Distributed Computing and Climate Change

GraWil writes "The BBC are reporting the launch of climateprediction.net. The aim of the project is to investigate the approximations that have to be made in state-of-the-art climate models which frequently give rise to inconclusive predictions. More info on the current state of climate modeling is given by the latest Intergovernmental Panel on Climate Change (IPCC) report which highlights the need to quantify uncertainties of climate projections. So now, in addition to finding ET or curing cancer, your PC can now contribute to our understanding of climate change."

Comments

No Linux

[trolman]

No Linux version for "months." How about folding at home for those of you disappointed masses.

Heat

[Clowning]

The heat from running these distributed computing apps causes climate change inside my apartment.

Re:Heat

[gl4ss]

well, actually there's a huge amount of electricity wasted in normal heating elements used for heating.

you know, imagine the power of a beowulf cluster that was used to heat some suburb with the waste heat(and for once, not a joke :).

if you're going to use the electricity to heat you might as well do something useful with it in the process.

-

Other projects

[Krunch]

Here are some other ditributed computng projects.

Global warming

[Krunch]

> Your PC can now contribute to our understanding of climate change.

And you can contribute to climate change itself too. Let's accelerate global warming by using 100% CPU at any time.

Re:Global warming

[Krunch]

USA is said to be one of the country that contribute the more to global warming. Isn't it paradoxal that it's also the one that contribute the more to climate prediction ? http://cpdn.comlab.ox.ac.uk/map/index.html

Re:Global warming

[caffeine_monkey]

You're joking, I assume, but this is a point they've answered in their FAQ:

Won't all those computers left on for 24 hours a day have a detrimental impact on the climate system?

Assume a computer running 24hrs/day requires, on average, 50W of power. If 100,000 computers join the climateprediction.net project, the project will require 5,000kW of power. There are 24 hours in a day, so each day the project will consume 120,000kW-hrs, or 432,000,000kJ of electrical energy.

That's a big number, so let's try and put it in perspective by calculating how much energy is necessary to boil water for a cup of tea. Let's use a tiny bit of physics to do it. Assuming a specific heat of water of 4.19 kJ/(kg-K), 0.237kg/cup of water, a necessary temperature rise from 20 degrees Celsius to 100 degrees Celsius, and that only one cup of water is boiled for each cup of tea, then about 80kJ/cup of energy are necessary (assuming our kettle is 100% efficient). This means that running the climateprediction.net project for one day is equivalent to boiling water for 5,400,000 cups of tea!

Is five and a half million cups of tea a lot? According to the Tea Council, some 37 million people in the United Kingdom drink, on average, 3.4 cups of tea per day. That's nearly 126 million cups of tea per day in the UK alone!!!

Each day, about 23 times more energy will be spent boiling water for tea in the United Kingdom than would be used by the computers involved in the climateprediction.net project. More seriously, a rough calculation suggests that 100,000 computers running 24hrs/day for one year at a power consumption of 50W will contribute approximately 0.0001% of the total amount of CO2 generated in one year. This is not an insignificant amount, but seems (to us) a worthwhile investment to better understand the climate system.

Assuming you are convinced this experiment needs to be done, there are basically two options: to buy a hangar-full of PCs and run it ourselves (not even an option right now, since the climate research community doesn't have the resources); or to recycle spare CPU out in the community, as we propose to do under the climateprediction.net experiment. Since the main environmental impact of a PC is in manufacture and disposal, not the power consumed in running it (never mind the air-conditioning costs and visual impact of that hangar on some innocent rural community), environmentalists will, we hope, approve of our strategy.

No Linux - but they are trying hard

[Angostura]

There may not be Linux port right now, but I have to say that I'm very impressed with the team's open attitude and the way that they are actively attempting to support various Unix flavours, including Mac OS X as quickly as they can. They take forum suggestions seriously, so get involved.

I'm a SETI and Folding user, but I have to say that I find this project very compelling. We know that cancer is serious, and there are big businesses looking to find answers. The question of climate change is potentially more serious, in my opinion. But we need to find out for sure

Can climate be predicted at all?

[Anonymous Coward]

"In sum, a strategy must recognize what is possible. In climate research and modeling, we should recognize that we are dealing with a coupled non-linear system, and therefore that the prediction of a specific future climate is not possible."

