B. Verghese, S. Devine, A. Gupta and M. Rosenblum, OS support for improving data locality on CC-NUMA compute servers, in Proc. of the 7th Int'l Conference on Architectural Support for Programming Languages and Operating Systems, 1996, pp. 279-289.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....Work Data locality on NUMA machines has been addressed in many projects over the last years. Among the projects focusing on improving data locality, a few approaches based on page migration have been proposed; most of them, however, only target tightly coupled architectures. Verghese et.al. [9] study the improvements of performance on CC NUMA systems, provided by OS supported dynamic migration and replication. This kind of pagemigration is based on the information about full cache misses collected via instrumenting the OS. Hot pages, i.e. pages to which a large number of misses are ....

....execution of applications. Besides that, SIMT comprises functionality enabling a transparent transfer between the simulation platform and a real cluster. Currently, the ARS migration algorithms are implemented on top of SIMT. 3. 2 Page Migration Algorithms Commonly used page migration mechanisms [9, 6] are based on competitive algorithms which migrate a page if the difference between the number of local references and the number of remote references from one node exceeds a predefined threshold. This scheme is easy to implement; a similar one, called Out U, is therefore also applied within ARS. ....

B. Verghese, S. Devine, A. Gupta, and M. Rosenblum. OS Support for Improving Data Locality on CC-NUMA Compute Servers. Technical Report CSL-TR-96-688, Computer System Laboratory, Stanford University, February 1996.

....Work Data locality on NUMA machines has been addressed in many projects over the last years. Among the projects focusing on improving data locality, a few approaches based on page migration have been proposed; most of them, however, only target tightly coupled architectures. Verghese et.al. [9] study the improvements of performance on CC NUMA systems, provided by OS supported dynamic migration and replication. This kind of pagemigration is based on the information about full cache misses collected via instrumenting the OS. Hot pages, i.e. pages to which a large number of misses are ....

....execution of applications. Besides that, SIMT comprises functionality enabling a transparent transfer between the simulation platform and a real cluster. Currently, the ARS migration algorithms are implemented on top of SIMT. 3. 2 Page Migration Algorithms Commonly used page migration mechanisms [9, 6] are based on competitive algorithms which migrate a page if the difference between the number of local references and the number of remote references from one node exceeds a predefined threshold. This scheme is easy to implement; a similar one, called Out U, is therefore also applied within ARS. ....

B. Verghese, S. Devine, A. Gupta, and M. Rosenblum. OS Support for Improving Data Locality on CC-NUMA Compute Servers. Technical Report CSL-TR-96-688, Computer System Laboratory, Stanford University, February 1996.

....several approaches based on page migration have been implemented. This section briefly describes a few of such approaches. In addition, a few approaches for thread migration have to be mentioned, as their techniques can in principle be applied to page migration as well. Verghese et al. [22] study the performance improvement on CC NUMA systems with OS supported dynamic migration and replication. This kind of page migration is based on the information about full cache misses collected via instrumentation in the OS. Hot pages, i.e. pages on which a large number of misses occur, are ....

....modification in the shared virtual memory and delivers the migration information as service replies further to the GUI of ARS enabling the visualization of the migration behavior. Migration algorithms play a central role in developing a migration system. Commonly used page migration mechanisms [17,22] are based on competitive algorithms which migrate a page if the difference between the number of local references and the number of remote references from one node exceeds a predefined threshold. This scheme is easy to implement, hence a similar one, called Out U, is also applied within ARS as a ....

B. Verghese, S. Devine, A. Gupta, M. Rosenblum, OS support for improving data locality on CC-NUMA compute servers, Technical Report CSL-TR-96-688, Computer System Laboratory, Stanford University, February 1996.

....does, but is intended only for the analysis of the interconnect trac with the goal to improve the modeling accuracy for network simulation systems. Data locality, on the other hand, has been addressed intensively since it has a severe in uence on performance of NUMA systems. Among these e orts [11, 4, 15, 2, 19, 16, 22], which are primarily based on compiler analysis and page migration. One of these e orts, especially closely related to the approach presented here, is the Dprof pro ling tool [2] developed by SGI. Dprof samples a program during its execution and records the program s memory access information as ....

B. Verghese, S. Devine, A. Gupta, and M. Rosenblum. OS support for improving data locality on CC-NUMA compute servers. Technical Report CSL-TR-96-688, Computer System Laboratory, Stanford University, Feb. 1996.

