Hi NEST-team,
I would like to run a simulation of > 1000 neurons, and to make my computing time faster I am looking into how to use parallel computing with NEST.
I have seen this https://www.nest-simulator.org/parallel-computing/ page but I can't seem to find anything about using the cloud (e.g. google cloud, colab etc.) and the use of GPU's. I am quite new to this so it could be I missed something.
- Is it possible to use a GPU for the nest simulations? - Are the some resources I could learn from?
Thank you!
Best, Daphne
Dear Daphne,
TL;DR: NEST does not support GPUs and can easily run simulations of up to 50k neurons with realistic numbers of synapses on a single computer with only CPUs.
It is a common misconception that GPUs will make everything faster. In fact it might make things much slower for a large class of problems.
For the simulation of strongly coupled spiking neuronal networks using point neuron model, the time to exchange events usually exceeds the time to update the individual neurons by far. As GPUs are foremost good in situations where you can run many operations on the same data and don't read/write from/to memory a lot, they usually don't help here.
If you are interested in running NEST on a cloud service, you might want to look at the Docker image for NEST: https://github.com/nest/nest-docker/
I suggest to start off by writing and running your simulation on your local computer using threads and possibly MPI as described in https://www.nest-simulator.org/parallel-computing/. https://nest-simulator.readthedocs.io/en/latest/examples/ has heaps of examples for all kinds of simulations. If this approach gets too slow, you can move to computer clusters, supercomputers and cloud services.
Cheers, Jochen!
On 02.02.20 18:45, Daphne Cornelisse wrote:
Hi NEST-team,
I would like to run a simulation of > 1000 neurons, and to make my computing time faster I am looking into how to use parallel computing with NEST.
I have seen this https://www.nest-simulator.org/parallel-computing/ page but I can't seem to find anything about using the cloud (e.g. google cloud, colab etc.) and the use of GPU's. I am quite new to this so it could be I missed something.
- Is it possible to use a GPU for the nest simulations?
- Are the some resources I could learn from?
Thank you!
Best, Daphne
NEST Users mailing list -- users@nest-simulator.org To unsubscribe send an email to users-leave@nest-simulator.org
-- Dr. Jochen Martin Eppler Phone: +49(2461)61-96653 ---------------------------------- Simulation Laboratory Neuroscience Jülich Supercomputing Centre Institute for Advanced Simulation
------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------ Forschungszentrum Juelich GmbH 52425 Juelich Sitz der Gesellschaft: Juelich Eingetragen im Handelsregister des Amtsgerichts Dueren Nr. HR B 3498 Vorsitzender des Aufsichtsrats: MinDir Volker Rieke Geschaeftsfuehrung: Prof. Dr.-Ing. Wolfgang Marquardt (Vorsitzender), Karsten Beneke (stellv. Vorsitzender), Prof. Dr.-Ing. Harald Bolt, Prof. Dr. Sebastian M. Schmidt ------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------
I’ve been working on OpenMP based offloading support for some time but given virtual functions only supported from 5.1 specification, it’s not so straight forward to enable devices.
On Mon, Feb 3, 2020 at 18:55 Jochen Martin Eppler j.eppler@fz-juelich.de wrote:
Dear Daphne,
TL;DR: NEST does not support GPUs and can easily run simulations of up to 50k neurons with realistic numbers of synapses on a single computer with only CPUs.
It is a common misconception that GPUs will make everything faster. In fact it might make things much slower for a large class of problems.
For the simulation of strongly coupled spiking neuronal networks using point neuron model, the time to exchange events usually exceeds the time to update the individual neurons by far. As GPUs are foremost good in situations where you can run many operations on the same data and don't read/write from/to memory a lot, they usually don't help here.
If you are interested in running NEST on a cloud service, you might want to look at the Docker image for NEST: https://github.com/nest/nest-docker/
I suggest to start off by writing and running your simulation on your local computer using threads and possibly MPI as described in https://www.nest-simulator.org/parallel-computing/. https://nest-simulator.readthedocs.io/en/latest/examples/ has heaps of examples for all kinds of simulations. If this approach gets too slow, you can move to computer clusters, supercomputers and cloud services.
Cheers, Jochen!
On 02.02.20 18:45, Daphne Cornelisse wrote:
Hi NEST-team,
I would like to run a simulation of > 1000 neurons, and to make my computing time faster I am looking into how to use parallel computing with NEST.
