LAMMPS WWW Site - LAMMPS Documentation - LAMMPS Mailing List Archives
Re: [lammps-users] Using LAMMPS on Cloud Supercomputers
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [lammps-users] Using LAMMPS on Cloud Supercomputers

From: Titusi Forum <titusiforum@...24...>
Date: Sat, 22 Jul 2017 08:52:54 -0700


1. free platform has 8hrs wall time limit. If you are looking to run 2weeks with 64 cores, please send an email to admin@...7010....
2. you can always call LAMMPS as library from your code. Alternatively, do you want your code to be available to everyone. If yes, then we can install it. Otherwise we can do a private setup for you. 
3. yes, not on free platform. Pls contact us if you need it.
4. Yes, we run simulations AWS, Google, Azure as well as some smaller university based clusters. Tens of thousands of cores are generally available at any point of time. All of them have state of the art hardware, reliability and availability. 

On Sat, Jul 22, 2017 at 7:52 AM, <xavier.bidault@...7013.....> wrote:

I gave it a little try. It is a very interesting tool! Here are some questions :

1) What is the limit of time for one single simulation with LAMMPS? For instance, can it run for 2 weeks with 64 cores ?

2) Is there a way to compile and use my own modified version of LAMMPS?

3) Is there a way to use more than 128 cores?

4) What are the charateristic of this HPC clusters? Does it have a name? In the top 500? What is the total number of cores ?

Best regards,
Xavier Bidault

De: "Titusi Forum" <titusiforum@...24...>
À: "Rajeev Dhongar" <rajeev.dhongar@...24...>, "LAMMPS Users Mailing List" <>
Envoyé: Vendredi 21 Juillet 2017 20:36:32
Objet: Re: [lammps-users] Using LAMMPS on Cloud Supercomputers

Dear Rajeev:
If model is running well on 128 core node then it should run fine on all other nodes.
Can you use the "Collaborate" tab on kogence and share your model privately with mukul@...7010....
I would fix the issues and revert it back private to you.

First time you fire a node, it takes about 1.5minutes for interactive display to get active. Next time you fire a simulation on same node, it will come instantaneously. Also, note that your simulations do not stop/break if you close your browser or shut your machine. You can always reconnect to visualization as long as your job has not finished.

On Fri, Jul 21, 2017 at 9:45 AM, Rajeev Dhongar <rajeev.dhongar@...24...> wrote:
Hi, This is a very useful project. 
Thx for introducing us to Kogence.
I tried running some kappa models on your site.
Stranegely, code runs well on 128core node but crashes on 16core node.
Any idea what is wrong?
I tried using Octave to do some 3D graphics.
But I am not being able to connect to interactive graphics. Can you help.


On Thu, Jul 20, 2017 at 6:22 PM, Titusi Forum <titusiforum@...24...> wrote:
Dear LAMMPS Community,

This may be useful to some of you. Kogence is offering free cloud supercomputing for LAMMPS simulations of materials. You can fire simulations on machines as good as 128cores on single node or cluster many such nodes together. This is a nonprofit effort to match excellent opensource such as LAMMPS with cloud supercompute power.

There are some 30+ public LAMMPS projects that you can fork, modify and execute. Alternatively, you can also create your own project from scratch. You can keep project private or share it with collaborators. Details are provided here :-

If you run into any issues, please feel free to contact me at mukul AT alumni DOT stanford DOT edu. I would also appreciate any feedback you might have for me.


Check out the vibrant tech community on one of the world's most
engaging tech sites,!
lammps-users mailing list

Check out the vibrant tech community on one of the world's most
engaging tech sites,!
lammps-users mailing list