Posts by ojum-le

1) Message boards : Number crunching : wah2_eu25 work units are BIG (Message 54595) Posted 1 Aug 2016 by ojum-le Post: Think about HadCM3-Model some years ago. A 160-year-simulation takes round about 6 month of calculating-time on a i7-920-machine. CPDN always was a project who need endless PC-power. I was shut down my CPDN-computing because of the hich energy-prices in germany :-( An i7-920 takes 260W always. Over the year it causes about 1000€/1100$ of energy-costs every year. The only senseful way to use CPDN in a private case is to use high-efficient Power-platform. Broadwell-EP (LGA2011-3) in a low-power design would be the efficient way to participate CPDN for credit-hunting. The 2-channel-platform (1151) is not really recommend for CPDN i mean. The FPU-Power is too much less for the energy-consumption. CPDN is a typical high-FPU-based computation. Look at spec.org to identify the best performance of "481.wrf" in the SPEC2006-FP-Rates-benchmark. These benchmark induces directly the CPDN-performance of the system. There you can see, that an actual Skylake 1151-CPU only has 80 Percent more FPU-power over an i7-920. Than take a look to an Broadwell/Haswell-EP-platform and u can see a much stronger increasement of FPU-power - depends on core-numbers - but the whole performance and efficiency in not comparable to S1151-platform.
2) Message boards : Number crunching : Reporting - Errors while computing - (Message 46346) Posted 2 Jun 2013 by ojum-le Post: Try to clean up your data-directory. C:\Boinc\data\projects\climateprediction.net\ Delete all files. I had the same issues like u.
3) Message boards : Cafe CPDN : Astonishing web pages (Message 46343) Posted 1 Jun 2013 by ojum-le Post: I love Schweinebraten :-)
4) Questions and Answers : Windows : Optimise PC build for CPDN (Message 46287) Posted 25 May 2013 by ojum-le Post: http://climateapps2.oerc.ox.ac.uk/cpdnboinc/result.php?resultid=15736358 <core_client_version>7.0.25</core_client_version> <![CDATA[ <message> Das Laufwerk kann einen bestimmten Bereich oder eine bestimmte Spur nicht finden. (0x19) - exit code 25 (0x19) </message> <stderr_txt> [hadcm3n_data_6.07_windows_intelx86.zip] End-of-central-directory signature not found. Either this file is not a zipfile, or it constitutes one disk of a multi-part archive. In the latter case the central directory and zipfile comment will be found on the last disk(s) of this archive. unzip: cannot find zipfile directory in hadcm3n_data_6.07_windows_intelx86.zip, and cannot find hadcm3n_data_6.07_windows_intelx86.zip.zip, period. Called boinc_finish </stderr_txt> ]]> My RAID dropped on this day. One if the 15k.7 had caused many struggle and the system crashed. But now it works correct. I think there were lost some files who was in SAS-Controller-RAM. But actually i dont get new work :-(
5) Questions and Answers : Windows : Optimise PC build for CPDN (Message 46276) Posted 23 May 2013 by ojum-le Post: If u can wait some months you can build up a socket-2011-System with Ivy-Bridge based Xeon-EX Series up to 15 cores per CPU. The Socket 1567 of the Xeon-E7 (Westmere-EX) reachd EOL and will be replaces by existing S2011!!! With these platform you can use: Sandy-Bridge-EP (Xeon E5-2600) up to 8 cores Ivy-Bridge-EP (E5-2600v2) up to 12 cores comming Q3/4 2013?! Ivy-Bridge-EX Q3/4 2013 up to 15 cores All with quad-channel RAM-interface and ECC support. S1155 based I7 with dual channel RAM is much slower in CPDN. A good way to find out the forecast-performance is the SPEC2006 benchmark (http://www.spec.org/cpu2006/). Ther is a Sub-benchmark in CFP2006 called 481.wrf - a weather forecast model based on fortran models similar to our models that needs much FPU-power. E5-2620 (350�): 204 points at 481.wrf 2xE5-2620 (700�): 406 points I7-3770k (270�): 129 points at 481.wrf PS: Haswell I7 (s1150) only increases 8 percent FPU-performance in comp. with Ivy-Bridge! :-( But s2011 based systems need much of energy. My I7-920 uses 260 watt electrical power when runs on full-load (160watt in idle). E5-2600 based systems will not take less energy i think :-/ Best power/perfomance have the mobile CPUs by Intel. But this CPUs u only find in notebooks and ITX-sytsems. SSD: 2 years i run CPDN on Intel Postville G2 80GB, and 2 more years with Crucial C300 128GB an i had no issues with them. Actually CPDN runs on Cheetah 15k.7/600gb because the installed SSD is to small. In summer i change to Samsung SSD 840 (TLC) with 512 GB. I think it is not necessary to use SLC. Good quality consumer-ssd (Samsung, Crucial, Intel, Toshiba, ...) should be adequate for CPDN. The cheapest SLC is WINKOM Powerdrive 60GB on SATAII cost 170�! For this money u get 256GB!
