From Grid5000
Jump to navigation Jump to search

Grid'5000 is a large-scale and flexible testbed for experiment-driven research in all areas of computer science, with a focus on parallel and distributed computing including Cloud, HPC and Big Data and AI.

Key features:

  • provides access to a large amount of resources: 15000 cores, 800 compute-nodes grouped in homogeneous clusters, and featuring various technologies: PMEM, GPU, SSD, NVMe, 10G and 25G Ethernet, Infiniband, Omni-Path
  • highly reconfigurable and controllable: researchers can experiment with a fully customized software stack thanks to bare-metal deployment features, and can isolate their experiment at the networking layer
  • advanced monitoring and measurement features for traces collection of networking and power consumption, providing a deep understanding of experiments
  • designed to support Open Science and reproducible research, with full traceability of infrastructure and software changes on the testbed
  • a vibrant community of 500+ users supported by a solid technical team

Read more about our teams, our publications, and the usage policy of the testbed. Then get an account, and learn how to use the testbed with our Getting Started tutorial and the rest of our Users portal.

Grid'5000 is merging with FIT to build the SILECS Infrastructure for Large-scale Experimental Computer Science. Read an Introduction to SILECS (April 2018)

Recently published documents and presentations:

Older documents:

Grid'5000 is supported by a scientific interest group (GIS) hosted by Inria and including CNRS, RENATER and several Universities as well as other organizations. Inria has been supporting Grid'5000 through ADT ALADDIN-G5K (2007-2013), ADT LAPLACE (2014-2016), and IPL HEMERA (2010-2014).

Current status (at 2024-05-20 22:52): 1 current events, 5 planned (details)

Random pick of publications

Five random publications that benefited from Grid'5000 (at least 2517 overall):

  • Vladimir Ostapenco, Laurent Lefèvre, Anne-Cécile Orgerie, Benjamin Fichel. Modeling, evaluating, and orchestrating heterogeneous environmental leverages for large-scale data center management. International Journal of High Performance Computing Applications, 2023, 37 (3-4), 10.1177/10943420231172978. hal-04047008 view on HAL pdf
  • Guillaume Le Berre, Christophe Cerisara, Philippe Langlais, Guy Lapalme. Unsupervised multiple-choice question generation for out-of-domain Q&A fine-tuning. 60th Annual Meeting of the Association for Computational Linguistics, May 2022, Dublin, Ireland. pp.732-738, 10.18653/v1/2022.acl-short.83. hal-03797343 view on HAL pdf
  • Maël Madon, Georges da Costa, Jean-Marc Pierson. Characterization of Different User Behaviors for Demand Response in Data Centers. 28th International Conference on Parallel and Distributed Computing : Parallel Processing (Euro-Par 2022), Aug 2022, Glasgow, United Kingdom. pp.53-68, 10.1007/978-3-031-12597-3_4. hal-03768237 view on HAL pdf
  • Ilya Meignan--Masson. Bridging the gap between profiling and monitoring in HPC systems with dynamically reconfigurable fine-grain data collection. 0 Université Grenoble Alpes. 2022. hal-03773464 view on HAL pdf
  • Shakeel Sheikh, Md Sahidullah, Fabrice Hirsch, Slim Ouni. Robust Stuttering Detection via Multi-task and Adversarial Learning. EUSIPCO 2022 - 30th European Signal Processing Conference, Aug 2022, Belgrade, Serbia. hal-03629785 view on HAL pdf

Latest news

Rss.svgCluster "estats" is now in the default queue in Toulouse

We are pleased to announce that the estats cluster of Toulouse (the name refers to Pica d'Estats) is now available in the default queue.

As a reminder, estats is composed of 12 edge-class nodes powered by Nvidia AGX Xavier SoCs. Each node features:

  • 1 ARM64 CPU (Nvidia Carmel micro-arch) with 8 cores
  • 1 Nvidia GPU (Nvidia Volta micro-arch)
  • 32 GB RAM shared between CPU and GPU
  • 1 NVMe of 2TB
  • 1 Gbps NIC
  • Since it is not a cluster of server-class machines (unlike all current other Grid'5000 nodes), estats runs a different default system environment, but other common functionalities are the same (kadeploy etc., except kavlan which is not supported yet).