I have seen this https://www.nest-simulator.org/parallel-computing/ page but I can't seem to find anything about using the cloud (e.g. google cloud, colab etc.) and the use of GPU's. I am quite new to this so it could be I missed something.
- Is it possible to use a GPU for the nest simulations?
- Are the some resources I could learn from?
Thank you!
Best, Daphne
NEST Users mailing list -- users@nest-simulator.org To unsubscribe send an email to users-leave@nest-simulator.org
-- Dr. Jochen Martin Eppler Phone: +49(2461)61-96653
Simulation Laboratory Neuroscience Jülich Supercomputing Centre Institute for Advanced Simulation
Forschungszentrum Juelich GmbH 52425 Juelich Sitz der Gesellschaft: Juelich Eingetragen im Handelsregister des Amtsgerichts Dueren Nr. HR B 3498 Vorsitzender des Aufsichtsrats: MinDir Volker Rieke Geschaeftsfuehrung: Prof. Dr.-Ing. Wolfgang Marquardt (Vorsitzender), Karsten Beneke (stellv. Vorsitzender), Prof. Dr.-Ing. Harald Bolt, Prof. Dr. Sebastian M. Schmidt
NEST Users mailing list -- users@nest-simulator.org To unsubscribe send an email to users-leave@nest-simulator.org
Dear Daphne,
If you need access to larger computing resources for NEST (or other neuroscience) simulations, Human Brain Project provides access to some of Europe's largest supercomputers through the ICEI/FENIX program. It requires a brief application and a usually quick review, but the process is much more lightweight than standard applications for supercomputing resources. For more information, see
https://www.humanbrainproject.eu/en/massive-computing/fenix-icei/
Please get in touch if you would like to know more about this option!
Best, Hans Ekkehard
On 3 Feb 2020, at 10:47, Jochen Martin Eppler <j.eppler@fz-juelich.demailto:j.eppler@fz-juelich.de> wrote:
Dear Daphne,
TL;DR: NEST does not support GPUs and can easily run simulations of up to 50k neurons with realistic numbers of synapses on a single computer with only CPUs.
It is a common misconception that GPUs will make everything faster. In fact it might make things much slower for a large class of problems.
For the simulation of strongly coupled spiking neuronal networks using point neuron model, the time to exchange events usually exceeds the time to update the individual neurons by far. As GPUs are foremost good in situations where you can run many operations on the same data and don't read/write from/to memory a lot, they usually don't help here.
If you are interested in running NEST on a cloud service, you might want to look at the Docker image for NEST: https://github.com/nest/nest-docker/
I suggest to start off by writing and running your simulation on your local computer using threads and possibly MPI as described in https://www.nest-simulator.org/parallel-computing/. https://nest-simulator.readthedocs.io/en/latest/examples/ has heaps of examples for all kinds of simulations. If this approach gets too slow, you can move to computer clusters, supercomputers and cloud services.
Cheers, Jochen!
On 02.02.20 18:45, Daphne Cornelisse wrote: Hi NEST-team,
I would like to run a simulation of > 1000 neurons, and to make my computing time faster I am looking into how to use parallel computing with NEST.
I have seen this https://www.nest-simulator.org/parallel-computing/ page but I can't seem to find anything about using the cloud (e.g. google cloud, colab etc.) and the use of GPU's. I am quite new to this so it could be I missed something.
- Is it possible to use a GPU for the nest simulations? - Are the some resources I could learn from?
Thank you!
Best, Daphne
_______________________________________________ NEST Users mailing list -- users@nest-simulator.orgmailto:users@nest-simulator.org To unsubscribe send an email to users-leave@nest-simulator.orgmailto:users-leave@nest-simulator.org
-- Dr. Jochen Martin Eppler Phone: +49(2461)61-96653 ---------------------------------- Simulation Laboratory Neuroscience Jülich Supercomputing Centre Institute for Advanced Simulation
------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------ Forschungszentrum Juelich GmbH 52425 Juelich Sitz der Gesellschaft: Juelich Eingetragen im Handelsregister des Amtsgerichts Dueren Nr. HR B 3498 Vorsitzender des Aufsichtsrats: MinDir Volker Rieke Geschaeftsfuehrung: Prof. Dr.-Ing. Wolfgang Marquardt (Vorsitzender), Karsten Beneke (stellv. Vorsitzender), Prof. Dr.-Ing. Harald Bolt, Prof. Dr. Sebastian M. Schmidt ------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------ _______________________________________________ NEST Users mailing list -- users@nest-simulator.orgmailto:users@nest-simulator.org To unsubscribe send an email to users-leave@nest-simulator.orgmailto:users-leave@nest-simulator.org
--
Prof. Dr. Hans Ekkehard Plesser Head, Data Science Section
Faculty of Science and Technology Norwegian University of Life Sciences PO Box 5003, 1432 Aas, Norway
Phone +47 6723 1560 Email hans.ekkehard.plesser@nmbu.no Home http://arken.nmbu.no/~plesser
Hans,
Is this program widely available to the outside of EU citizens?