6) Message boards : climateprediction.net Science : Climate is too complex for accurate predictions (Message 46222) Posted 14 May 2013 by ojum-le Post: Most of the Links are down - pleas try to update the links. The numeric models like HadAP/CM have a general uncertainty. They only will calculate what the programmers have implemented. Many causes in climate are known and described scientifically correct, others are unknown. The models can only create (calculate) what we would exped. I think, at the end of 2010 decade we will get a basically new generation of computers for scientific problems. The answer for actual uncertainties could be the cognitive computers based on neural network techniques in opposition the von-Neuman-architecture. Few months ago i found a paper (http://journals.ametsoc.org/doi/abs/10.1175/2008JCLI1869.1) "Constraints on Model Response to Greenhouse Gas Forcing and the Role of Subgrid-Scale Processes" in which they used a software neural network to improve parameter-settings for Had-Models we calculated. Also 2001 was published a dissertation "Zur Anwendung Neuronaler Netze in der Klimatologie" to the neural network-techniques by Andreas Walter (german weather forecast DWD) find on amazon (http://www.amazon.de/Zur-Anwendung-neuronaler-Netze-Klimatologie/dp/3881483780). He said "neural networks have big potential for modeling non-linear Input-Output-Relations. If we found a way to set objective criteria, with neural networks the climatology can be get a huge knowledge acquisition". The original paper is in german language. Computers who can learn autonomously und find coherences in the System we actually dont know/see. Neural Networks are already available only in Software. Next step was the presentation of a Processor based on Neorons and Synapses based on the structure of human brain called "Synapse" and evolved by IBM in 2011. Maybe this fully new architecture can revolutionize the climate-prediction. Then we talk about speed-increases in dimensions. So be patient...
7) Message boards : Number crunching : Project has no tasks available (Message 45209) Posted 2 Nov 2012 by ojum-le Post: yeah, 8x HadCM3 160-years with ~1500h per WU an my PC have employment for some months. Otherwise we are lucky, because we do not consume so much expensive electrical power. Last year ~2200 kwh per PC (but with "green enegry" :-) ). I hope this time-out doesnt take to much time, and when CPDN will restart we got some new numeric models with 64 Bit and Multi-Core-Support. http://climateprediction.net/science/pubs/volcompearthsci092009.pptx - last page In between i think its not wrong to find out any problems on the server hard- and software. Weeks of unforseen disreachability (PSU dead, HDD full, Databank inconstistant, Hacker-Forces) are really not cool. Other question: Will our last years calculatings build up one of the fundaments of 2013 IPCC-Report?
8) Message boards : climateprediction.net Science : When comes HadGEM-Models? (Message 45099) Posted 16 Oct 2012 by ojum-le Post: On study a paper of climate-simulation with usage a neuronal network in climate-modelling i found in the conclusion, that the results of this experiment can help to set up better initial-vectors/parameters AND the new next generation Models HadGEM are based on these knowledges. Then i saw, that the MET-Office (http://www.metoffice.gov.uk/research/modelling-systems/unified-model/climate-models)is already using these new Models. The IPCC 2007 is saw is based on HadGEM models. Now the question, when will arrive the next generation models the millions of Home-Computers in CPDN Network? Are the original HadAM/CM also run on "standard PCs" in scientific climate institutes with x86-cores?, whats about parallelism? And in comparison with real scienfitic computers, how more or less performant is the CPDN-network actually? thanks PS: in these Semester i visit a study course of Prof. Stefan Rahmstorf from PIK-Potsdam :-) Maybe i could ask him personal too, aboput the computers an models and CPDN.
