...
Main Topics | Schedule | Speaker | Affiliation | Type of presentation | Title (tentative) | Download |
|
|
|
|
|
|
|
Dinner Before the Workshop | 7:30 PM | Only people registered for the dinner |
|
|
| |
|
|
|
|
|
|
|
Workshop Day 1 | Wednesday June 12th |
|
|
|
|
|
|
|
|
|
| TITLES ARE TEMPORARY (except if in bold font) |
|
Registration | 08:00 |
|
|
|
|
|
Welcome and Introduction Amphitheatre | 08:30 | Marc Snir + Franck Cappello | INRIA&UIUC&ANL | Background | Welcome, Workshop objectives and organization |
|
| 08:45 | Bill Kramer | UIUC | Background | NCSA updates and vision of the collaboration |
|
| 09:00 | Marc Snir | ANL | Background | ANL updates vision of the collaboration |
|
| 09:15 | Frederic Desprez | Inria | Background | INRIA updates and vision of the collaboration | |
Big systems | 9:30 | Bill Kramer | UIUC | Background | Update on BlueWaters |
|
| 10:00 | Break |
|
|
|
|
| 10:30 | Mitsuhisa Sato | U. Tsukuba & AICS | Background | AICS and the K computer | |
CANCELED | 11:00 | Paul Gibbon | Juelich | Background | Meeting the Exascale Challenge at the Juelich Supercomputing Centre. |
|
Resilience&fault tolerance and simulation | 11:00 | Marc Snir | ANL&UIUC | Report | ICIS report on Resilience |
|
11:30 | Vincent Baudoui | Total & ANL | Joint-Results | Round-off error and silent soft error propagation in exascale applications | ||
| 12:00 | Lunch |
|
|
|
|
Numerical Algorithms | 13:30 | Bill Gropp | UIUC | Background | Topics for Collaboration in Numerical Libraries |
|
14:00 | Paul Hoveland | ANL | Background | Argonne strategic plan in applied math |
| |
| 14:30 | Marc Baboulin | INRIA | Background | Using condition numbers to assess numerical quality in high-performance computing applications |
|
| 15:00 | Luke Olson | UIUC | Background | Opportunities in developing a more robust and scalable multigrid solver |
|
15:30 | Break | |||||
| 16:00 | Frederic Nataf | INRIA&P6 | Background | Toward black-box adaptive domain decomposition methods |
|
Resilience&fault tolerance and simulation Chair: Franck Cappello | 16:30 | Bogdan Nicolae | IBM | Joint Result | AI-Ckpt: Leveraging Memory Access Patterns for Adaptive Asynchronous Incremental Checkpointing | |
17:00 | Martin Quison | INRIA | Result | Improving Simulations of MPI Applications Using A Hybrid Network Model with Topology and Contention Support | ||
| 17:30 | Adjourn |
|
|
|
|
| 18:45 | Bus for Diner |
|
|
|
|
|
|
|
|
|
|
|
Workshop Day 2 | Thursday June 13th |
|
|
|
|
|
|
|
|
|
|
|
|
Programming Models (cont.) | 08:30 | Jean-François Mehaut | INRIA | Result | Progresses in the European FP7 Mont-Blanc 1 project and objectives of its follow up: Mont-Blanc 2 |
|
| 09:00 | Rajeev Thakur | ANL | Background | Update on MPI and OS/R Activities at Argonne |
|
| 09:30 | Andra Ecaterina Hugo | INRIA | Results | Composing multiple StarPU applications over heterogeneous machines: a supervised approach |
|
| 10:00 | Celso Mendes | UIUC | Background | Dynamic Load Balancing for Weather Models via AMPI |
|
| 10:30 | Break |
|
|
|
|
Big Data, I/O, Visualization | 11:00 | Dries Kimpe | ANL | Results | Triton: Exascale Storage |
|
| 11:30 | Gilles Fedak | INRIA | Result | Active Data: A Programming Model to Manage Data Life Cycle Across Heterogeneous Systems and Infrastructures |
|
| 12:00 | Matthieu Dorrier | INRIA | Joint Result | Data Analysis of Ensemble Simulations: an In Situ Approach using Damaris |
|
| 12:30 | Ian Foster | ANL | Background | TBA |
|
| 13:00 | Lunch |
|
|
|
|
|
|
|
|
|
|
|
Mini Workshop1 |
|
|
|
|
|
|
Resilience | 14:00 | Ana Gainaru | UIUC | Results | Challenges in predicting failures on the Blue Waters system. |
|
| 14:30 | Xiang Ni | UIUC | Results | ACR: Automatic Checkpoint/Restart for Soft and Hard Error Protection. |
|
| 15:00 | Tatiana Martsinkevich | INRIA & ANL | Result | On the feasibility of message logging in hybrid hierarchical FT protocols |
|
| 15:30 | Mohamed Slim Bouguerra | INRIA & ANL | Result | Investigating the probability distribution of false negative failure alerts in HPC systems |
|
| 16:00 | Break |
|
|
|
|
| 16:30 | Amina Guermouche | UVSQ | Result | Multi-criteria Checkpointing Strategies: Response-time versus Resource Utilization |
|
| 17:00 | Thomas Ropars | EPFL | Result | Towards efficient replication of HPC applications to deal with crash failures |
|
| 17h30 | Mehdi Diouri | INRIA | Result | ECOFIT: A Framework to Estimate Energy Consumption of Fault Tolerance Protocols for HPC Applications |
|
| 18:00 | Adjourn |
|
|
|
|
|
|
|
|
|
|
|
Mini Workshop2 |
|
|
|
|
|
|
Numerical Algorithms and Libraries | 14:00 | Jean Utke | ANL | Result | Designing and implementing a tool-indedendent, adjoinable MPI wrapper library |