- Working Group I - IPCC Report on Climate Change, 2001

Sounds great, but...

[ThesQuid]

Regardless of the data, a lot of people will only see this issue through the prism of their preconceived political agendas. I'm not against good data, far from it, but this is such a highly charged subject I'd like to know if they are going to be completely open about the data and the methods applied to it. That MIGHT help.

Re:Sounds great, but...

[zaphodbblx]

While I run seti@home because its something positive to believe in, I won't run these other programs for one simple reason PROFIT. Most of these other programs doing a lot of big time number crunching on things that are probably going to be patented(well except the global warming thing) and they want US to do their work for them. Will we share in the profits..highly doubtful

Re:Sounds great, but...

[A Naughty Moose]

You want a share of the profit? Buy yourself $1000 (roughly the cost of a descent modern day home pc) worth of stock in pharmacutical companies, and then donate your cpu cycles to the project. You'll be doing your part to increase the price of the stock, which will then become your share of the profits.

Re:Climate change?

[iCat]

they allready have come to the conclusion that there is a climate change

Maybe that is because Climate Change is real. You are either ignorant or ill informed. I suggest you address this.

Re:Climate change?

[hanwen]

To me it sounds like they allready have come to the conclusion that there is a climate change. Supporting scientist that allready are political biased is "dead in the water".

Wake up and smell the coffee. There is climate change, and it is very, very likely to have been caused by human intervention. Check out for yourself what the IPCC has to say on this: 2001 climate change report summary for policy makers. The question now is: how big will the changes be, and what will the consequences be? Calculating this takes a lot of CPU cycles.

Tomorrow's weather report isn't perfect either

[Tau Zero]

Right now the error bars on the climate projections are huge, something close to a 2.5:1 ratio between the best-case and worst-case projections. Even if the only accomplishment is to cut the error bar in half, that would be great progress.

It's kind of like knowing that you have a 60% chance of rain tomorrow, and knowing that the rain will be as heavy showers and will blow through between 1:30 and 4:45 PM. The latter information is far more useful for planning your day than the former.

Re:glowbull warmongering

[Krunch]

I understand the joke but the fact is while USA represents a very small percentage of the world population, it is responsible for 20% of total CO2 emission in the world (pdf). CO2 is well known to contribute to global warming and some countries engaged themselfs to reduce their CO2 emission. But not the USA... This is going offtopic and it gives a negative point of view of US. Mod me down.

Experiences of a beta tester

[grid geek]

I've been running climateprediction.net as a beta tester for the last couple of months. My experiences with it so far have been good, running it on a PIII-733 and a AMD 1GHz Duron laptop. No major crashes or faults.

Compared to SETI each work unit takes forever. None of this one unit every 8 hours business, when they say it takes a committment they mean it, 90 days of 24/7 operation to finish one unit on the Duron, so I guess there is unlikely to be anyone hitting the 100,000 unit mark any time soon!

A bit about the program - The work unit itself is broken down into 3 segments. There's an upload of results so far at the end of each one and a daily connection to confirm how much cpu time you've used in the last day and what checkpoint you've reached. If you don't do this it doesn't ask, it just checks if you have a connection and if not waits until you do. The program check points every couple of minuutes but can roll back a bit if you reboot (not a huge amount but its not as frequent as SETI).

Overall I've had no problems with it apart from it crashing out of virtual memory once when I'd left it running without a network connection for 2 weeks.

Re:Experiences of a beta tester

[pointwood]

Try checking one of the big primes from http://mersenne.org/ - that'll take you a year or something like that on a top end machine :)

This isnt for charity

[Anonymous Coward]

please be aware that this is a commercial project

http://www.climateprediction.net/misc/sponsors.php

all of those companies SELL services based on this data, so iam sure they would very much like the public to do their work while they sit back and reap all this lovely free data, even the UK Goverments Met Office isn't free and if you would like weather data (like what its like in your area) you have to pay for it (unlike the USA which offers access to its data streams/imaging for free)

so go ahead if you want to donate your CPU to companies such as
" Risk Management Solutions (RMS) is the world's leading provider of products and services for the quantification and management of natural hazard risks."

then go right ahead, Me ? ill just keep looking for aliens thanks, at least mankind will benefit instead of a few shareholders in a faceless corporation.

