This page gives the procedure to ask for a special usage permission, e.g. to use more resources than what is allowed in the Grid'5000 usage policy.
Asking for special permission
In order to make a request for a special usage permission, please send an email to the Grid'5000 executive committee. Your request should include:
- your affiliation (laboratory, team, and who is your advisor if you are a student)
- a description of the motivations and the expected scientific impact of your experiments (10-20 lines)
- detailed information about what you are asking for in terms of resources:
- number of nodes
- date/time of the reservation
- duration of the reservation
- explanation about why this work is not doable under the Grid'5000 usage policy (that is, which rules are a problem for this work, and why). (5-10 lines)
Please allow at least two full business days to give us time to evaluate your request. Also, as a general policy, we will not cancel other users' reservations, which is another reason for making such exceptional requests in advance.
Special permissions granted per year
This section lists authorized usages of Grid'5000 by the executive committee that did not fit well with the Usage Policy.
- 1 node on gemini during specific time range from 2021-10-13 to 2021-10-15 for Maxime Gonthier : Scheduling under memory constraints using GPU
- 2021-10-13/14: 11h-19h
- 2021-10-15: 8h-16h
- 4 nodes in Lille (chetemi or chiclet) for Eglantine Karlé for a week of computations, starting from 2021/09/10, valid until 2021/09/19 : Compare different methods for ranking problems for large datasets.
- 1 node of grvingt or dahu for 100 hours from 2021-09-07 to 2021-09-24 for Rania Talbi: Experiments on encrypted data classification for manuscript submission.
- 30 nodes of dahu and 3 nodes of yeti (Grenoble) for Loïc Dachary for 9 days from 2021-08-06: benchmark demonstrating that a novel object storage architecture delivers the expected results in an experimental environment at scale.
- 2 nodes of dahu (Grenoble) for Alain Tchana/Kevin Nguetchouang for one month from 2021-07-07 to 2021-08-06: scalabilité des systèmes de virtualisation pour le stockage (QEMU/KVM)
- 24 nodes of parasilo (Rennes) for Isabelly Louredo Rocha for 8 days from 2021-05-14 to 2021-05-21: characterizing Machine Learning (ML) and Deep Neural Network (DNN) Workloads on CPU machines using Spark framework
- 4 nodes of dahu (Grenoble) for Alain Tchana for 11 days from 2021-04-27: scalabilité des systèmes de virtualisation pour le stockage (QEMU/KVM)
- 25 nodes of paravance (Rennes) for Yasmine Djebrouni for 5 days from 2021-04-16: Characterization of deep learning and Machine Learning workloads on Spark Apache
- 5 nodes on chifflot (Lille) for Brij Srivastava for 14 days from 2021-04-01: reconnaissance de locuteurs, anonymisation de la voix
- 1 node on chifflet (Lille) for Stéphane Rouget for 48h from 2021-04-01: measure the performance of TTile, a comprehensive methodology for finding optimized tile configurations with imperfectly nested microkernels ("beyond perfect") and outer tile loops optimized via analytical modeling.
- 1 node on Gemini (Lyon) for Ayman MAKKI with unrestricted job prolongation using oarwalltime from 8th to 21th Februray: An automatic and efficient method for Neural Architecture Search using Bayesian Optimisation
- 1 node on Gemini (Lyon) for Thierry Gautier from 9:00 12th oct for 2 days: MUMPTech project
- 1 node on Gemini (Lyon) for Thierry Gautier from 9:00 28th sept for 5 days: MUMPTech project
- 1 node on gros (Nancy) for Dimitri Saingre from 15:00 Sep 7th to Sep 30th : Crypto-currency smart contracts study.
- 10 nodes on Parasilo (Rennes) for Isabelly Rocha from 09:00 23th june to 09:10 3rd july : Deep Learning using Spark framework.
- Jobs of 19h with 3 nodes on chifflot (Lille) for Adnan Zeddoun from 18th june to 19th july : Deep Learning on backgammon
- chifflot-7 and chifflot-8 (Lille) continuously for Manuel Diaz zapata from 13:00 11th june to 01:00 13th june : training for two neural networks
- 5 nodes on Parasilo (Rennes) for Yasmine Djebrouni from 09:00 27th april to 13:00 1st May : Collect and analyze experiments traces on Distributed Machine Learning.
- 12 nodes on Paravance (Rennes) for Ronan Alexandre Cherrueau from 18:06 4th february to 12:52 5th february : Practical session for OpenStack course (network issue in Nantes the 4th february - routed network Kavlans unusable).
- 15 nodes on Ecotype (Nantes) for Adrien Lebre from 12:30 3rd february to 12:30 5th february : Practical session for OpenStack course.
- 2 taurus nodes on Lyon for Zeina Houmani from Wednesday September 18 at 8AM To Thursday September 19 at 1AM (17 hours) : work on data-driven microservice architecture.
- 1 node on Lille (chifflot cluster) from 15th August 2019 to 16th August 2019 for Fabien Teytaud : AI algorithms international competition in Macao.