    For the experimentations, it is recommended to deploy Ubuntu L4T.

    More information in the Jetson page.

    The cluster was funded by a CNRS grant.

    -- Grid'5000 Team 9:51, March 6th 2024 (CEST)

    Rss.svgThe big variant of Debian 12 "Bookworm" environments is ready for deployments

    We are pleased to inform you that the big variant of Debian 12 (Bookworm) environments is now supported for deployments in Grid'5000. Check `kaenv3 -l debian12%` for detailed information.

    Notably, the NVIDIA driver has been updated to version 535.129.03, and CUDA has been upgraded to version 12.2.2_535.104.05_linux for the amd64 architecture.

    The default environment available on nodes will continue to be debian11-std for the foreseeable future.

    Please refer to the updated wiki documentation¹ for guidance on Debian 12-min|nfs|big usage.


    -- Grid'5000 Team 14:21, Jan 22nd 2024 (CEST)

    Rss.svgCluster "montcalm" is now in the default queue in Toulouse

    We have the pleasure to announce that the "montaclm" cluster is now available in the default queue of the Toulouse site, which makes the site full-fledged again!

    This cluster consists of 10 HPE Proliant DL360 Gen10+ nodes with 2 CPUs Intel Xeon Silver 4314 (16 cores per CPUs), 256 GB of DDR4 RAM, and 894GB SSD.

    Jobs submitted on the Toulouse site will run by default on this cluster.

    Beside the "montcalm" cluster, the "edge-class" cluster "estats" is still available in the testing queue for now.

    In order to support the SLICES-FR project, the site infrastructure has been funded by CNRS/INS2I and the "montcalm" cluster has been funded by University Paul Sabatier (UT3).

    -- Grid'5000 Team 10:30, 18 Jan 2024 (CET)

    Rss.svgNew "edge-computing"-class nodes in Toulouse's testing queue: cluster Estats with 12 Nvidia AGX Xavier SoCs

    A new cluster named "estats" is available in the testing queue of the Toulouse site, composed of 12 "Edge computing"-class nodes.

    Estats is composed of 12 Nvidia AGX Xavier SoCs¹. Each SoC features:

  • a ARM64 CPU (Nvidia Carmel micro-arch) with 8 cores
  • a Nvidia GPU (Nvidia Volta micro-arch)
  • 32 GB RAM shared between CPU and GPU
  • a 2TB NVMe
  • 1 Gbps NIC
  • The 12 modules are packaged in a chassis manufactured by Connecttech⁴.

    Since it is not a cluster of server-class machines (unlike all current other Grid'5000 nodes), estats runs a different default system environment. This environment includes Nvidia's Linux for Tegra²³ overlay on top of the Grid'5000 standard environment. This means:

  • a Debian 11 system like other clusters, but
  • a special Linux kernel,
  • several specific tools and services,
  • and several incompatible tools (e.g. Cuda).
  • This default environment does not include the required Tegra-specific version of Cuda.

    To benefit from the whole Nvidia stack with e.g. the specific Cuda version and DL accelerators support for Nvidia Tegra, it is advised to deploy on the node the Nvidia-supported Ubuntu 20.04 OS with the full L4T support, using kadeploy. You can use the ubuntul4t200435-big environment. E.g.:

    ftoulouse$ oarsub -q testing -t exotic -p estats -t deploy -l nodes=1 -I

    ftoulouse$ kadeploy3 ubuntul4t200435-big

    This tutorial page explains how this ubuntul4t200435-big environment is built and how to...

    Read more news

    Grid'5000 sites

    Current funding

    As from June 2008, Inria is the main contributor to Grid'5000 funding.


    Logo INRIA.gif




    IMT Atlantique
    Université Grenoble Alpes, Grenoble INP
    Université Rennes 1, Rennes
    Institut National Polytechnique de Toulouse / INSA / FERIA / Université Paul Sabatier, Toulouse
    Université Bordeaux 1, Bordeaux
    Université Lille 1, Lille
    École Normale Supérieure, Lyon

    Regional councils

    Provence Alpes Côte d'Azur
    Hauts de France