 |
|
|
1)
Message boards :
Cafe CPDN :
Scotland team
(Message 46323)
Posted 31 May 2013 by old_user596405 Post: I got a couple too but my cache is set wide open at 10 days so any new ones floating about i can catch :) Got a bunch after spotting Margaret's helpful spot about new models. Thanks! What I had to do was increase both caches to 10 days until new models downloaded after setting no new tasks for GPUGrid. Then revert to zero + 0.01 days for a special GG reason. Er, now have 34 'live' models (run in rotation depending on number of cores available - currently 13 at a time). Helped by using a third PC (oldest, replaced by a new build) for GPU testing until ready to sell. Sorry about greed but aim is to reach a milestone of completed models before winding down CPDN to share cores with WCG on a rotational basis. Will be lucky to hit milestone by end of this year so much CPDN crunching to do! Apparently there are new projects in the WCG pipeline (of special interest to me) as well as an improved user interface. Standard BOINC, at last?
Yes, last night started to update stats which quickly reported Dave Roberts. Unfortunately 1 second later the script canceled with a "member left" message. Member Kevin who is no longer active. This is a lie. No time to dig into code (already one other known bug) as I'm mega busy and very code rusty. Will try a manual workaround today. Incidentally I thought we had a Dave Roberts in the past. Like the BBC project but definitely not there (in Iain's final BBC stats). Good to get the occasional new member though. Perhaps being in page 1 of leading teams' tables does help - as someone recently commented :) Cheers ism |
|
2)
Questions and Answers :
Getting started :
Cannot attach to climateprediction
(Message 46298)
Posted 28 May 2013 by old_user596405 Post: Success! Tried the command line method. ..BOINC\BOINCCMD --attach_project http://climateprediction.net <strong 32 character account key (from the Your Acccount page)> Note that this requires the BOINCCMD executable not the main BOINCMGR one which I had incorrectly tried first time. Thanks Mike for the clue about the command line method. Details from the BOINC manager help. Back in production. Even got a single European WU right away. :) Being a temporary system (until sold), instead of deploying regular security software, am just using Windows Defender (set to ignore BOINC folders) as well as Windows Firewall. Taking Mike's pointer about allowing firewall to run BOINC manager, have now done that - just to make sure. Does not really explain why GPUGrid and WCG both attached ok though! Ian |
|
3)
Questions and Answers :
Getting started :
Cannot attach to climateprediction
(Message 46296)
Posted 28 May 2013 by old_user596405 Post: How are you attaching from the new Boinc installation? I connected my new PC via the wizard and it seemed to work OK (that was about a month ago). New Windows 8 install on a clean drive then a new BOINC install. Started BOINC then Tools / Add project or account manager / Add project / selected CPDN - i.e. the usual method. Or is there another? When starting BOINC for first time it immediately started the wizard. Same problem. Been ages since I reattached CPDN. |
|
4)
Questions and Answers :
Getting started :
Cannot attach to climateprediction
(Message 46293)
Posted 28 May 2013 by old_user596405 Post: Hi Knew about this problem in March. Thought it was sorted. Having rebuilt an old machine prior to selling it, over next few weeks will use it to test graphics cards running GPUGrid. Thought that I might as well crunch some CPDN models at the same time (8 on this i7-920 !) until GPU tests are all satisfactorily completed. Ok, models are currently scarce but do get the odd one or two getting to my two main systems - usually "full resolutions". Sadly, after a new installation, BOINC refuses to attach (project not found). GPUGrid attached ok. In the meantime, have attached and downloaded WCG units just to keep the old system busy :) Moan IanSM |
|
5)
Message boards :
Cafe CPDN :
Scotland team
(Message 46178)
Posted 10 May 2013 by old_user596405 Post:
It's been a while since we had a new member who went on to make a good contribution. Meantime I'm putting a spanner in the works! Have built two new systems to replace two older rigs but with half the number of cores available for CPDN crunching. i.e. 2 x i5-3570k each with 4 "hard" cores but no additional (hyper-) threads to replace the older i7-9xx systems each with 8 threads. The graphics cards (for GPUGrid) also need dedicated CPU cores so we are down from a maximum of 12-14 running models to only 4-6 (depending on number of GPUs in use). The i5s are quicker so may be equivalent to 7-8 of the first gen i7s. Enough of the techy stuff. Bottom line is that my RAC should drop from over 5k recently to maybe around 2.5k. We shall see the next time a new batch of work units is released. Sorry about that! IanSM P.S. Someone could kindly ask IanB if he might upgrade his 1st gen i7s to current 3rd gen i7s (which still provide hyperthreading)! P.P.S. Now have one i7-920 (8 model) cruncher for sale which has made a good contribution to the project over the last 3 years! Not allowed to advertise here though :) |
|
6)
Message boards :
Cafe CPDN :
Scotland team
(Message 46167)
Posted 8 May 2013 by old_user596405 Post: WOW, not only are we at 8th in RAC today but we've also moved into 20th place in total credit, so on the front page of the team rankings in both categories - well done everybody! ;-) Just spotted this myself. You beat me to it !! First time in top 20 for both total and RAC. Ian |
|
7)
Message boards :
Cafe CPDN :
Scotland team
(Message 45973)