9) Message boards : Number crunching : Intel Xeon Phi Support? (Message 44760) Posted 29 Aug 2012 by ojum-le Post: Hello, Intel's High-Performance Computing Card "Xeon Phi" will released soon. This Card have a FPU-Power of 1TeraFLOP. It looks, Xeon Phi is based on old Intel x86 Core-Design (Pentium with SIMD and AVX) with 62 cores connected with a ring-bus-system. This card have own Memory (min. 8GB)and an own LINUX-Operatingsystem which communicates with the HOST-OS via TCP/IP via PCI-Express. Question to the Engeneerers of CPDN: Could/would it be possible to use this special HPC-Card for Climate-Modelling? Could this Card solve the problem of existing GPU-calculation for CPDN which can not run, because we need x86 compatible cores. Bye
10) Message boards : Number crunching : Upload Failure (Message 44108) Posted 27 Apr 2012 by ojum-le Post: I also hope for solving the problem: in que: 48 files aprx. 850 MBytes! 5 EU-Models finished
11) Message boards : Number crunching : More FPU or Integer Power needed? (Message 43613) Posted 27 Dec 2011 by ojum-le Post: your calculation is not correct i think. You have to multipy results with the core-number: Integer: 11954 (4core)= 11954x4=47816 6690 (8core) = 6690x8=53520 (some little increase) FPU: 4360 (4core) = 4360x4=17440 3767 (8core) = 3767x8=30136 (giant increase, same TDP) It seems Intel SMT could boost your calculating power, especially FPU-calculations got most profit of SMT. But this is theoretical Benchmark, and how we know, HadCM3N-Models 'hate' Intel-SMT. Maybe, the best Benchmark for CPDN is the 481.wrf at the SPEC CPU-Benchmark. http://www.spec.org/cpu2006/Docs/481.wrf.html Intel's I7 3960x generates 233 wrf-Points! Best Value ever for signle-CPU. http://www.spec.org/cpu2006/results/res2011q4/cpu2006-20111121-18924.html Otherwise i think, CPDN-Executes doesn't use all Compiler-optimizations for modern and different CPU-types. Is there any outlook to run the models on nVidia's TESLA-Chip?
12) Message boards : Number crunching : persistant miscalculation of RAC? (Message 43501) Posted 3 Dec 2011 by ojum-le Post: Hello, Since i am running 8 HadCM3N Models on my I7 the recent average value goes in the dumps. http://de.allprojectstats.com/show.php?projekt=21&id=1175444 My machine got 3000-4000 credits per day, but the RAC gows down and down. Actually the RAC for this computer is ~1700! How could this happen? Could you be able, to fix this miscalculation? The BOINCStats own calculated RAC shows a value, that looks trustworthyer :-) http://boincstats.com/stats/host_graph.php?pr=cpdn&id=1175444 Byebye
13) Message boards : Number crunching : More FPU or Integer Power needed? (Message 43445) Posted 21 Nov 2011 by ojum-le Post: This is exactly the Reason, why i go back to AMD. For a long time Intel CPUs were the best in Price an Price/Money. But when i look to the actually prices of XEON and I7, i think AMD is the most better choice. If you only want power, u have to use a Intel but for what price?! Some GFlops more will cost u some hundreds of Dollar/Euro more. Next year AMD offers an Single-Socket Opteron 3000 up to 3Ghz on Socket G34! And next yea AMD launches a new Server-Platform with "terramar" CPU up to 20 Cores. :-) Intel is OUT and only for people, who have to much money. PS: yes i also saw, that the power of 1100T and 1090T is very good for price.