Registration server seems in trouble

[Frodo420024]

As others have noted, getting the client to accept the registration seems buggy. I've tried a dozen times, and my registration is now known to the server - but the client refuses to accept it, thus I cannot start calculating.

Either a bug in the registration process, or /. has hit hard again...

inconclusive predictions?

[Anonymous Coward]

I predict the chance of inconclusive predictions today is 60%

Good distributed computing client software?

[InvisiBill]

Has anyone come up with a decent client program for any of these projects? I've been running distributed.net's RC5 client for years. Of all the projects I've tried, it's the only good program I've come across. It actually runs as a service with 0 priority, so it really does use unused cycles, unlike the screensavers which only work when you're away from your PC (and if you use a screensaver). Have the other programs gotten better, or are they the same as when I looked at them way back?

Distributed computing changes climate...

[Junior J. Junior III]

See, all those jokes about overclocked Athlons contributing to global warming are now coming true.

LOL

[sw155kn1f3]

Did someone imagine a distributed cluster all around the planet heating atmosphere to the point no one lives on the earth anymore?
What do you need to cool it? Oceans?

Running CPDN on Linux and some other things.

[BuilderBob]

Rumours on the forum are that it can be run under WineX or some such things, as well, of course, as VMWare.

There is no problem with running the model on Linux though, the model itself will run under any operating system with enough power, it was originally written for a Cray and is still used at the Met Office on Cray-like machines (specifically a a T3E, I think).

The model will (and does) currently run on Linux, quite happily, the problem with running CP.net on Linux is that the program used to control the model is currently windows only, as is the visualisation software.

As for running the model without the control program, there are two problems, the first is that the interface is....not good. It uses Fortran namelists for most of the non-compiled variables and input files with specifications that were dreamt up by Satan on LSD (It's always a good sign when the program itself doesn't follow the file specifications). The CP.net team have created a "virtual grad-student" (their words) which will look after your model and redo any calculations it needs, as well as deciding when to report back to CP.net and take a coffee break. Having sat waiting for the model to run/crash I wish I had a toy like that, even if I did have to make the coffee.

The second problem is that the model is balanced on a knife edge. There is a continuous battle between realistic physics (more complicated functions, shorter integration timesteps, slower model) and getting some work done (longer timesteps, simple physics, etc.). A part of this project will be to find out which parameters can be changed in such a way as to make the model fall over and become an ice planet or any of the other non physical but numerically feasible solutions.

It will take a long time to run each model, as the website says, but this is pretty much the simplest model which would produce a useful result, even on a 2.6Ghz Athlon you won't get more than about a day every six minutes (3 minutes for the atmosphere, 3 minutes for the ocean) for the full model, 50 years is 360*6*50 = 108000 minutes (75 days) on 24/7, luckily (?) a good portion of the models will fail before then, some will take longer as the results are checked if they look extreme. The real physical differences produced will only be a subset of the results from the experiment.

The model can go faster, e.g. a variation has been developed by the MetOffice where the Ocean model can runs upto 10 times faster than in the CP model, the main reason for this speed up? Iceland was deleted from the map :) (in terms of size, I think Ireland and the UK are next)

The data which will result in this project will hopefully be able to give a quantitative prediction of how bad things might get if we (say) double the amount of carbon dioxide in the atmosphere, it can tell us (=scientists) how likely it is that New York will be flooded or El Nino will shutdown. Whether or not the data will be 'open' is anybody's guess. Checking the sponsers, at least one of them is an insurance company that insures based on weather forecasts (good crop weather, flooding, etc.), I have little doubt about the commercial value of the data (c.f Cancer research programs).

As for the people who want access to the model source (and the source for the visualisation programs I guess). Are you completely out of your mind :) It's half a million lines of Fortran which has been written by many many people over about 10 years. Having access to the source in this case would benefit nobody. It only does 1 thing, model the climate. The atmosphere model has about 50 different options for the physics schemes, 10 different dynamical schemes and noise filtering options, all of which need to be set up properly to have any chance of working. The 'simple' ocean model has another load of options, then the 'complex' ocean model has another load of options, then there are multiple way to couple the atmosphere and ocean together. (Also, *shock horror*, it has bugs in it.)