- 48 nodes on Nantes (ecotype cluster) from 17th April 2019 to 19th April 2019 for Jonathan Pastor : Determining a model of the electrical consumption of servers and coolers in the Ecotype server room.
- 1 node on Lille (chifflet cluster) from 26th March 2019 to 10th April 2019 for Brij Mohan Lal Srivastava : Training of the state-of-the-art i-vector systems in the context of the Comprise project.
- 64 nodes on Nancy (grvingt cluster) from 22th March 2019 to 01st April 2019 for Tiago Carneiro Pessoa : Evaluation of productivity-aware languages for HPC in the context of distributed branch and bound.
- 14 nodes on Nantes from 25th February 2019 to 27th February 2019 for Ronan-Alexandre Cherrueau : OpenStack lab IMT Nantes.
- 4 nodes on Lyon (nova cluster) from 4th February 2019 to 8th February 2019 for Tarek Menouer : Scheduling context in Docker Swarm toolkit.
- 10 nodes on Lyon (nova cluster) from 24th December 2018 to 31th December 2018 for Yasmina Bouizem : Evaluation of FaaS frameworks and trying to measure the energy consumption for each framework during the execution of the workload.
- Permission to bypass the limit of 2 advance reservations in the site of Lyon in order to use all the nodes of the site for all the nights for one week in December 2018 for Maha Alsayasneh (Erods, LIG): Study of performance bottlenecks of different distributed system stacks.
- 8 nodes on Rennes (paravance cluster) from 10th December 2018 to 23th December 2018 for Vania Marangozova-Martin (Erods, LIG) : Elasticity of multiple execution containers.
- 8 nodes on Rennes (paravance cluster) from 03th November 2018 to 18th November 2018 for Vania Marangozova-Martin (Erods, LIG) : Elasticity of multiple execution containers.
- 48 nodes on Nantes (Ecotype cluster) from 14th May to 18th May 2018 for Marie Delavergne, Ronan Alexandre Cherrueau and Adrien Lebre : Evaluation of NewSQL databases.
- Permission to bypass the day/night boundary (i.e. have jobs running from 9am to 8pm) until 2018-05-31, for Amir Teshome Wonjiga: experiments on cloud security monitoring that require few nodes but a long time.
- 18 nodes on Nantes (Econome cluster) from 27th February to 2nd March 2018 for Linh Nguyen : Experiment on optimization of VM boot time.
- 12 nodes on Lyon (taurus cluster), for 7 days, January 8th to January 15th, for Mathilde Boutigny, Issam Raïs and Laurent Lefèvre (LIP, Lyon): experiment on energy savings
- 7 nodes in Rennes for 2 days (parasilo/paravance clusters), Monday January 22th 9:00 to Wednesday January 24th 19:00, for Clément Elbaz (Christine Morin, Myriads team, INRIA Rennes), presentation at FIC 2018 in Lille (https://www.forum-fic.com/)
- 7 nodes in Rennes for 2 days, October 16 and 17, for Clément Elbaz (Christine Morin, Myriads team, INRIA Rennes), presentation at "rencontres Inria-Industrie" (Paris)
- 20 paravence nodes in Rennes for 5 days in a row: from Friday May 5th 9am to Tuesday May 10th 9am, for Vikas Jaiman (Erods team, LIG)
- 7 nodes in Rennes for 2 days during the whole days starting at 8 AM for a presentation at FIC 2017 for Anna Giannakou
- 4 taurus machines in Lyon for 3 days in a row, once, for Hélène Coullon, Issam Raïs, Laurent Lefèvre, Christian Perez, to be completed by October 9th, 2016
- 22 Suno nodes, once a week during the complete day, between July 14th and August 31st, 2016, for Hardik Soni
- long jobs on the production resources in Nancy to Bernard Maigret, on the graoully cluster between July 21st and July 31st 2016
- 15 nodes in Lyon (job 801706), 2 nodes in Nancy (job 807704), 15 nodes in Rennes (job 740896), 15 nodes in Reims (job 188450), for Adel Nadjaran Toosi, between February 5, 2016 and February 8, 2016 (80h), for an experiment on geographical load balancing requiring 3 consecutive days.
- 20 nodes on stremi @ Reims, between 15:00 and 23:59 CET on November 17th, 18th and 19th, 2015 (during SC'15), for Daniel Balouek-Thomert, to demonstrate a collaboration between Avalon and the NewGeneration-SR startup around plugins for green scheduling in datacenters.
- 4 nodes on econome and 8 nodes on paranoia between Friday, Oct. 23rd, 19:00 CEST to Friday Oct. 30th, 19:00 CET. This is for the Harness project's final review. Reservation by George Ioannidis
- 15 nodes for paravance for the whole week between June 1st and June 6th, 2015. This is for Daniele Calandriello, from the Sequel team at Inria
- Daily reservation of a Taurus node for the Seed4C project from November 17th, 2014 to February 2015.
