BBC CCE - What year are you on?

[Evil-Dragon]

I'm on 1927 at the moment of writing, doing about 1 year per day.

[Strongbow]

1940 on my old Intel

and 1950 and 1972 on my Opteron

[Dee]

I'm on 1935 :)

[spikey_richie]

Done just over a year (15months) in 4 hours on my AMD 3400+

Done the same amount in 6 hours on my P4 3ghz.

[Frederic Brillouet]

I'm at the end of '32. Done 214 hours over this, is that normal?

[NeoGen]

How long does it take a timestep for you?

214 hours for only 12 years seems rather low really...

[Frederic Brillouet]

Something like 2,5. I read that i needed to defragment my hdd, and since i did that, the project has speeded up a little bit.

[NeoGen]

Most people like to watch the graphics/screensaver, but (assuming that you haven't) if you disable that you'll get a noticeable speed increase.

[Frederic Brillouet]

the graphic thing is always turned off. Since I don't know how to oc, my speed won't go up very fast, or am i wrong? By the way, what cpu-coolers would you guys advise? I have a big Zalman trying to cool my cpu. I've bought a can of compressed air to remove the dust that was between the cooling fins. The processor is only 2 degrees cooler than it was before. Should i change this crappy thing for an arctic freezer 64 pro?

[Empty_5oul]

which zalmen version is it? - if you can post a link to the same one from either their own site or a retailer.

what is the actual temparature? and what are you using to determine this?

[Frederic Brillouet]

CNPS7000B-AlCu is my cooler. I have an msi mainboard equipped with a core center chip. There's also a core center utility that allows you to oc your cpu a little bit. This utility measures system and cpu temperatures. This utility provides me with information on my cpu temperature. I hope this clears the issue a little bit.

[Lagu]

Hi Frederic

My experience of compressed air is that it isn´t enough. I have bough a little bottle containing Isoprophylalcohol of my pharmacist. I got this tip from Vaughan and it is perfect.
Using a cotton pin you can clean every blade on your fan from dust and smell of cooking if it have hit your fan. It dryed fast and make no damage of the blade.

Lagu ;)

[dAVE]

I hope this is the right place to post this. My recent credit has stopped rising. I am still going along at just over 2sec/TS and doing a year every 14 hours approx. 3 days ago I got up to about 6800 points with a recent average credit of 317.52 then things started to drop. The BOINC manager says that I am down to 310.2 and I am in the team as 302.19. My computer number is 144423. But my years still keep crunching on as normal. I am up to 1956.
I thought that I had just reached my maximum Recent Average Credits but I see that people who have longer secs/TS are still rising. I have probably clicked something I shouldn't have. Any ideas? Dave. :cry:

[Empty_5oul]

i don't think you clicked something you shouldn't. There is very little to change within BOINC itself

if you have viewed the progression recently you can hit "ctrl,alt and del" to check it isnt still open. It will be using a fair percentage of your CPU if it is running in the background somewhere.

have you installed or unistalled any software recently, or added/removed any hardware since noticing the change?

[dAVE]

No new programs and the two BOINC processes are each taking a steady 47% to 50% each as normal. The trickles are still going in at 1 or 2 a day and my user total graph is still going up in a roughly straight line. It is just that my user average has peaked and is now going down. Dave. :?

[Empty_5oul]

monitor it over the next couple of days, there is a possibility that you WU were slightly slower than usual hence the reuced RAC. If you did something processor intensive or rebooted the machine this would be the reason.

See what happens in the next few trickles. hopefully it was a one off!

[NeoGen]

The Recent Average Credit should never stop growing as long as you keep crunching 24/7.

But if you notice performance (s/TS) decreasing maybe you should stop it a bit and do some optimizing. A defragmentation is known to speed up things a little.
Also, if you use cpu intensive applications or games for long periods on the computer where the experiment is running, it will get less cpu time, even tough it is running.

[dAVE]

I'm off for 4 days so I'll leave it running as it seems to be going through the years OK and see if it has straighted itself out on my return. Dave.

