Message boards :
Number crunching :
No Tasks Available
Message board moderation
Previous · 1 · 2 · 3 · 4
Author | Message |
---|---|
Send message Joined: 27 Jan 05 Posts: 16 Credit: 790,158 RAC: 0 |
Keeping a machine going exclusively for ClimatePrediction crunching doesn't sound right. |
Send message Joined: 16 Jan 10 Posts: 1081 Credit: 6,972,865 RAC: 3,926 |
More than 5,000 new EU models in the queue now ... |
Send message Joined: 30 Nov 10 Posts: 4 Credit: 114,510 RAC: 0 |
Why not? |
Send message Joined: 15 May 09 Posts: 4341 Credit: 16,497,933 RAC: 6,477 |
Well, they didn't last long. Only 4 left now. Still 3 Moses ones. yet to see if they are re-issues or not. |
Send message Joined: 31 Dec 07 Posts: 1152 Credit: 22,058,355 RAC: 529 |
Why not? Some people worry about the added carbon footprint of running a machine just to try to predict the effects of a larger carbon footprint. |
Send message Joined: 30 Nov 10 Posts: 4 Credit: 114,510 RAC: 0 |
I'm on 100% renewable energy. |
Send message Joined: 18 Dec 13 Posts: 62 Credit: 1,078,935 RAC: 0 |
I'm on 100% renewable energy. Due to the economics of electricity production, unless you are either off-grid or on a 100% renewable local grid, demand reduction also counts. This does complicate things when running the processors harder to compute complex fluid dynamics models adds to power drain on the computer, but the latter would seem to be an acceptable compromise. I'm also getting my electricity from a company generating only through renewables, but it's provided through a pooled national grid, and it's a pretty still night. |
Send message Joined: 31 Aug 04 Posts: 391 Credit: 219,888,554 RAC: 1,481,373 |
What I'm figuring -- there are maybe 200000 -that's the most contributing anything here, ever, and 500 watts per machine (less per cpu) that's what -- maybe - exactly 100 megawatts. 1 tenth of a big nuke plant, maybe 20 big diesel locomotives. Less than one supertanker. For the whole lot of us computing. Unless that math is totally wrong, and it could be, I'm never going to worry about my contribution here contributing to global warming. (one ten-thousandth of one percent -- sheesh) But I am running on 10% renewable and 90% nuclear, which is almost as good. But, please people, do not worry that running cpu cycles here is poisoning the atmosphere. Do the math! Please. <edit> PS when I said nuclear, I meant fission, like uranium. NOT the free thermonuclear (but not renewable) power that comes from the nearest star. |
Send message Joined: 27 Jan 05 Posts: 16 Credit: 790,158 RAC: 0 |
Eirik, if BOINC uses only spare capacity of machines already in use, that would be 0 tenths of a big nuke plant, 0 big diesel locomotives, 0 supertankers. (I'm not sure how units of locomotives and supertankers can be used to measure rate of electricity use, mind). Using your maths, if there are 10 other BOINC like projects in the world all running machines doing nothing but BOINC, there would be one unnecessary nuclear power station and all the unreconciled waste that brings. Please people, do worry about leaving your machine on just to run BOINC, it poisons the atmosphere. One ten-thousandth x 7 billion = lots. Niall, for those of Eirik's persuasion, please invent something which lets machines run only when the wind's blowing. |
Send message Joined: 15 May 09 Posts: 4341 Credit: 16,497,933 RAC: 6,477 |
Starting to wonder if the use of computers to crunch climate models and the energy used should have it's own thread? Eirik, if BOINC uses only spare capacity of machines already in use, that would be 0 tenths of a big nuke plant, 0 big diesel locomotives, 0 supertankers. Not true, computers are not like class A amplifiers which use the same energy regardless of volume. They use significantly more energy when running flat out which they do when crunching. I too am on a plan that states I only use renewable electricity. This does not absolve me from responsibility in conserving what I use however as if I use less, then on a grid system, that energy can be used elsewhere meaning less oil/gas etc is needed to power someone else's needs. Against that, during winter, the computer reduces my need for other heating. I am sure there is a lot more to be said on this but I will leave that to others. |
Send message Joined: 27 Jan 05 Posts: 16 Credit: 790,158 RAC: 0 |
Well I'd accept that argument if it were true but it depends on this premise "They use significantly more energy when running flat out which they do when crunching". Here are some measurements from University of Pennsylvania which make that premise questionable. They do measurements when running SETI or Folding: https://secure.www.upenn.edu/computing/resources/category/hardware/article/computer-power-usage |
Send message Joined: 1 Jan 07 Posts: 942 Credit: 34,163,943 RAC: 5,381 |
Well I'd accept that argument if it were true but it depends on this premise "They use significantly more energy when running flat out which they do when crunching". That's not what the Kill-a-Watt meter plugged into my wall outlet says. I'll grab some figures later - both the machines I have power monitoring on are rather busy just now. I'm wondering if they've accounted for the modern BOINC default settings which suspend activity (especially GPU activity) when they detect user activity? Most people who do flat-out crunching probably change those default settings. |
Send message Joined: 1 Jan 07 Posts: 942 Credit: 34,163,943 RAC: 5,381 |
