Hi David, Not sure what you mean by the term "close coupled code". SPAT do occasionally have students running coupled code (such as coupled ocean-atmos models), although given the time an complexity to initiate the setup of these, we usually provide students with restricted local access to one of our LINUX NAS devices. Mike Mussard has installed our data visualisation software (IDL) on a couple of the UG cluster PCs to enable some UG project students to do some simple data analysis using this software. However, for summer and MSci students, it's usually easier to provide them with local accounts on our NAS servers than to either try and get them access to the College HPC or have specific software installed on the UG teaching clusters. Rich -----Original Message----- From: physics-departmental-computing-bounces@imperial.ac.uk [mailto:physics-departmental-computing-bounces@imperial.ac.uk] On Behalf Of David Colling Sent: 29 May 2014 19:27 To: Hudson, Jony J Cc: Carter-Cortez, Victoria; physics-departmental-computing; Mussard, Mike J Subject: Re: [Physics-Departmental-Computing] IT Facilities in Physics Do any groups have project students running close coupled code? On 29/05/14 19:18, Jony Hudson wrote:
An Amazon EC2 "micro" instance only costs $0.02 per hour, which is about £2 per week. They're good for mucking around and running long-running jobs on. A "small" costs about twice as much and is pretty capable with 1.7GB or memory and 160GB of attached storage. I use them all the time at home for things, and my bill never seems to amount to much.
Maybe worth just setting these up on demand for projects that need them? Most projects I've done have some small budget and maybe that could be used on this sort of thing. And might be cheaper than messing around setting our own stuff up?!
Anyway, just an idea.
Jony
-- Centre for Cold Matter, The Blackett Laboratory, Imperial College London, London SW7 2BW T: +44 (0)207 5942986 http://www.imperial.ac.uk/people/jony.hudson http://www.imperial.ac.uk/ccm/research/edm http://gorilla-repl.org/ http://j-star.org/ --
On 29 May 2014, at 18:46, David Colling <d.colling@imperial.ac.uk> wrote:
Hi All,
Should we actually build a small cluster specifically for projects? This needn't be very large or very powerful but good enough to perform project simulations. The College HPC is in high demand and it seems a shame to use the resources on UG projects when there are (in research terms) more important things to be doing with them.
Best, david
On 29/05/14 17:44, Mannall, Adrian B wrote:
I'm on leave but Mike Mussard can probably make a cluster machine available for Victoria to use without the log off and power saving issues.
Regards Adrian
Sent from Samsung Mobile
-------- Original message -------- From: j.spencer@imperial.ac.uk Date:29/05/2014 13:57 (GMT+00:00) To: "Colling, David J" Cc: "Carter-Cortez, Victoria" ,physics-departmental-computing Subject: Re: [Physics-Departmental-Computing] IT Facilities in Physics
Hi Victoria, David,
I frequently hit the same issue. The undergraduate labs are not suitable for running research simulations.
In the past I have got my MSci and UROP students access to the College's HPC resources (http://www3.imperial.ac.uk/ict/services/hpc). This might be suitable if the code is C/C++/Fortran/etc and the student has the support of their supervisor and has some experience with Linux (or guidance from someone who does).
Best wishes,
--James
On Thu, 29 May 2014, David Colling wrote:
Hi Victoria,
Not that I know of. I am forwarding this to the Departmental Computing list because there are helpful and friendly people from ICT on it and because I would be curious to know if this situation occurs with project students in other groups. We have spoken about a project cluster several times but decided that it would hardly be used.
Comments welcome, david
On 28/05/14 14:04, Carter-Cortez, Victoria wrote:
Hi Dr Colling,
I am and undergraduate student currently undertaking a computational project and have been trying to run my code in the Department's lab. I have attempted this a couple of times and have been signed of by the computers on every occasion. Unfortunately, the code takes a very long time to run and I expect it will take a good number of days to complete.
I was wondering if there are any other IT facilities that I could access to make this process faster?
I hope this is the correct means for making such an enquiry.
Kind regards,
Victoria
_______________________________________________ Physics-Departmental-Computing mailing list Physics-Departmental-Computing@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/physics-departmental-comp uting
-- James Spencer Computational Science Support Specialist Thomas Young Centre/Condensed Matter Theory Imperial College http://www.cmth.ph.ic.ac.uk/people/j.spencer/ http://www.github.com/jsspencer
_______________________________________________ Physics-Departmental-Computing mailing list Physics-Departmental-Computing@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/physics-departmental-compu ting
_______________________________________________ Physics-Departmental-Computing mailing list Physics-Departmental-Computing@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/physics-departmental-comput ing
_______________________________________________ Physics-Departmental-Computing mailing list Physics-Departmental-Computing@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/physics-departmental-computing