs from previous work in several ways. In our work we specifically target large scale data intensive applications on large scale machines. The application emulators presented in this paper lie in between pure analytical models and full applications. They provide a simpler, but ....

R. Bagrodia, S. Docy, and A. Kahn. Parallel simulation of parallel file systems and I/O programs. In Proceedings of the 1997 ACM/IEEE SC97 Conference. ACM Press, Nov. 1997.

....[34] and TPVM [12] projects. Both projects do not address how a PVM program can be executed in a multi threaded environment without changing the programming interface. Most of previous MPI research is focused on distributed memory machines or workstation clusters, e.g. 9] The MPI SIM project [6, 7] has used multithreading to simulate MPI execution on distributed memory machines as we will discuss in Section 3.1. Thread safety is addressed in [3, 26, 29] However, their concern is how multiple threads can be invoked in each MPI node, but not how to execute each MPI node as a thread. These ....

R. Bagrodia, S. Docy, and A. Kahn. Parallel Simulation of Parallel File Systems and I/O Programs. In Proc. of SuperComputing'97.

....component for the target architecture s file system can be used. In this scenario, this layer can simulate a parallel file system s cnodes, ionodes, disks, and caching strategies. This larger, and more complete simulation system is beyond the scope of this thesis. The reader is referred to [BDK97] for more details. 36 CHAPTER 5 Experiments and Results 5.1 Benchmarks The IBM SP2 at UCLA was selected both as the target machine as well as the host machine. Each node of the SP2 is a POWER2 node with 128Kb of cache, and 256Mb of main memory. The nodes are connected with a high performance ....

....other communication used in collective I O. To correctly simulate the effects of disk contention, a more detailed disk model that simulates the queuing and contention at the disk, must be used. A study using this model along with modeling a parallel file system and caching effects can be found in [BDK97]. 43 CHAPTER 6 Conclusion In this thesis, we have presented the design and implementation issues of a simulation library for parallel I O programs which use MPI IO. Using the simulator, we show good speedups from parallel execution. We have also shown the usefulness and extensibility of the ....

[Article contains additional citation context not shown here]

Rajive Bagrodia, Stephen Docy, and Andy Kahn. "Parallel Simulation of Parallel File Systems and I/O Programs." In Proceedings of 1997 Supercomputing Conference, San Jose, CA, 1997. To appear.

....as a function of various hardware and system software characteristics that include the number of processors, interconnection network characteristics, or message passing library implementations. The simulator can also be used to evaluate the performance of parallel file systems and I O systems [5]. Supported capabilities include a number of different disk caching algorithms, collective I O techniques, disk cache replacement algorithms, and I O device models. MPI SIM simulates the application program by using direct execution to simulate local code fragments and parallel discrete event ....

Bagrodia, R., S. Docy, and A. Kahn, "Parallel Simulation of Parallel File Systems and I/O Programs", Proc. Supercomputing '97, San Jose, 1997.

.... include validated state of the art processor and memory hierarchy models based on SimpleScalar [14] interconnection network models using the PARSEC parallel simulation language [9] large scale parallel program simulations using the MPI Sim simulator [28,10] and parallel I O system simulators [8]. A unified application representation based on a combination of static and dynamic task graphs has been developed that can serve as a common workload representation for this wide range of performance models. Ongoing research within POEMS, is developing techniques to integrate subsets of two or ....

....as a function of various hardware and system software characteristics that include the number of processors, interconnection network characteristics, and message passing library implementations. The simulator also can be used to evaluate the performance of parallel file systems and I O systems [8]. Supported capabilities include a number of different disk caching algorithms, collective I O techniques, disk cache replacement algorithms, and I O device models. The parallel discrete event simulator uses a set of conservative synchronization protocols together with a number of optimizations to ....

Bagrodia, R., S. Docy, and A. Kahn, "Parallel Simulation of Parallel File Systems and I/O Programs", Proc. Supercomputing '97, San Jose, 1997.

....processor SP benchmarks and using 16 host processors) Recently, MPI SIM was extended to model aspects of the computer architecture beyond those of the communication system. Detailed models of I O systems, parallel le systems, and I O data caching and placement algorithms were added to MPI SIM[BDK97] 7 Acknowledgments All the data presented in this paper was collected on the IBM SP2 at UCLA s Oce of Academic Computing, granted to UCLA by IBM Corporation under their Shared University Research Program. We are thankful to the members of the SESAME project at UCLA for their comments and ....

R. Bagrodia, S. Docy, and A. Kahn. Parallel Simulation of Parallel File Systems and I/O Programs . In SuperComputing, 1997.

....processor SP benchmarks and using 16 host processors) Recently, MPI SIM was extended to model aspects of the computer architecture beyond those of the communication system. Detailed models of I O systems, parallel file systems, and I O data caching and placement algorithms were added to MPI SIM[BDK97] 7 Acknowledgments All the data presented in this paper was collected on the IBM SP2 at UCLA s Office of Academic Computing, granted to UCLA by IBM Corporation under their Shared University Research Program. Special thanks to M. Dhagat for implementing the data parallel programs and to S. Docy ....

R. Bagrodia, S. Docy, and A. Kahn. Parallel Simulation of Parallel File Systems and I/O Programs . In SuperComputing, 1997.

.... state of the art processor and memory hierarchy models based on SimpleScalar [Burger97] interconnection network models using the PARSEC parallel simulation language [Bagrodia98] large scale parallel program simulations using the MPI SIM simulator [Prakash98] and parallel I O system simulators [Bagrodia97, Bagrodia 98b] A unified application representation based on a combination of static and dynamic task graphs is being developed that can serve as a common workload representation for this wide range of performance models [Adve99a] In ongoing research within POEMS, we are developing techniques ....

....as a function of various hardware and system software characteristics that include the number of processors, interconnection network characteristics, and message passing library implementations. The simulator also can be used to evaluate the performance of parallel file systems and I O systems [Bagrodia97] Supported capabilities include a number of different disk caching algorithms, collective I O techniques, disk cache replacement algorithms, and I O device models. The parallel discrete event simulation uses a set of conservative synchronization protocols together with a number of optimizations ....

Bagrodia, R., S. Docy, and A. Kahn, "Parallel Simulation of Parallel File Systems and I/O Programs", Proc. Supercomputing '97, San Jose, 1997, PAGE NOS.

....in parallel. All communication is performed using a communicator which describes the group of communicating processes. Only member processes may use a given communicator. This paper assumes that the program does not have any I O commands; simulation of the I O constructs is described in Bagrodia et al. (1997). In the subset of MPI we simulate, all collective communication functions are implemented in terms of point to point communication functions, and all point to point communication functions are implemented using a set of core nonblocking MPI functions. The core functions include MPI Issend, a ....

Bagrodia, R., S. Docy, and A. Kahn, Parallel Simulation of Parallel File Systems and I/O Programs. In Supercomputing 97, 1997.

....parallel file system library designed to sit on top of multiple UFS instances and provide a wide variety of parallel file system capabilities, such as caching, prefetching, and data distribution. The COMPASS environment described in this paper used the parallel I O system simulator detailed in [1] and is perhaps the only simulator that combines the ability to do integrated interconnection network, I O and file system, and scalability studies. It has also been used for the simulation of data parallel programs compiled to message passing codes [25] Additionally, the simulator itself is ....

R. Bagrodia, S. Docy and A. Kahn. "Parallel Simulation of Parallel File Systems and I/O Programs," SuperComputing'97, 1997.

No context found.

R. Bagrodia, S. Docy, and A. Kahn. Parallel Simulation of Parallel File Systems and I/O Programs. In Proc. of SuperComputing'97.

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