[spikey_richie]

Does the credit get allocated once you finish a unit? Or is it calculated say once a week based on the current progress?

[Empty_5oul]

it is known as a "trickle". The WU is massive taking many hours. So the project knows you are still processing the unit you send a trickle to it every so often.
For each trickle you receive credit - the time for these varies on the processor.
I think you are granted what you claim as you claim it - it works differently to other BOINC projects without the WU validaiton.

[Strongbow]

I have just checked the progress of my leading core and I'm 50.86% through it and cruching the year 2002! ...just 78 more years left then! :? :roll: :cry:

Just that for 52 days solid crunching!!! I'm starting to feel sorry for my workstation, after it's finished I may even give it a nice cold boot as a present from me! :cool:

[NeoGen]

52 days... :shock:

You'd be 2nd ranked on our team if you were registered on the Uptime project :)
http://en.uptime-project.net/page.php?page=toplist&content=league&lid=8550

[Strongbow]

Ooo! Best join then!

Does it have any overheads?

[NeoGen]

It's just a little app that stays on the system tray, takes about 5 megs of RAM, and every half hour (or other defined time period) it polls the system for the uptime and sends it over to the site. Doesn't do absolutely nothing else besides that.

But one thing to be aware of is that if you want to put several machines on it, you have to create a separate account for each. Using the same reg data on several machines at the same time will get your values all screwed up.
That's why you see people with two and three accounts there but with different names.

[Strongbow]

I've just installed it and it says I've been up for 21days! :?

Ah ha! That's when I did a reboot after playing with windoze! Bugger!

[NeoGen]

When you started it, you chose "method 2" or "method 3" for the uptime polling, I hope?
I don't know the technical differences, but I know that method 1 has a limit of 49 days only.

[Strongbow]

Yep I chose method 2!

[NeoGen]

And to put the thread back on topic again...
DMMc seems to be the only one capable of matching Bozo on RAC... but unfortunately he started late, so has a 13,000 points handicap. :P

[Strongbow]

I'm just hanging on in there until my machine turns up in a week or two :hiding: , then hopefully I can decommison this old laptop and stomp ahead again!

[Frederic Brillouet]

Done just over a year (15months) in 4 hours on my AMD 3400+

Done the same amount in 6 hours on my P4 3ghz.

15 months in 4 hours????? How did you oc your computer to create such a beast?

[Strongbow]

He must be talking about a different project, possibly the standard climate change project, as going by his specification the BBC CCE would take him approximately a little over 16hours to do a full 12months if he gave it 100% time on his core.

[spikey_richie]

How do I join the uptime team?

[Strongbow]

Here you go...
http://www.amdusers.com/modules.php?name=Forums&file=viewtopic&t=2281

[spikey_richie]

Thanks for the link, but still clueless :oops:

[NeoGen]

After registering an account on the site, on your profile page you got menus on the left for joining a league (team) or showing which one you belong to currently.

If you happen to have already registered, but not be logged in, click the register link again and supply your username/password to login.

[spikey_richie]

Sorted now NeoGen, thanks to blackheath :D but thank you too :cool:

[spikey_richie]

I'm on the 4th of June, 1924 - Just had a total eclipse (the world went white!)

http://spikey-richie.co.uk/gallery/albums/userpics/10001/eclipse.jpg

[Frederic Brillouet]

Nice :D didn't know we were supposed to get a result like that :D

[spikey_richie]

It has cleared now, maybe it was just a rendering issue...

[dAVE]

Hi, I'm back now, Cornwall in the spring is great and "She who must be obeyed" is in the pink. Whilst I was away the RACs were 332.1, 338.2, 344.5, 338.2 & 327.8. Since I have been back and suspending BIONIC, for various tasks that I really bought this computer for, the RACs have been 351.1, 356.8, and 339.3. My s/TS has remained constant so the number crunching on the project seems to be OK. It looks like there must be some sort of delay built in to the stats. Thanks for your help, Dave.

[drezha]

Cornwall in the spring is great


Off topic but yeah it does.

