I received a task that appears to need >3000h of my CPU. I will not be able to finish this in the time given. Such a task blocks my computer without getting a result. Think such large tasks should only be given to very powerful computers and users with an appropriate active time.

Thanks for conidering this input.
thomas

Hi Toko,

The deadline is ignored, and the project will accept the result uploads whenever they occur.

Secondly, the data is uploaded every model year (i.e., roughly every 18 hours or so). So your computer will be producing results and sending them to the servers almost immediately.

The researchers do not expect every model to get to 160 years, simply doing as much as you can, or want to, is sufficient.

-Cheers,

Mike

---

I'm a volunteer and my views are my own.
News and Announcements and FAQ

Toko, if you disable the globe screensaver the model will process faster. You\'ll still be able to look at your globe when you want to by using the View graphics button in your boinc manager.

---

Cpdn news

... The deadline is ignored, and the project will accept the result uploads whenever they occur.
...

The only possible issue that may occurr is you crunch for other BOINC projects as well. BOINC will figure out at some point in time that you will miss the deadline; it will then allocate more time fro CPDN in an attempt to make it to the deadline. This may mean that BOINC will not download work from other projects for a long time.

---

Hi Toko,

The deadline is ignored, and the project will accept the result uploads whenever they occur.

Secondly, the data is uploaded every model year (i.e., roughly every 18 hours or so). So your computer will be producing results and sending them to the servers almost immediately.

The researchers do not expect every model to get to 160 years, simply doing as much as you can, or want to, is sufficient.

-Cheers,

Mike

Thanks Mike - very much comforting my processor (cooled down few degrees ;))
Screensaver never turned on.

Toko, if you disable the globe screensaver the model will process faster. You\'ll still be able to look at your globe when you want to by using the View graphics button in your boinc manager.

Tnx mo.v
thomas

... The deadline is ignored, and the project will accept the result uploads whenever they occur.
...

The only possible issue that may occurr is you crunch for other BOINC projects as well. BOINC will figure out at some point in time that you will miss the deadline; it will then allocate more time fro CPDN in an attempt to make it to the deadline. This may mean that BOINC will not download work from other projects for a long time.

Tnx, nice to know
thomas

The current CPDN workunits are rather large lol. Even my Core 2 Duo looks like it will take around 1900-2k hours to finish a WU. But its got a 50% share so i guess i should finish it fairly easily.
Another thing to think on is the fact there is only one size WU available so i don\'t suggest CPDN for older machines. I wouldn\'t try this on my 2Ghz Celery for example (supposedly its fast enough but i doubt it). But maybe a late p4, core 2 or AMD dual core i would.
Whats sort of machine is this on?
Also estimated time is just that an estimate it could be wildly wrong. CPDN Wu was listed as 2850hrs on my Core 2 Duo E6400 yet so far im processing at least 1/3 faster than the estimate, so wait and see if the estimate is accurate.

---

Rather than relying on the boinc estimates which can be wildly inaccurate particularly with a new model, it\'s probably easier just to notice how long your computer takes between trickles (= to complete a model year) and multiply this by 160.

For example, a computer only running cpdn and trickling once a day (about 3.2 seconds/timestep running 24/7) will take about 5 months. Then factor in any holidays when the computer will be turned off, a bit of down time for maintaining the computer, and an extra week or two in case you crash the model and need to restore your backup.

The minimum recommended CPU is 1.6GHz but slightly slower computers can run the current model as long as they don\'t share time with other boinc projects.

---

Cpdn news

I noticed that the 5.15 application does not report time on the client in t/s anymore. It\'s MFlOps/GFlOps.

I kind of liked t/s, but can live with the new way.

---

I\'m still seeing s/ts, I assume this is on the show graphics screen after you\'ve pressed \'Z\' to hide the BBC sidebar?

---

I'm a volunteer and my views are my own.
News and Announcements and FAQ

LOL i cant even get the graphics to display. Oh well im saving a few cycles for the app anyway. I found my t/s times in the trickles section. Although alarmingly its much longer when running 2 models side by side and slightly longer with every trickle.

---

I\'m guessing you have a hyperthreaded CPU rather than a dual core CPU. There is a long-term advantage to running two, but it\'s less (overall throughput is increased by ~15% rather than ~90% as it would be with a dual core).

