Abstract
The partitioned global address space (PGAS) programming model has gained attention as a robust model suitable for a diversity of emerging concurrent architectures. PGAS offers more scalability over the former distributed shared memory system (DSM) by supporting asynchronous execution based on message passing. Combining asynchronous communication with the facility to make the location of data transparent, applications written in PGAS languages have to trade off the benefits of concurrent architectures with the overhead caused by accessing distant memories. Here we present an effective simulation model to reflect the cost of distant memory accesses on a PGAS system. The model, called Gasimo, simulates a generic PGAS execution environment on top of a cluster of homogeneous dual-core machines. Gasimo is a parallel extension of a particular DSM simulator, called DSiMCluster, which has been implemented on top of a discrete event simulation (DES) engine known as HASE.
Author supplied keywords
Cite
CITATION STYLE
Marurngsith, W., & Ibbett, R. N. (2010). Gasimo: A global address space simulation model. In SIMUTools 2010 - 3rd International ICST Conference on Simulation Tools and Techniques. ICST. https://doi.org/10.4108/ICST.SIMUTOOLS2010.8671
Register to see more suggestions
Mendeley helps you to discover research relevant for your work.