Though I'd hate to argue the point Devon is better...

Guess where I live? ;) :lol:

[dAVE]

Where do you go for team help on the BBC project?
By the way Somerset has the best cider!
Should we have a West Country Forum?
Dave.[/quote]

[NeoGen]

Where do you go for team help on the BBC project?
Right here. :)
I've been at the Q&A section of the project for a couple of months and answered over 100 questions during that time. Are you having any problem with your model or boinc?

[dAVE]

NeoGen. Thanks to your advice and from Empty-5oul earlier in the thread I seem to be running OK. I was just confused by the way that the RACs have a delay built in so that if you suspend one day it doesn't show up for a couple of days.
By the way does anyone else's model report on the 1st of December each model year. I kind of assumed that the trickle would go in on the 1st of January.
PS, I am up to 1969 now.
Thanks for the help, Dave.

[dAVE]

Oh dear, I spoke to soon. My model has just finished with a "computation error" It was sending a trickle and then gave 3 reports of
" handle_trickle_down failed: null pointer"
It then deferred the scheduler request for 1 minute before resheduling the CPU: application exited and "Computation for task hadcm3l_1bjf_00062065_1 finished"
I have a back up, do I go back to that or was my model no good and so I should start a new one? :-( Dave.

[NeoGen]

Always try the backup to see if it wasn't an unfortunate event. If it continues, cool! :) (But if it crashes again in the same way, then you do have to get a new model...)

Speaking of which, there is one bug you can't go around yet... if you reach year 2013 it will crash :-(
http://bbc.cpdn.org/forum_thread.php?id=3156

[Steve Lux]

Well, that's the end of the world on the Mayan calendar anyhow, so who will need the data?

[NeoGen]

Surely not the Mayans :lol:

[Frederic Brillouet]

haha, lol :lol: let's see if mine will do the same when i get there. first i want to gain some points at qmc

[NeoGen]

I'm afraid your current model will not live beyond next week... :-( check the other thread I posted.

[Frederic Brillouet]

my model is at 10.8 percent at the moment. why wouldn't it last longer than next week?

[NeoGen]

Unless your model is already one of the new ones, this is what will happen...
http://www.amdusers.com/modules.php?name=Forums&file=viewtopic&t=2419#29337

[Frederic Brillouet]

well isn't that great. How am i going to explain to my dad that he'll have to endure the continuous humming of my computer for another 4 months or so? Not to mention the additional electricity usage!

[dAVE]

Oh dear NeoGen I need your help yet again. I realise that mayhem has broken out and that you are busy but In cannot get onto the BBC site as all the servers are overloaded and, anyway, you guys seem to be ahead of the game. You remember I said that my BBC BOINC was sending trickles on the 1st of December each year. Well, when it killed my model, the new one started on 01/12/1920. Thinking it was corrupt I took the opportunity to uninstall it and download the main Berkley BOINC. I reattached to the project and all my preferences were loaded too easily. It then downloaded two new models and both started on 01/12/1920, which is clearly wrong. Any ideas? Dave :?

[Empty_5oul]

it is probably worth you selecting "reset project". This will delete what you have an start over. You shouldnt lose much work as your old one was wrong (from the BBC) and now you have two new ones :?

anyway. click reset and you should end up with just 1 model to predict.

[dAVE]

New model loaded and it started on 01/12/1920 too. It is crunching OK but I'm worried that I will get to the end only to find that the date error kills me! I guess I'll just have to wait until the mayhem dies down on the BBC site and start a new thread there in the hope that one of the programmers can sort it out. Thanks for all the help guys even if I do keep dropping out and not helping the team. Dave. :cry: Why isn't there an emoticon for the old and wrinkly?

[NeoGen]

Yep, if it is something wrong with boinc or the project, resetting, or even detaching and reattaching, will fix it.

On the other hand I don't know how different these new workunits might be, so a different starting time wouldn't be so hard to believe... :roll:

[dAVE]

What date did your new model start on? Dave.

