Download presentation
Presentation is loading. Please wait.
Published byJennifer King Modified over 11 years ago
1
InterCell foundations with ParXXL Render Large Scale Computations Interactive Jens Gustedt INRIA Nancy – Grand Est AlGorille Stéphane Vialle Supélec Metz campus
2
InterCell lowlevel ParXXL 2 Jens Gustedt Jens Gustedt INRIA RILLEpour la g rithmesALGO Stéphane Vialle Goals Provide a programming paradigm and an interface: Simplicity: easy to use close to existing programming habits Performance: competitive responsive easy to evaluate Interoperability: close to known standards, heterogenity of computing material, OS, administrative domains
3
InterCell lowlevel ParXXL 3 Jens Gustedt Jens Gustedt INRIA RILLEpour la g rithmesALGO Stéphane Vialle
4
InterCell lowlevel ParXXL 4 Jens Gustedt Jens Gustedt INRIA RILLEpour la g rithmesALGO Stéphane Vialle Round trip time: > 0.13 s Synchronization Frequency: < 7.5 Hz Problem 1: Communication Latency
5
InterCell lowlevel ParXXL 5 Jens Gustedt Jens Gustedt INRIA RILLEpour la g rithmesALGO Stéphane Vialle Overruling Latency
6
InterCell lowlevel ParXXL 6 Jens Gustedt Jens Gustedt INRIA RILLEpour la g rithmesALGO Stéphane Vialle Problem 2: Data Management Access? Management?
7
InterCell lowlevel ParXXL 7 Jens Gustedt Jens Gustedt INRIA RILLEpour la g rithmesALGO Stéphane Vialle Delegation of responsibilities
8
InterCell lowlevel ParXXL 8 Jens Gustedt Jens Gustedt INRIA RILLEpour la g rithmesALGO Stéphane Vialle Isolation in welldefined parts Control Communication
9
InterCell lowlevel ParXXL 9 Jens Gustedt Jens Gustedt INRIA RILLEpour la g rithmesALGO Stéphane Vialle Two commonly used paradigms Message Passing Standardization (MPI) Efficiency (distributed env.) Simplified Data Control Memory Blowup Extra Copies Shared Memory Standardization (pthread, shm_open) Latency Problems Data Consistency Problems Efficiency (parallel machines) Random Access during Computation
10
InterCell lowlevel ParXXL 10 Jens Gustedt Jens Gustedt INRIA RILLEpour la g rithmesALGO Stéphane Vialle parXXL unifies these paradigms Low Level Computational Efficiency (distributed and / or parallel) Simplified Data Control Memory Efficiency (data size) Transfer Efficiency High Level Abstraction for Cellular Computation Implementation of commonly used Cell-Networks
11
InterCell lowlevel ParXXL 11 Jens Gustedt Jens Gustedt INRIA RILLEpour la g rithmesALGO Stéphane Vialle parXXL achievements 2008 Low Level Packaging and Portability Code Stability: Size Up (fill whole InterCell cluster) Robustness (long runs = transient failures) Transfer Efficiency: Data collectors. Collect distinguished data from the cells Mix communication models (shared memory, files & messages)
12
InterCell lowlevel ParXXL 12 Jens Gustedt Jens Gustedt INRIA RILLEpour la g rithmesALGO Stéphane Vialle ParXXL achievements 2008, cont. High Level Cellular Computation: Find a compromise between synchronized and de-synchronized Cell-Computation Develop new modes of asynchronous computation (cont.) Implementation of commonly used Cell-Networks: Optimize the parallel deployment (avoid scaling problems) (cont.)... changed 40000 lines of code (sep 07 – sep 08)
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.