OK, I promised some figures. Both sets of figures that follow apply to the 'system unit' - mini-tower computer - only: monitor and all other ancillaries are separately powered. Power figures are measured at the wall socket with a 'Kill-a-Watt' style plug-through meter. Power draw displayed on the meters fluctuates, so I've given an idea of the range. 1) Dell Optiplex 9020 - a modern off-the-shelf business machine, less than a year old. Fitted with Intel i5 'Haswell' CPU, which has an integrated HD 4600 GPU. At idle - Windows desktop displayed, but no applications running: draw 25-27 watts. Under load - BOINC running five tasks, four on CPU and one on iGPU: draw 80-102 watts. 2) homebuilt high-power PC. It could probably be classed as a gaming machine, but I use it for crunching instead. CPU is Intel i7 'Ivy Bridge', also with integrated HD 4000 GPU. Also fitted with two NVidia GTX 670 GPUs, and multiple fans. It's about two years old. At idle - Windows desktop but nothing else: draw 85-87 watts. CPU loaded - BOINC running six tasks, GPU crunching suspended: draw 147-150 watts. Full load - BOINC using all three GPUs in addition to the CPU: draw 380-390 watts. I justify the power usage because of the scientific work I'm doing (as well as it being an interesting hobby). The machines are both in a domestic setting in a cool country (northern England), so the exhaust heat is welcome - yes, even in June! I'd find it harder to justify using the power in either a warm country, or in a high-density data center, where extra power has to be expended on air conditioning to extract 'waste' heat. |
Send message Joined: 22 Mar 06 Posts: 144 Credit: 24,695,428 RAC: 0 |
Ahhh goody, a discussion that isn't about credits and model failures (and that should have its own thread). Every now and then I too wonder if it is all worth is, should we be clocking up all these kW, should we ...., but I always come to the same conclusion. Yes. There is no way I can figure out if the carbon generated is worth it, so I look to a higher power - the IPCC. The day the IPCC says enough is enough, then I will pull the kill switch and stop, but until then I keep chugging on. Afterall, is there anyone who would disagree with the 97% ;-) I can see the models changing as time goes on until one day it will be mitigation modelling that takes priority (yeah, call me the doom munger :-( ), but there will be models for some time to come. I'm lucky in that I live in a country that has 60-70% of its grid electricity generated from renewable sources, but there is still a CO2 & monetary cost. Earlier this year I updated my PC and optimised it for CPDN, whilst still cracking through all my work and home stuff. It's a well equipped Xeon, currently running 10 tasks, and here are some figures. PC at idle, no screen. 85 Watts. Screen only no PC. 46 Watts. CPDN 10 tasks overhead. 92 Watts. (fewer tasks will use less power and vice versa.) To avoid confusion the total power used will be the total of those numbers (223 Watts) when I'm sitting in front of the screen with CPDN working. If I assume the PC is on 12 hours a day for work and personal use, the additional overhead of using the PC for CPDN work 24/7 is 1180 kWh per year, compared to roughly 660 kWh if I just used the PC with no CPDN during work hours. As I only use 5,800 kWh per year (and no gas) it's a sizable chunk, but worth it. |
Send message Joined: 31 Dec 07 Posts: 1152 Credit: 22,058,355 RAC: 529 |
I see that we are back to no work being available. That little batch of hadam3p_eu�s were nice while they lasted, but, they didn�t last long. I guess that�s what happens when you have a pack of hungry computers waiting to gobble them down. Hope there will be more soon. |
Send message Joined: 31 Dec 07 Posts: 1152 Credit: 22,058,355 RAC: 529 |
I see that the Scientists behind the hadam3p_eu project have dump a ton of new WU�s into the hopper. I can�t remember when there were more than 50,000 WU�s available at one time. |
Send message Joined: 30 Jan 14 Posts: 70 Credit: 60,900 RAC: 0 |
Hi Jim, Yes, a big new batch of runs! These are a continuation of our recent weather@home 2014 UK Flooding experiment. If you recall, that experiment involved running an ensemble of models that represent "the world that might have been" without climate change, and another ensemble based on actual observations of the recent winter. This new batch is a subset of the natural ensemble (based on actual observations). We want to increase the number we have of these so we have the same size ensembles for all the subsets of the natural ensemble (this makes the statistics more robust). More specifically, we want to find out whether different models used to generate the natural SSTs (see the experiment setup page for more details) lead to significantly different answers to the question of whether or not climate change altered the risk of extreme precipitation. I hope this helps! Best wishes, Hannah Hannah Rowlands -- No longer Communications Officer for climateprediction.net, as of October 2015 |
Send message Joined: 15 Feb 06 Posts: 137 Credit: 33,452,399 RAC: 5,451 |
Thanks Hannah, It is always helpful to know what the batches of tasks are hoping to achieve. For those of us with a science background, the information you give makes our number crunching much more meaningful. Please keep us up-to-date when new batches of tasks are sent out. Thanks again. |
Send message Joined: 29 Jan 12 Posts: 3 Credit: 79,756 RAC: 0 |
Hi, Hannah. Thanks for the info! I just returned to this project after a long absence. I am excited to be involved again and looking forward to new and different experiments in the future. Bill |
Send message Joined: 31 Aug 04 Posts: 391 Credit: 219,888,554 RAC: 1,481,373 |
Yup - more work welcome, Thanks for saying what it's worth.. Will keep contributing.. |
©2024 climateprediction.net