Pp. 153–167 of the Proceedings
Next: Introduction //TRACE: Parallel Trace Replay with Approximate Causal Events Michael P. Mesnier1, Matthew Wachs, Raja R. Sambasivan, Julio Lopez, James Hendricks, Gregory R. Ganger, David O'Hallaron Carnegie Mellon University Introduction Background & motivation Trace replay models Synchronization and the effect on I/O I/O throttling Design overview Discovering data dependencies Discovering compute time Approach 1 Approach 2 Putting it all together Detailed design The causality engine Throttled mode Unthrottled mode Trace replay Preparing traces for replay Replaying the traces Evaluation Experimental setup Experiment 1 (think-limited) Experiment 2 (I/O throttling) Experiment 3 (I/O sampling) I/O sampling discussion Experiment 4 (Node sampling) Related work Conclusion Acknowledgements Bibliography About this document ... Michael Mesnier 2006-12-22
Michael P. Mesnier1, Matthew Wachs, Raja R. Sambasivan, Julio Lopez, James Hendricks, Gregory R. Ganger, David O'Hallaron
Carnegie Mellon University