- Reservation of 2 nodes for a complete week in Nantes, from November 10th to November 17th 2014 (Jonathan Pastor)
- Yulin Zhang ( 10 nodes of Lyon (3 hercule, 3 orion and 4 taurus) for about 8 hours on October 29th, 2014
- Reservation of all the platform by Simon Delamarre for the Spring School Tutorials and challenges: June 17th 14:00 -> 18:00 and June 18th 14:00 -> 18:00.
- Reservation of lyon, rennes, luxembourg and nancy for the COMPAS Grid'5000 tutorial on April 22th between 09:00 and 12:30
- Reservation of ~1500 cores on luxembourg, lyon and reims for the IC2E Grid'5000 tutorial on March 10th between 14:00 and 22:00
- Long usage of one node by Georges Markomanolis on the Stremi cluster in Reims on February 20th, 2014. The job is about simulating with SimGrid 130GB of Time-Independent traces of LU benchmark, class D, 4096 MPI processes.
- Use of 200 machines on Nancy to prepare an entry to the Scale challenge, on February 7th, 2014, between 9:00 and 19:00. Afin de finaliser notre soumission au challenge SCALE de CCGrid, for user tbuchert.
- Use of Grid'5000 for a few day for the mid-term review of European project Seed4C, from October 14th to October 17th. Involved people are Yulin Zhang (hzhang), Lamiel Toch (latoch) and Guillaume Verger (gverger)
- Use of Grid'5000 for the Scale challenge. 14-16 mai, for
- D. Balouek, A. Lèbre, F. Quesnel and
- Eugen Feller, Christine Morin, Matthieu Simonin, Anne-Cécile Orgerie, and Yvon Jégou (Won 2nd prize!)
- Use of Grid'5000 for tutorials at SC'Camp, between July 10-16, 2011, for 30 to 40 students
- Use of clusters parapide (Rennes) and griffon (Nancy) from 9:00 to 18:00 for a demonstration of Computer-Go on April 13-14th 2011.
- Use of Grid'5000 for the Tapei GO competition, Taiwan on March 9th, 2011, from 3 to 9 am.
- Review of the XtreemOS european project, December 2nd, 2010.
- Super Computing and Distributed Systems Camp, Catay, Santander, Colombia, August 15-22, 2010 : created between 20 and 30 accounts for an OAR tutorial
- Challenge Humans vs Computer Go competition, in Rennes and Nancy, July 20th, 2010.
- Challenge trecvid, pour les utilisateurs du groupe irim. Pas d'exceptions à la charte pour la durée et le nombre des jobs.
- Challenge de programmation lors de PASCO 2010, du 21 au 23 juillet, de 7h à 22h, à Grenoble.
- Démo OGF-29 : tout Grid'5000, le 21 juin de 18h15 à 19h30 heure de Chicago (le 22 juin de 1h15 à 2h30 heure de Paris)
- Démo au colloque ANR STIC à la cité des sciences (4-7 janvier 2010) : 8 noeuds à Lyon pour 84h.
- SC'09 (sans discussions préalables)
- 20 machines pour chaque site pendant trois jours de demo (72h) a partir de Mardi 8h (GMT-8). La demonstration fait partie du projet HIPCAL. Son but est de montrer l'execution des applications medicales sur une infrastruture virtuelle grande echelle dont l'infrastructure physique est Grid5000.
- Demos orientees green, 50 noeuds du cluster de Sagittaire de Lyon (sagittaire numeros 30 a 79) du vendredi 13/11 a 9h du matin au vendredi 20/11 a 7h du matin
- Démonstration dans le cadre du Meeting du Consortium du Projet Européen AutoI (Autonomic Internet) à WaterFord (Irlande), 6-9 juillet 2009, https://www.grid5000.fr/mediawiki/index.php/Special:G5KReports?menu=renderexperiment&id=585
- Tutorial lors de la conférence "Latin American Conference on High Performance Computing". http://eventos.saber.ula.ve/eventos/internalPage.py?pageId=4&confId=55, 21 et 22 Septembre 2009
- Ecole d'été INRIA-CEA : 8-19 juin 2009 : bug 2145 : http://www.inria.fr/actualites/colloques/cea-edf-inria/2009/grid/index.en.html : 20aine de participants
- démonstrations à SigMetrics (Seattle): le 17 juin 2009 : bug 2133 : Best student demo award
- démonstrations aux OpenDays aux laboratoires Bell Labs Alcatel Lucent France : 2-5 juin 2009 : bug 2133 :
- GridComp Review : 29-30 janvier 2009 : bug 1774 : http://gridcomp.ercim.org/ :
- Plugtest 2008 : 20-24 octobre 2008 : pas de bug ? : http://www-sop.inria.fr/oasis/plugtests2008/MonteCarloPricingContest.html
- GridComp Review : 16-17 juillet 2008 : bug 1407 : http://gridcomp.ercim.org/ :
- 4th Plugtest 2007 : 28 octobre-1er Novembre 2007
- 3rd Plugtest 2006 : 28-30 novembre 2006 : bug 329 :
- 2nd Plugtest 2005 : 10-11-12 octobre 2005 : pas de bug : http://www.etsi.org/plugtests/grid.htm