---

I'm a volunteer and my views are my own.
News and Announcements and FAQ

Sadly this is a true Dual Core (C2D)
If apps run side by side i go from 1.9123 sec/ts to 2.1045 sec/ts which is a significant increase (but mainly of the second WU not the first).
Also by the 9th trickle my first WU has increased to 1.9715 sec/ts. Im worried how slow this is gonna get by the end. Maybe i need more RAM but i thought 1Gb would be enough for a Linux Machine

---

1.9 to 2.1 s/ts isn\'t too bad if it\'s accurate (10%), but one thing to remember is that the displayed s/ts speed is averaged over the entire lifetime of the model. The more accurate way is to look at the CPU time reported against the trickles on the website, work out how many seconds the latest one took, and divide that figure by 25920 (number of timesteps per trickle).

I suspect the reason the times have gone up might be saturation of the memory bus. Is it one of the ones with a smaller cache?

---

I'm a volunteer and my views are my own.
News and Announcements and FAQ

Its only an E6400 lol i didn\'t splash out on the extreme. Maybe for my next cruncher...i mean desktop yes i want a faster desktop. thats my story and im sticking to it.

---

im finding this annoying, its gonna take me 40 hours atm to do 1 year (or 1 feed back,) 5.65 s/ts

do also do other stuff on it however still gonna take sosay 3000 hours to compute, thats

125 days of non stop computing,

if at 2 hours a day
thats just over 4 years!!!

or at 4 hours a day
thats just over 2 years!!!

tempted to do other mainly projects like prime grid (22 mins to compleat) or seti (1hour 20 mins)

well got my frends pc working on this (been working for like 5 months still hasent noticed lol)

my own pc may b another 1.

just for u to know, my pc specs are below

amd 2000+
512 mb ram
10 gb virtual memory
120 gb hdd

im finding this annoying, its gonna take me 40 hours atm to do 1 year (or 1 feed back,) 5.65 s/ts

do also do other stuff on it however still gonna take sosay 3000 hours to compute, thats

125 days of non stop computing,

if at 2 hours a day
thats just over 4 years!!!

or at 4 hours a day
thats just over 2 years!!!

tempted to do other mainly projects like prime grid (22 mins to compleat) or seti (1hour 20 mins)

well got my frends pc working on this (been working for like 5 months still hasent noticed lol)

my own pc may b another 1.

just for u to know, my pc specs are below

amd 2000+
512 mb ram
10 gb virtual memory
120 gb hdd

Yes this is certainly not a project for short Wu\'s.
There are probably many projects better suited to you if you like your Wu\'s short.
Beware that 1hr 20mins is a short Seti WU my slower rigs often run over 10hrs on a seti long WU so it might not be as short as you expect. A good project if you want to have a set time limit for Wu\'s is rosetta@home. In your Rosetta preferences you can set your own Wu length.
the machine you have quoted is just marginal to run a CPDN WU if you check the min specs they ask for.

Also out of interest why 10gb virtual memory? Thats an absolute waste of hard disk space. If your using swap your going to slow down massively as is. If your using much over a gig of swap you might as well not have RAM your going to be that slow. generally its not recommended to tell windoze what to do with swap file size. It almost manages it competently on its own. If your on Linux you only really need a swap twice the size of your RAM and even then you are unlikely to use any at all.
If you wish to look into another project try the Team BOINC@AUSTRALIA Website there is a list of active projects down the left hand side of the page with an explanation of what they are trying to do.
Also im unsure if im reading this right but we dont install BOINC on other peoples (even friends) machines without their permission. It could be considered theft. But then again i may have just misunderstood and i hope i have.

---

well got my frends pc working on this (been working for like 5 months still hasent noticed lol)

ROFLMAO ^^ I should try that on my brother. His nice new Intel laptop would do sooooo well on MalariaControl and he just doesn\'t have the motivation to sign up... Too bad this is against my \"Admin Ethics\" , else I might really give it a try... Gotta make do with my own Celeron (unfortuately that has a benchmark handicap running Linux and they still give benchmark credit over there) and a fast AMD that seems to have the wrong architecture... or maybe try to persuade my brother after all ;-)

On the Linux box try installing BOINC 5.8.1 it does wonders for your benchmarks
On my C2D it up things considerably. It almost benchmarks as good as a windoze machine.

---