|
| 14:30 | Laurent Hascoet | INRIA | Result | The adjoint of MPI one-sided communications |
|
| 15:00 | Stefan Wild, | ANL | Result | Loud computations? Noise in iterative solvers |
|
| 15:30 | Jed Brown | ANL | Result | Vectorization, communication aggregation, and reuse in stochastic and temporal dimensions |
|
| 16:00 | Break |
|
|
|
|
| 16:30 | Yushan Wang | INRIA P11 | Result | Accelerating incompressible fluid flows simulations using SIMD or GPU computing |
|
| 17:00 | Frederic Hecht | INRIA/P6 | Result | TBA |
|
| 18:00 | Adjourn |
|
|
|
|
|
|
|
|
|
|
|
| 18:45 | Bus for diner |
|
| Lyon |
|
|
|
|
|
|
|
|
Workshop Day 3 | Friday June 14th |
|
|
|
|
|
|
|
|
|
|
|
|
Mini Workshop1 (cont.) |
|
|
|
|
|
|
Resilience | 08:30 | Di Sheng | INRIA | Result | Optimization of Google Cloud Task Processing with Checkpoint-Restart Mechanism |
|
| 09:00 | Guillaume Aupy | INRIA | Result | On the Combination of Silent Error Detection and Checkpointing |
|
| 09:30 | Discussion |
|
|
|
|
| 10:00 | Break |
|
|
|
|
Mini Workshop3 | 10:30 | Guillaume Mercier | INRIA | Result | Topology Management and MPI Implementations Improvements |
|
Programming and Scheduling | 11:00 | Vincent Lanore | INRIA | Result | Static 2D FFT adaptation through a component model based on Charm++ |
|
| 11:30 | Anne Benoit | INRIA | Result | Energy-efficient scheduling |
|
| 12:00 | François Tessier | INRIA | Result | Communication-aware load balancing with TreeMatch in Charm++ |
|
| 12:30 | Discussions |
|
|
|
|
| 13:00 | Closing and Lunch |
|
|
|
|
|
|
|
|
|
|
|
Mini Workshop2 (cont.) |
|
|
|
|
|
|
Numerical Algorithms and Libraries | 08:30 | François Pellegrini | INRIA | Result | Shared memory parallel algorithms in Scotch 6 |
|
| 09:00 | Luc Giraud | INRIA | Result | TBA |
|
| 09:30 | Discussions |
|
|
|
|
| 10:00 | Break |
|
|
|
|
Mini Workshop4 | 10:30 | Kate Keahey | ANL | Result | TBA |
|
Clouds | 11:00 | Gabriel Antoniu | INRIA | Result | TBA |
|
| 11:30 | Christian Perez | INRIA | Result | TBA |
|
| 12:00 | Eddy Caron | INRIA | ResultTBA | Seed4C: Secured Embedded Element and Data privacy for Cloud Federation |
|
| 12:30 | Discussions |
|
|
|
|
| 13:00 | Closing and Lunch |
|
|
|
|
...
Energy-efficient scheduling
Jean Utke
Designing and implementing a tool-indedendent, adjoinable MPI wrapper library
The efficient computation of gradients by the "adjoint-mode" of algorithmic differentiation (AD) entails the inversion of MPI communication graphs. The logic to be implemented for adjoining non-blocking communication patterns is sufficiently complex to warrant a design of components that is independent of the algorithmic differentiation tool that provides the context in which the adjoint communication is to take place. We discuss (i) how we account for the different data models implied by the AD tool as well as the target language, (ii) the implementation choices among the possible adjoint communications, and (iii) the currently known limitations of our approach. We hope for feedback from the community regarding this design particularly with respect to performance and current developments in the MPI standard.
Laurent Hascoet
In this talk, I will survey recent works on energy-efficient scheduling. The goal is to minimize the energy consumption of a schedule, given some performance constraints, for instance a bound on the total execution time. I will first revisit the greedy algorithm for independent tasks in this context. Then I will present problems accounting for the reliability of a schedule: if a failure may occur, then replication or checkpoint is used to achieve a reliable schedule. The goal remains the same, i.e., minimize the energy consumption under performance constraints
Jean Utke
Designing and implementing a tool-indedendent, adjoinable MPI wrapper library
The efficient computation of gradients by the "adjoint-mode" of algorithmic differentiation (AD) entails the inversion of MPI communication graphs. The logic to be implemented for adjoining non-blocking communication patterns is sufficiently complex to warrant a design of components that is independent of the algorithmic differentiation tool that provides the context in which the adjoint communication is to take place. We discuss (i) how we account for the different data models implied by the AD tool as well as the target language, (ii) the implementation choices among the possible adjoint communications, and (iii) the currently known limitations of our approach. We hope for feedback from the community regarding this design particularly with respect to performance and current developments in the MPI standard.