....no physical home , and main memory acts as a huge cache. The hardware transparently migrates and replicates the data to the current accessor, adapting dynamically to the memory access pattern and hopefully placing the data closer to the subsequent accessors. As an alternative of COMA, Verghese [11] has augmented NUMA RC machines with additional operating system support for dynamic page migration, aimed at reducing the remote memory access latencies. Although these techniques are aimed at multiprocessors they have much in common with the very large reconfigurable fabrics we are considering. ....

Ben Verghese, Scott Devine, Anoop Gupta, and Mendel Rosenblum. OS support for improving data locality on CCNUMA compute servers. Technical Report CSL-TR-96688, Computer System Laboratory, Stanford University, 1996.

....most important performance issue of NUMA architectures among factors like memory contention and false sharing [13] which affect performance. An improvement of data locality can be achieved through a static data distribution during compiling time [8, 10, 18] and a dynamic redistribution in run time [12, 15]. In the compiler approach, locality is analyzed during the compiling procedure using the information about the context of data structures and functions. Communication among processors is minimized via initially effective data distribution[10] communication overlap [8] and elimination of ....

....data, decides the location of shared pages, and moves incorrectly allocated pages to nodes which require them more frequently. As the most information can be offered by the hardware monitor, many troubles related to extracting symbol information from binaries as it is the case in [12] and [15] are avoided. The tool environment includes another tool, the SIMulation Tool (SIMT) for architecture design and performance evaluation. SIMT is an execution driven multithreads simulator which simulates the parallel execution of SPMD programs on top of virtual shared memory systems with various ....

Ben Verghese, Scott Devine, Anoop Gupta, and Mendel Rosenblum. OS support for improving data locality on CC-NUMA compute servers. Technical Report CSL-TR-96-688, Computer System Laboratory, Stanford University, February 1996.

....of one iteration will be repeated throughout the execution of the program. The other algorithm checks periodly for hot memory areas and migrates the pages with excessive remote references. Performance evaluations on an SGI Origin2000 show a significant improvement in throughput. Verghese etc. [16] study the improvements in performance provided by OS supported dynamic migration and replication. This kind of page migration is based on the information about full cache misses. Results of their experiments show a performance increase of up to 29 for some workloads. The implemented approaches ....

Ben Verghese, Scott Devine, Anoop Gupta, and Mendel Rosenblum. OS support for improving data locality on CC-NUMA compute servers. Technical Report CSL-TR-96-688, Computer System Laboratory, Stanford University, February 1996.

No context found.

B. Verghese, S. Devine, A. Gupta and M. Rosenblum, OS support for improving data locality on CC-NUMA compute servers, in Proc. of the 7th Int'l Conference on Architectural Support for Programming Languages and Operating Systems, 1996, pp. 279-289.

No context found.

B. Verghese, S. Devine, A. Gupta and M. Rosenblum. OS support for improving data locality on CC-NUMA compute servers. In Proceedings of the 7th Int'l ConferenceonArchitectural Support for Programming Languages and Operating Systems, pages 279--289, 1996.

No context found.

B. Verghese, S. Devine, A. Gupta and M. Rosenblum. OS support for improving data locality on CC-NUMA compute servers. In Proc. of ASPLOS, pages 279-289, 1996.

No context found.

B. Verghese, S. Devine, A. Gupta, and M. Rosenblum. Os support for improving data locality on cc-numa compute servers. In In Proceedings of the 7th International Conference on Architectural Support for Programming Languages and Operating Systems, October 1996.

No context found.

B. Verghese, S. Devine, A. Gupta, and M. Rosenblum. Os support for improving data locality on cc-numa compute servers. In In Proceedings of the 7th International Conference on Architectural Support for Programming Languages and Operating Systems, October 1996.

No context found.

B. Verghese, S. Devine, A. Gupta, and M. Rosenblum. OS Support for Improving Data Locality on CC-NUMA Compute Servers. Technical Report CSLTR -96-688, Computer System Laboratory, Stanford University, February 1996.

No context found.

B. Verghese, S. Devine, A. Gupta, and M. Rosenblum. OS Support for Improving Data Locality on CC-NUMA Compute Servers. Technical Report CSL-TR-96688, Computer System Laboratory, Stanford University, February 1996.

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