| Author |
Message |
|---|
|
|
|
i was just wondering if anyone is getting anything like me when my computer downloads the tasks for woodall primes they are showing 1444 hours for everyone, ive aborted them to get new ones etc and its always the same exact times, just wondering what i can do to get this to work and not have insane hours to completion so i can actually do the tasks, my second computer i run is doing them fine showing avg 50 hours or so so im just wondering what i can do to make it better, andy help would be great, thanks
____________
|
|
|
|
|
|
Hello. Just because it says 1444 hrs dose not mean thats how long it will take. Your DCF (duration correction factor) could be off?? Try and crunch some and see if the time to complete drops, also if you have a slow rig you will get times like that, and again that dose not mean thats how long it will take. Do you have other cpu projects going on? Did you just start to crunch CW prime? Also they are about 30/45 hrs on a fast rig, longer on older rigs.
Edit..yes I see you just joined not too long ago.
____________
My first Prime
20208428036625*2^666666-1 (SGS)
PROUD MEMBER OF TEAM CARL SAGAN |
|
|
|
|
|
Boinc is notoriously bad at estimating WU completion times for primegrid as the WU's have vastly different lengths. If you see stupid numbers like 1440 hours just ignore it and start crunching the WU - the estimate given by Boinc has nothing to do with how long it will actually take. I downloaded Woodall LLR's yesterday with an estimate of 240 hours. I've only been going 28 hours and i'm more than halfway through all of them ;-)
____________
 |
|
|
|
|
|
thanks everyone, ill just let that computer crunch on em for awhile and see how it goes lol, just not used to ever seeing that high of a number on a project before |
|
|
|
|
thanks everyone, ill just let that computer crunch on em for awhile and see how it goes lol, just not used to ever seeing that high of a number on a project before
I expect the times to drop especially if you have never run Woodall's before on that host. As a point of reference, my old Opteron 240 took 466,141.92 seconds for a Woodall. But your i7 CPU Q 740 @ 1.73GHz host with hyperthreading enabled will be worse. The good news is that you get two WU per core done but it will take close to twice the time as one WU.
Just a warning: Watch out for the WU deadline (not challenge) as it is very short for slow computers even running 24 hrs a day only on Woodalls. |
|
|
|
|
|
One other note.
You also are crunching PPS Sieve under CUDA..
Boinc calculates their runtime if they were running CPU, but when they are run under GPU and take significantly less time, the DCF gets re-adjusted.
Unfortunately, the DCF is on a per project basis.
So, after running WUs on an "extremely fast machine" and then running on a different type of WU, it doesn't know what to do...
I see it all the time when switching between GPU and CPU here.
____________
|
|
|
 rroonnaalldd Volunteer developer
Volunteer tester
Send message
Joined: 3 Jul 09
Posts: 1213
ID: 42893
Credit: 34,634,263
RAC: 0
                 
|
One other note.
You also are crunching PPS Sieve under CUDA..
Boinc calculates their runtime if they were running CPU, but when they are run under GPU and take significantly less time, the DCF gets re-adjusted.
Unfortunately, the DCF is on a per project basis.
So, after running WUs on an "extremely fast machine" and then running on a different type of WU, it doesn't know what to do...
I see it all the time when switching between GPU and CPU here.
I think all this depends of the version of the used boinc-client...
Newer version since 6.12.X should have different DCF-paths for CPU and GPU but only the CPU-side will be displayed in the Boinc-manager and the DCF for the CPU-side should be adjusted on per app basis.
____________
Best wishes. Knowledge is power. by jjwhalen
 |
|
|
|
|
|
A Pentium M 1.73GHz now reached 56% after 126 hours and claims still 136 hours to go (in reality it should be only about 100 hours). For sure tomorrows deadline can't be reached.
Is it possible to restrict those slow computers from server side to avoid them downloading the long running wus (Woodall, Cullen, PSP, SoB)?
I'll let finish my wu for the challenge, but probably the server will sent out an additional copy before my wu gets returned.
____________
 |
|
|
|
|
A Pentium M 1.73GHz now reached 56% after 126 hours and claims still 136 hours to go (in reality it should be only about 100 hours). For sure tomorrows deadline can't be reached.
Is it possible to restrict those slow computers from server side to avoid them downloading the long running wus (Woodall, Cullen, PSP, SoB)?
I'll let finish my wu for the challenge, but probably the server will sent out an additional copy before my wu gets returned.
I have found that with less than 11 and a half hours to go the system downloads another WU.