Posted 19 Apr 2013 by old_user596405 Post: Ian, Depends how many CPU cores you are running 100% bearing in mind that GPUs will grab one or more cores. Here, my recent rebuild of one PC has now reduced CPDN units to 3 (i5-3570k without HT) from 7 (i7-950 with HT), GPUGrid using the 4th/8th cores. Even 2 GPUs share only one CPU for GG but other GPU projects seem to need one each (as I have recently confirmed). So I'm crunching 10 cores compared to your 14? Is that what you have worked out? However, the Ivybridge i5 is quicker anyway (e.g. regional models take just over 1 second per time step compared to almost 2.5 seconds with the other rig - an i7-920) so perhaps I'm crunching more like the equivalent of 12+ 1st generation i7 cores. Minimal overclocking here as well. Do not worry. My second rig will also be rebuilt with a second Ivybridge i5-3570k very soon. So you will become the main cruncher again for the team! Will finish remaining work units over next week so back to another waiting period which will be ably filled by WCG again. :) IanSM P.S. Suggest any continuation of this stuff is better done over in our forum now you have discovered it has been down for a few days ! |
|
8)
Message boards :
Cafe CPDN :
Scotland team
(Message 45968)
Posted 18 Apr 2013 by old_user596405 Post: Hi Ian So, what is the state of play with the Scottish Forums? Can't update our stats as the site has been DOWN for several days ! iansm |
|
9)
Message boards :
Cafe CPDN :
Scotland team
(Message 45924)
Posted 13 Apr 2013 by old_user596405 Post: We have had 119 members since 23 Nov when Colin Neill joined. Mind you, 118 does ring a recent bell. ?? Anyway, the stats actually pops-up a message when someone joins or leaves (and pauses the script) so could not miss that. Unless stats app is flawed. Getting a bit creaky in its old age! Any new revamp of the CPDN stats pages ( there was a rumour last year) then I'm afraid our version won't be adjusted. My programmer's brain cells retired a few years ago! IanSM |
|
10)
Message boards :
Cafe CPDN :
Scotland team
(Message 45892)
Posted 11 Apr 2013 by old_user596405 Post: No new member detected ! :) |
|
11)
Message boards :
Cafe CPDN :
Scotland team
(Message 45891)
Posted 11 Apr 2013 by old_user596405 Post: Team Scotland seems to have acquired a new member and, as usual, I can't work out who it is! I'm sure someone will enlighten me - meantime, welcome, whoever you are, thank you for joining the team and happy crunching! ;-) Missed new member myself. Don't check Scotland list often - only to see if I'm ahead of IanB ! Will know next time I do a stats update as it picks up the join date. BTW, a change of plan. I'll continue to run CPDN until I complete 2000 models so it will be a few months esp. after reducing available CPU cores to 6 from 12-14 after upgrading my second machine. Quicker cores though so probably equivalent to 8-9 of the old CPUs. Cheers IanSM |
|
12)
Message boards :
Cafe CPDN :
Scotland team
(Message 45678)
Posted 21 Mar 2013 by old_user596405 Post: Hi Bad news (maybe). Once I've completed 30 models in progress, will be bringing to an end 7 years of running Climate project models. About 1,750 of them! Reason? Computer upgrades. Recently upgraded motherboard and CPU in PC 1. For faster and more energy efficient processing but reduced to a 4 core CPU only. Furthermore, only 2 cores are available for CPU projects as the other pair are dedicated to my main GPU project - GPUGrid. Am in top 30 IN RAC and top 40 in total credits out of over 18,000 users. Indeed, GG has been my no. 1 project for some time. Later will be upgrading PC 2 for the same benefits of faster proceesors and lower power costs. Therefore, will be down from an original maximum of 16 cores (models) in 2 PCs. Or from 12 since it became mandatory to dedicate a CPU core to each GPU (have 4) in GG. The CPU is required to "feed" the very fast GPU processors. Plan to switch the eventual 4 free CPU cores to POEM and WCG only. Both generate very short work units of a few hours instead of days or weeks with CPDN. Decision also partly based on inconsistent availability of CP models as well as what would be a higher productivity (throughput / turnaround time) for two worthy projects. Anyway, 7 years of CP is no bad. :) Cheers, Iansm |
|
13)
Message boards :
Cafe CPDN :
Scotland team
(Message 44821)
Posted 7 Sep 2012 by old_user596405 Post: "SP" wrote: Looks like another new member has joined team Scotland in the last few days ...... Meantime, welcome, whoever you are and we hope to issue a more personal welcome soon! http://climateapps2.oerc.ox.ac.uk/cpdnboinc/show_user.php?userid=684687 Stuck or given up! No trickles yet. iansm |
|
14)
Message boards :
Cafe CPDN :
Scotland team
(Message 43971)
Posted 4 Apr 2012 by old_user596405 Post: We can assume that if BOINCstats states that 23 was highest ever then 22 will be the new record. BOINCstats does not record highest ever for RAC so we can assume that 12 was also a record. So both numbers have reached highest positions during the last week. Interestingly, we now have up to 10 fewer active crunchers (22) than in the early CPDN years. Higher productivity! Incidentally, I see that an original member has recently restarted after over 4 years absence. Was also a BBC cruncher. Maybe more will return ? |
|
15)
Message boards :
Cafe CPDN :
Scotland team
(Message 43362)
Posted 3 Nov 2011 by old_user596405 Post: Team Scotland appears to have acquired two new members in CPDN recently but I can't work out who they are to welcome them - does anyone know? Iansm probably does, if he's around! ;-) Sabine has started trickling away ok. The other is Ice (joined Oct 13) but no tasks downloaded yet. Probably the inconsistency of WU availability in recent weeks but decent stocks of two model types now available - so no excuses! |
|
16)
Message boards :
Number crunching :
Why haven't I recieved credit in a long time?