Laurent Hascoet
The adjoint of MPI one-sided communications
Computing gradients of numerical models by the adjoint mode of algorithmic differentiation is a crucial ingredient for model optimization, sensitivity analysis, and uncertainty quantification of many large-scale science and engineering applications. The adjoint mode implies a reversal of the data dependencies and consequently a reversal of communications in parallelized models. Building on previous studies regarding the adjoining of MPI The adjoint of MPI one-sided communications
Computing gradients of numerical models by the adjoint mode of algorithmic differentiation is a crucial ingredient for model optimization, sensitivity analysis, and uncertainty quantification of many large-scale science and engineering applications. The adjoint mode implies a reversal of the data dependencies and consequently a reversal of communications in parallelized models. Building on previous studies regarding the adjoining of MPI two-sided communications, we investigate the construction of adjoints for certain one-sided MPI communications
...
Investigating the probability distribution of false negative failure alerts in HPC systems
As large parallel systems increase in size and complexity, failures are inevitable and exhibit complex space and time dynamics. Several key results have demonstrated that recent advances in event log analysis can provide precise failure prediction. The state-of-the-art in failure prediction provides a ratio of correctly identified failures to the number of all predicted failures of over 90\% and its able to discover around 50\% of all failures in a system. However large part of failures are not predicted and considered as false negative alerts. Therefore, developing efficient fault tolerance strategies to tolerate failures requires a good perception and understanding of failure prediction properties and characteristics. In order to study and understand the properties and characteristics of the false negative alerts, we conduct in this paper a statistical analysis to discover the probability distribution of such alerts and their impact on fault tolerance techniques. To this end we study failures logs from different HPC production systems. We show that: (i) surprisingly the false negative distribution has the same nature as the failure distribution; (ii) after adding failure prediction we were able to infer statistical models that describes the inter arrival time between false negative alerts and so current fault tolerance can be applied on these systems; (iii) the current failures traces contain a high amount of correlation between the failure inter arrival time that can be used to improve the failure prediction mechanism. Another important result is that checkpoint intervals can still be computed from existing first order formula when failure distribution is purely random.
...
In this talk, we revisit traditional checkpointing and rollback recovery strategies, with a focus on silent data corruption errors. Contrarily to fail-stop failures, such latent errors cannot be detected immediately, and a mechanism to detect them must be provided. We consider two models: (i) errors are detected after some delays following a probability distribution (typically, an Exponential distribution); (ii) errors are detected through some verification mechanism. In both cases, we compute the optimal period in order to minimize the waste, i.e., the fraction of time where nodes do not perform useful computations. In practice, only a fixed number of checkpoints can be kept in memory, and the first model may lead to an irrecoverable failure. In this case, we compute the minimum period required for an acceptable risk. For the second model, there is no risk of irrecoverable failure, owing to the verification mechanism, but the corresponding overhead is included in the waste. Finally, both models are instantiated using realistic scenarios and application/architecture parameters.
Dries Kimpe
Triton: Exascale Storage
In this talk, I will present a status update of our work on Triton, a newly designed exascale era storage system. In addition to Triton specific information, the presentation will also include a brief discussion about the tools and techniques that help us in implementing and designing Triton. One such tool is the use of discrete event simulation to quickly evaluate algorithms at scale before implementing them in Triton.
Tatiana Martsinkevich
On the feasibility of message logging in hybrid hierarchical FT protocols
...
Optimization of Google Cloud Task Processing with Checkpoint-Restart Mechanism
...
Modern hardware architectures featuring multicores and a complex memory hierarchy raise challenges that need to be addressed by parallel applications programmers. It is therefore tempting to adapt an application communication pattern to the characteristics of the underlying hardware. The MPI standard features several functions that allow the ranks of MPI processes to be reordered according to a graph attached to a newly created communicator. In this talk, we explain how the MPI implementation of the MPI_Dist_graph_create function was modified to reorder the MPI process ranks to create a match between the application communication pattern and the hardware topology. The experimental results on a multicore cluster show that improvements can be achieved as long as the application communication pattern is expressed by a relevant metric. We also show several areas in MPI implementations where similar techniques can be beneficial.
Eddy Caron
Seed4C: Secured Embedded Element and Data privacy for Cloud Federation
In this talk we introduce the design of a secure federated cloud from end to end. We discussed the core of this platform based on a High Performance Computing middleware that uses federated clouds and other virtual resources as well classic HPC resources. We propose an architecture to ensure a high level of security from personal devices to the targeted virtual machine. The Seed4C platform improved security in each layer. With DIET Cloud, we are able to deploy a large-scale, distributed and secure HPC platform that spans across a large pool of resources aggregated from different providers through a secure way