If you don't want any more then set this project on this computer to No New Work and when the WU finishes manually upload it.
Conan
____________
 |
|
|
|
|
A Pentium M 1.73GHz now reached 56% after 126 hours and claims still 136 hours to go (in reality it should be only about 100 hours). For sure tomorrows deadline can't be reached.
Is it possible to restrict those slow computers from server side to avoid them downloading the long running wus (Woodall, Cullen, PSP, SoB)?
I'll let finish my wu for the challenge, but probably the server will sent out an additional copy before my wu gets returned.
No, the server does not sense what to send to where. What you can do though, is create multiple profiles for preferences under "your account -> PrimeGrid preferences" and then add the slow computer(s) to one profile and the quick ones to the other. You than can say per profile which subprojects you want them to run.
____________
PrimeGrid Challenge Overall standings --- Last update: From Pi to Paddy (2016)
 |
|
|
|
|
|
Finished the wu after 226h45min and got 3033 credits (2d16h overdue).
IMHO the server knows the requested calculation time or cpu speed (FLOPs), otherwise the server would always send wu until the host limit has been reached. Thus if the wu has a deadline of 7 days (604,000s) the server should have an interest not sending this wu to a host which has an expected runtime much hiher (more than 816,000s in this case).
I disabled now PG on this host.
____________
|
|
|
|
|
Finished the wu after 226h45min and got 3033 credits (2d16h overdue).
IMHO the server knows the requested calculation time or cpu speed (FLOPs), otherwise the server would always send wu until the host limit has been reached. Thus if the wu has a deadline of 7 days (604,000s) the server should have an interest not sending this wu to a host which has an expected runtime much hiher (more than 816,000s in this case).
I disabled now PG on this host.
The server uses the setting for your contact frequency (time) and how many days work you request to hold.
e.g. If (as someone said) it downloads when there are 11 hours to go you have o.5 days work minimum set.
It takes time to set your preferences to what you need, some times you still want to adjust them for different work types.
____________
Member team AUSTRALIA
My lucky number is 9291*2^1085585+1 |
|
|
|
|
|
Hello, new to Primegrid and BOINC in general. I was brought here by Numberphile for Woodall Primes, only to find that my little Lenovo laptop isn't quite up to it. It estimates 13 days of calculation for each task. Is this considered an acceptable time? Or should I focus on other projects for the sake of my computer's health? |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 14011
ID: 53948
Credit: 433,223,095
RAC: 925,100
                              
|
Hello, new to Primegrid and BOINC in general. I was brought here by Numberphile...
Welcome to PrimeGrid!
only to find that my little Lenovo laptop isn't quite up to it. It estimates 13 days of calculation for each task. Is this considered an acceptable time? Or should I focus on other projects for the sake of my computer's health?
Two very different questions there.
The deadline for Woodall tasks is 14 days, but if your computer is actually working on the task it will periodically report its progress to our server, and the server will extend the deadline if needed, up to 56 days, as long as your computer continues to make progress. So if you crunch 24/7 and finish in 11 days, that's fine.
As to whether it's good for your computer, that's more complicated.
Paradoxically, longer LLR tasks are actually easier on the computer than short tasks. The reason is that shorter tasks fit better into the CPU's cache than do longer (and therefore larger) tasks. Since cache is much faster than main memory, the CPU is able to run short tasks more quickly, which means they run hotter on the short tasks. Running 11 days worth of 10 minute tasks will draw more power, produce more heat, and be more stressful to your computer than running a single 11 day task.
____________
My lucky number is 75898524288+1 |
|
|
|
|
|
I am also new to PrimeGrid and here from Numberphile. I was just wondering if it is possible to write software to use GPUs to find Woodall numbers? Wouldn't that make it much faster?
If this is already an option and I have missed it, please let me know. I can't find a way to use a GPU to find Woodall numbers in the PrimeGrid Preferences. |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 14011
ID: 53948
Credit: 433,223,095
RAC: 925,100
|
I am also new to PrimeGrid and here from Numberphile. I was just wondering if it is possible to write software to use GPUs to find Woodall numbers? Wouldn't that make it much faster?
If this is already an option and I have missed it, please let me know. I can't find a way to use a GPU to find Woodall numbers in the PrimeGrid Preferences.
GPUs operate differently than CPUs, with different strengths and different weaknesses. Not all problems are good candidates for solving on GPUs. Testing Woodall numbers for primality is one of those that appears to not be very amenable to running on a GPU.
____________
My lucky number is 75898524288+1 |
|
|