[NeoGen]

I'm actually not running it... :oops:

Anyone reading this that is crunching on BBC, could you check on that please? :)

[Strongbow]

Each new model starts on 1st December 1920 so you're fine.

The BBC CCE models were all reset today so when a client running the old model reports back to the server (roughly at 3rd December on each year's model) then it will get an exit code 99 and will be forced to automatically download the fixed model in which it will start all over again at 1st Dec 1920.

[dAVE]

Thanks to both of you again. One day I won't be a newby and no longer a pain in the b*ckside. Dave.

[Empty_5oul]

we are pleased you post asking for help. It is better to get it right than be running an old/incorrect model.

[Evil-Dragon]

After enduring 37 days of CPU time i've finally started hitting near the year of my birth. I'm now up to 1980 (37.44%) and nearly starting 1981.

I'll finish this damn model if it kills me (it says i've got 44 days worth of CPU time left to do on this model as well :o )

[Steve Lux]

How do you tell what year you are on? Is it part of the screen saver? I have simply been crunching on one core - not really paying attention to the progress.

[Evil-Dragon]

It's on the screensaver/graphics. On 1982 now.

[Steve Lux]

Thanks for the info.. I'm on year 1945 at the moment.

[Evil-Dragon]

On 1995 now (46.4%)

My god these models take a long time. I'm figuring that the year 2000 is the 50% mark, if thats right then i've still got 80 years worth of modelling to do. :Eyecrazy:

EDIT: 18th Oct: 50% hit (49 days, 17 hrs, 39 mins) (December, Year 2000)

[Evil-Dragon]

Update: 100 years of modelling done (1920-2020)

62% done, 61 days 11 hrs CPU time done. Only another 30 days worth of CPU time to do and 60 years of modelling years left. :happy6:

My graph: http://bbc.cpdn.org/graph.php?resultid=562549 (the earth is screwed!)

[Evil-Dragon]

I'm finally coming to the end of my model:

89.2% done, 87 days 20 hrs CPU time done. 10 days CPU time remaining.

Check the temperature on this sucker: http://bbc.cpdn.org/graph.php?resultid=562549

[PcManiac]

Hey Guys, this might be a little off-topic, but if you type BBC CCE in google, AMDusers has the top links! :D

[drezha]

When I tried downloading the orignial CPDN WU, it downloaded a BBC one...

[NeoGen]

Hey Guys, this might be a little off-topic, but if you type BBC CCE in google, AMDusers has the top links! :D
I saw that happening for Segment Pi too sometime ago. But it didn't last long unfortunately.

[PcManiac]

I just started with this project, so maybe I can help get to 18th :)

[Evil-Dragon]

PcManiac: Be warned the models take approx 100 days (3 months) to compute on a AMD X2 4200+ (2.2Ghz)

Unfortunately they need to be so big because they are predicting from 1920 to 2080 with temperature, pressure, rain/snow, carbon dioxide levels, etc.. etc..

It's been fun but also scary to watch my model get hotter as time progresses. A scary future for us all. Something needs to be done before it gets too late.

[Evil-Dragon]

I'm on my last day of modelling BBC CCE (13hrs remaining)

It's took just over 98 days to complete one model and will be the last BBC CCE WU i will do on my main machine. My secondary machine just started a CDPN@Home WU though, so that will keep it busy for the next 3 months. I might even start a CDPN WU on my main machine after the BBC one is done, i really don't know yet. I think it deserves a rest!

[NeoGen]

You gotta be really persistant to hang on to a workunit for 98 days!!
:lol:


Congrats Evil-Dragon!:icon_thumright:

[Evil-Dragon]

I've had my fair few crashes (bluescreens, lockups, etc) as well where i thought everything was lost!! Luckily i never lost anything (phew)

It's finished now and sending back over 100mb in result data files. If anyone has the patience then CDPN needs some help with our rank there.

[NVM]

is this project not nearing its end? hmmm, going to go do some research.

[NeoGen]

BBC I don't know, but Seasonal Attribution is practically over. They are just receiving their last workunits to wrap up and analyze the results. No new workunits are being issued for some weeks already.