14) Message boards : Number crunching : More FPU or Integer Power needed? (Message 43424) Posted 18 Nov 2011 by ojum-le Post: Sooo, Let's have a look to the SPECs new Opteron 6200 Series: The SPECfp results of the Opteron 6276 (16c/2,3Ghz) looks very very good: http://www.tecchannel.de/server/prozessoren/2038251/cpu_test_amd_opteron_6262_he_und_6276_mit_16_core_bulldozer/ The 5th or 6th diagram "Specfp_rates" and on spec.org u can see the details: http://www.spec.org/cpu2006/results/res2011q4/cpu2006-20111025-18763.html http://www.spec.org/cpu2006/results/res2011q4/cpu2006-20111025-18753.html this is a 2-Socket-Machine => ~360GFlops!!!! with old 2x6180SE Top-model: http://www.spec.org/cpu2006/results/res2011q1/cpu2006-20110131-14298.html 320GFlops Intel Xeon E7-4870 10Core 2-Socket: http://www.spec.org/cpu2006/results/res2011q2/cpu2006-20110425-15899.html 370GFlops BUT: Price of E7-4870: ~4000Euro per CPU Opteron 6276(2,3Ghz): 750Euro!!!! Opteron 6272(2,1Ghz): 480Euro! I think the new Opterons could be the best CPUs in power AND price for next months or years! My next System could only be a Multi-Socket-Opteron :-) ... then i discovered the SPEC-descriptions on: http://www.spec.org/cpu2006/CFP2006/ "481.wrf C,Fortran - Weather - Weather modeling from scales of meters to thousands of kilometers. The test case is from a 30km area over 2 days." Could this the benchmark, with the most concordance to our Climate-Models? PS: Sandy-Bridge-E I7-3960X reaches 140 GFlops - haaahaa :-)
15) Message boards : Number crunching : More FPU or Integer Power needed? (Message 43397) Posted 16 Nov 2011 by ojum-le Post: Hello, for next year i am planning a System-Upgrade to a Multi-Socket-System. Now i think about, whats the better CPU-Brand-choice for CPDN? The coming up Opteron 6200 Series (Bulldozer) has an increased Integer-Power (~220GFLOPS with 2x6276) cause AMD means this were the most used kind of CPU-Caltulatings. So the Opteron 6200 has comparatively "bad" Floating-Point-Performance. For FPU i think Intel is the better choice. But what we do need for CPDN primarily? thanks
16) Message boards : Number crunching : thank you Les Bayliss for Information in News and Announcements (Message 43337) Posted 30 Oct 2011 by ojum-le Post: It seems, 2011 was the year with the most server-outages in all cpdn history :-) Let's hope, that next year differs form actually.
17) Message boards : Number crunching : thank you Les Bayliss for Information in News and Announcements (Message 43327) Posted 30 Oct 2011 by ojum-le Post: The main-question is: When will the scripts restarted? Some days or weeks or never?
18) Message boards : Number crunching : wrong RAC? (Message 42853) Posted 3 Sep 2011 by ojum-le Post: Hello, http://de.allprojectstats.com/show.php?projekt=21&id=950024 (end of page) This is one of my computers, participating CPDN. The Average-Value (RAC) couldn't be correct i think. Last 2 Weeks constantly more then 3500 credits/day, but RAC jupmps not over the 2300c/day-mark. What causes this discrepance? Thanks, mujo
19) Questions and Answers : Preferences : i7 920 Performace-Problem (Message 38576) Posted 23 Dec 2009 by ojum-le Post: Hi, PC is runnging fast, but there is another Problem with BOINC. Since yesterday, my Graph in the Program is some strange: A Reboot doesnt Help. As i Upgraded my System, i had complete reinstalled My OS on a formatted Disk. But With the new installed OS i get NOT a new PC-Number. Its the same like months before with the Q9550. http://climateapps2.oucs.ox.ac.uk/cpdnboinc/show_host_detail.php?hostid=950024 Same ID only CPU and RAM was changed, maybe the is the Problem? Today i tryed to join WCG with the Error \"No Work for this Type of computer\", i dont know the original english text, cause i use german Interface. For the Future: I will never install an OS at a Friday.
20) Questions and Answers : Preferences : i7 920 Performace-Problem (Message 38573) Posted 23 Dec 2009 by ojum-le Post: Hello, Now i have placed an another 2GB-Module. Now i have 4 GB in DualChannel-mode. I must wait for the next first Trickles, but i feel its much faster with 4 as with 2 GB. Maybe next Year comes an Upgrade to 12GB, but not with the actually RAM-Prices. So, i can Crunch now :-) Merry Christmas and New Year to the CPDN Operators and ALL Users.

Next 20