Itaru.
On Tue, Feb 4, 2020 at 2:44 PM Hans Ekkehard Plesser < hans.ekkehard.plesser@nmbu.no> wrote:
Dear Daphne,
If you need access to larger computing resources for NEST (or other neuroscience) simulations, Human Brain Project provides access to some of Europe's largest supercomputers through the ICEI/FENIX program. It requires a brief application and a usually quick review, but the process is much more lightweight than standard applications for supercomputing resources. For more information, see
https://www.humanbrainproject.eu/en/massive-computing/fenix-icei/
Please get in touch if you would like to know more about this option!
Best, Hans Ekkehard
On 3 Feb 2020, at 10:47, Jochen Martin Eppler j.eppler@fz-juelich.de wrote:
Dear Daphne,
TL;DR: NEST does not support GPUs and can easily run simulations of up to 50k neurons with realistic numbers of synapses on a single computer with only CPUs.
It is a common misconception that GPUs will make everything faster. In fact it might make things much slower for a large class of problems.
For the simulation of strongly coupled spiking neuronal networks using point neuron model, the time to exchange events usually exceeds the time to update the individual neurons by far. As GPUs are foremost good in situations where you can run many operations on the same data and don't read/write from/to memory a lot, they usually don't help here.
If you are interested in running NEST on a cloud service, you might want to look at the Docker image for NEST: https://github.com/nest/nest-docker/
I suggest to start off by writing and running your simulation on your local computer using threads and possibly MPI as described in https://www.nest-simulator.org/parallel-computing/. https://nest-simulator.readthedocs.io/en/latest/examples/ has heaps of examples for all kinds of simulations. If this approach gets too slow, you can move to computer clusters, supercomputers and cloud services.
Cheers, Jochen!
On 02.02.20 18:45, Daphne Cornelisse wrote:
Hi NEST-team,
I would like to run a simulation of > 1000 neurons, and to make my computing time faster I am looking into how to use parallel computing with NEST.
I have seen this https://www.nest-simulator.org/parallel-computing/ page but I can't seem to find anything about using the cloud (e.g. google cloud, colab etc.) and the use of GPU's. I am quite new to this so it could be I missed something.
- Is it possible to use a GPU for the nest simulations?
- Are the some resources I could learn from?
Thank you!
Best, Daphne
NEST Users mailing list -- users@nest-simulator.org To unsubscribe send an email to users-leave@nest-simulator.org
-- Dr. Jochen Martin Eppler Phone: +49(2461)61-96653
Simulation Laboratory Neuroscience Jülich Supercomputing Centre Institute for Advanced Simulation
Forschungszentrum Juelich GmbH 52425 Juelich Sitz der Gesellschaft: Juelich Eingetragen im Handelsregister des Amtsgerichts Dueren Nr. HR B 3498 Vorsitzender des Aufsichtsrats: MinDir Volker Rieke Geschaeftsfuehrung: Prof. Dr.-Ing. Wolfgang Marquardt (Vorsitzender), Karsten Beneke (stellv. Vorsitzender), Prof. Dr.-Ing. Harald Bolt, Prof. Dr. Sebastian M. Schmidt
NEST Users mailing list -- users@nest-simulator.org To unsubscribe send an email to users-leave@nest-simulator.org
--
Prof. Dr. Hans Ekkehard Plesser Head, Data Science Section
Faculty of Science and Technology Norwegian University of Life Sciences PO Box 5003, 1432 Aas, Norway
Phone +47 6723 1560 Email hans.ekkehard.plesser@nmbu.no Home http://arken.nmbu.no/~plesser
NEST Users mailing list -- users@nest-simulator.org To unsubscribe send an email to users-leave@nest-simulator.org
-
Daphne Cornelisse -
Hans Ekkehard Plesser -
Itaru Kitayama -
Jochen Martin Eppler