| Author |
Message |
|---|
|
|
|
Hello,
I have a machine with 2x580 dedicated to PrimeGrid only and other ATI rigs dedicated to Milkyway@Home as main with PrimeGrid as a backup. On web preferences, the cache is set to 0.50 days for all projects so that my ATI rigs aren't filled with too much Prime wus when Milkyway is down. BUT in order to pump primegrid wus to 2x580 machine, the cache is set to 4 days only in that machine's manager. They all work based on preferences.
However, i am unable to get more than 2 wus in my Nvidia machine. Am i doing something wrong here?
So basically i want the nvidia one to have many primewus, but i want my ati rigs to have less primegrid wus. Afaik, what i do is the right solution. But somehow it doesn't work.
Thank you for your guidance. |
|
|
|
|
|
Hello again,
Strange thing, when i restart the computer (eventually BOINC), it downloaded 64 wus. But still, would someone give me a confirmation that i am doing the right thing? Or correct me?
Thanks. |
|
|
|
|
|
Just wondering what you BOINC client reports as the time to completion for an unstarted work unit for Primegrid?
Peter
____________
My Top 5000 Primes:
110059!+1 is prime! (FPS)
14493618614235*2^666666-1 is prime! (SGS) |
|
|
|
|
Just wondering what you BOINC client reports as the time to completion for an unstarted work unit for Primegrid?
Peter
It's 8:35 (mins/secs) but i don't see its connection with the problem. Since the cache is only 2 wus again, each time a wu is finished there is an idle interval between requesting, downloading and starting.
|
|
|
|
|
|
Very unusual... And your sure the cache is set correctly in BOINC Options > Preferences?
Just on a side note, you can set the client to report completed tasks immediately by placing:
<cc_config>
<options>
<report_results_immediately>1</report_results_immediately>
</options>
</cc_config>
In cc_config.xml in your boinc data folder
Peter
____________
My Top 5000 Primes:
110059!+1 is prime! (FPS)
14493618614235*2^666666-1 is prime! (SGS) |
|
|
|
|
Very unusual... And your sure the cache is set correctly in BOINC Options > Preferences?
Just on a side note, you can set the client to report completed tasks immediately by placing:
<cc_config>
<options>
<report_results_immediately>1</report_results_immediately>
</options>
</cc_config>
In cc_config.xml in your boinc data folder
Peter
Yep, it is very unusual. I have a config file like this:
<cc_config>
<log_flags>
<task>0</task>
<file_xfer>0</file_xfer>
</log_flags>
<options>
<dont_contact_ref_site>1</dont_contact_ref_site>
<max_file_xfers_per_project>4</max_file_xfers_per_project>
<report_results_immediately>1</report_results_immediately>
</options>
</cc_config>
But i don't think it's related to cc_config. I'm using 6.12.26 version but still the same problem. |
|
|
|
|
|
It's probably due to Duration Correction Factor (DCF). I think the issue was i oc the card after i initiated the BOINC. Therefore the wu estimation is lowered and interfered with DCF.
So, your theory on wu estimation is actually correct:) Thank you. Let me try it now. |
|
|
 rroonnaalldd Volunteer developer
Volunteer tester
Send message
Joined: 3 Jul 09
Posts: 1213
ID: 42893
Credit: 34,634,263
RAC: 0
                 
|
<dont_contact_ref_site>1</dont_contact_ref_site>
<max_file_xfers_per_project>4</max_file_xfers_per_project>
What is the purpose of this settings?
____________
Best wishes. Knowledge is power. by jjwhalen
 |
|
|
|
|
<dont_contact_ref_site>1</dont_contact_ref_site>
<max_file_xfers_per_project>4</max_file_xfers_per_project>
What is the purpose of this settings?
Check this site for purpose:
http://boinc.berkeley.edu/wiki/Client_configuration |
|
|
|
|
|
You might want to try re-running the BOINC CPU benchmarks from Advanced > Run CPU Benchmark to see if that will correct the issue. Although if it is GPU related, then it should fix itself in time!
____________
My Top 5000 Primes:
110059!+1 is prime! (FPS)
14493618614235*2^666666-1 is prime! (SGS) |
|
|
|
|
<dont_contact_ref_site>1</dont_contact_ref_site>
<max_file_xfers_per_project>4</max_file_xfers_per_project>
What is the purpose of this settings?
Check this site for purpose:
http://boinc.berkeley.edu/wiki/Client_configuration
remove this line
max_file_xfers_per_project>4</max_file_xfers_per_project
just for testing and try again. I think that work enough time so duration factor is not a problem. |
|
|
|
|
|
Well currently, my caches are filled up but i increased "maintain additional work for" to 10 days and my each computer approximately has 150 wus in progress. I wish them to have more because these rigs are working for 7/24. |
|
|
rroonnaallddVolunteer developer
Volunteer tester
Send message
Joined: 3 Jul 09
Posts: 1213
ID: 42893
Credit: 34,634,263
RAC: 0
|
|
But all units have to be reissued to the queue if your units wouldn't validated...
Take look in the thread Side bets on wingmen?. Some SOB-units have no valid results after a half year.
From project-side is a wu-limit a must have. The biggest problem for the Seti-project was their number of open wu's and as a result an increased size of the project-db.
____________
Best wishes. Knowledge is power. by jjwhalen
|
|
|
mfbabb2Volunteer tester
Send message
Joined: 10 Oct 08
Posts: 510
ID: 30360
Credit: 11,549,356
RAC: 0
             
|
Well currently, my caches are filled up but i increased "maintain additional work for" to 10 days and my each computer approximately has 150 wus in progress. I wish them to have more because these rigs are working for 7/24.
If you are constantly connected to the internet, it should not matter -- you will refill as soon as the WUs are processed.
____________
Murphy (AtP)
 |
|
|
|
|
Well currently, my caches are filled up but i increased "maintain additional work for" to 10 days and my each computer approximately has 150 wus in progress. I wish them to have more because these rigs are working for 7/24.
If you are constantly connected to the internet, it should not matter -- you will refill as soon as the WUs are processed.
But my internet connection is via wireless usb and windows 7 cuts power to USB if the internet connection is idle more than an hour. Therefore, to prevent idle machine, i fill the queue as much as possible. |
|
|
|
|
. But my internet connection is via wireless usb and windows 7 cuts power to USB if the internet connection is idle more than an hour. Therefore, to prevent idle machine, i fill the queue as much as possible.
This sounds like Power Management Settings, and its probably the "USB Selective Suspend Setting". Its possible its also (or only) the Wireless Adaptor Setting - need to play with both as it depends on how you are setup and how you use the beast.
To get to it:
- Right Click Desktop, select "Personalize"
- Bottom right click on the words "screen saver"
- Click "Change Power Setting"
You will then see two Power Plans (Balanced and Power Saver), for whichever one has the radio circle checked, click "Change Plan Settings"
- Click "Change Advanced Power Settings"
In the resulting box, you will see two of the selections titled: "Wireless Adaptor Settings" and "USB Settings".
Look inside those two choices and select whatever is appropriate for you, the latter will depend on how you use the beast and the end result you need. Its likely both settings are set for active management of USB and Wireless, suspending their operation if it detects no activity after a period. It shouldnt make a difference to you as it (they) should spring to life on demand, however something appears to be either stopping that or is not reacting fast enough and BOINC "thinks" you have no link.
Some experimentation needed to discover which it is (or both). I cant be absolutely certain its the root cause of your issue, but it looks like it. If it is, you can then reduce the cache size you are holding - win-win for both yourself and your Projects.
Regards
Zy |
|
|
