| 個人檔案Microsoft HPC Institute ...相片部落格清單 | 說明 |
|
5月9日 What is next?The main problem we are facing now is scalibility (although more model functionality is also an interesting possibility). We want to be able to analyze a program for more than a few processes in a given model. On this front I have some ideas about how we will accomplish it. More on this when it becomes available. A Formal Model of MPI Process InteractionNo formal model for MPI process interaction currently exists. A formal model is desirable because it
In our paper (currently under review) we describe a formal model of MPI process interaction, detailing the protocols used in its implementation, and how it was validated using examples from the MPI standard document. We also demonstrate how a programmer might use the model to find bugs in their code, as well as gain a deeper understanding of the MPI standard itself.
The model (in +CAL) can be downloaded from our website. We use the TLC model checker from Microsoft Research for this work. 1月18日 The beginningAs I am fond of saying -"Its time to get this show on the road."
The University of Utah 's Microsoft HPC Institute is focused on developing techniques to ease the process of developing for a cluster computing environment. We are launching a number of research initiatives that include the following:
|
||||
|
|