(Message 42675)
Posted 26 Jul 2011 by old_user596405 Post: Credits will get updated!! A step in the right direction is that trickle records (see individual "trickle information for result ...." pages) have now been updated. Previously stopped on July 10 / 11. Patience is good. You never know, maybe credits will be calculated today. :) I had 15 x HadCM3N work units all suspended with 2 hours to completion. Been waiting to see records getting updated again before now letting them all rush at once for the finishing line. |
|
17)
Message boards :
Cafe CPDN :
Scotland team
(Message 42582)
Posted 5 Jul 2011 by old_user596405 Post: That's what I like to see - it was the climate projects that attracted most of our team into distributed computing, after all. Seconded. :) |
|
18)
Message boards :
Number crunching :
CPDN Credit Not Getting Out?
(Message 42451)
Posted 23 Jun 2011 by old_user596405 Post: Please note however, that unlike in the past, this project seems to be entering a phase where the amount of work at any one time is limited. Tens of thousands of computers, most running this project part time, is not what's needed now. Rather a thousand or so fast, dedicated computers, that can get the results back quickly. Exactly, Les. But this is a serious point to ponder. For over a week (since returning from a short holiday), I have kept the door open in one of my 3 quick and dedicated (24/7) systems - i.e. one of your "thousand". As expected, no new models so will continue to keep the door open. However, when a new batch is released, will I receive any? Or will many of the "tens of thousands" grab them first? Would not be happy if that happened. Missed the last batch of cm3N models when on holiday. My main GPU project is GPUGrid. Their scientists demand fast turnaround so applications are optimised for mid to high end GPU cards. Users of light weight cards are politely but firmly advised to redirect them to other projects. Incidentally, GPUGrid does not "pay" as well as other (Nvidia) GPU projects. This is a good thing so the project attracts users who are committed to the importance of the research. Rather than being credit hunters and gatherers :) How about advising users with low powered / part time systems to wait until CPDN returns to those heady days of multiple model types and copious supply? There are some good alternative projects for more modest systems - e.g. Rosetta, Malaria and the very popular WCG where they are happy to use any hardware for multiple research causes. Even a very slow machine can complete a WU in less than a day. Are there performance stats by CPU so, like some other projects (mainly GPU), lists can then be compiled for "Highly Recommended" / "Recommended" / "Not Recommended" systems? |
|
19)
Message boards :
Number crunching :
HadCM3N - Error Messages on Completion
(Message 42003)
Posted 22 Apr 2011 by old_user596405 Post: 4 of my 5 HadCM3n full resolution ocean models apparently "completed" earlier today - Task Ids 12758267, 12740235, 12740230 and 12740229. Each uploaded the final trickle at time step 1,036,800. The first 3 have status marked as "completed" but the 4th (12740229) is marked as "error while computing". All 4 tasks ran constantly 24/7, none "failed" then restored from backup - the usual reason for a error status which does not get corrected when model eventually finishes. All have identical stderr messages. Link to the host's summary page - http://climateapps2.oucs.ox.ac.uk/cpdnboinc/results.php?hostid=1057969 My assumption is that all 4 are ok so the scientists will get clean results otherwise a waste of resource and time (started 21 days ago)! All those stderr messages are, if irrelevant, confusing to say the least. |
|
20)
Message boards :
Number crunching :
HadCM3 Full Resolution model low credits
(Message 41983)
Posted 13 Apr 2011 by old_user596405 Post: HadCM3n credits have been "upgraded" to 12,441.60 for a completed model (40 year trickles @ 311.04). Previously just 3,249.60 (40 x 81.24). Thank you Jonathan! |
©2024 climateprediction.net