Mirror (Feb 26 2009)


log in

Advanced search

Message boards : Technical News : Mirror (Feb 26 2009)

Previous · 1 · 2
Author Message
Profile Bryan Price
Send message
Joined: 3 Apr 99
Posts: 3
Credit: 509,972
RAC: 0
United States
Message 870120 - Posted: 27 Feb 2009, 18:02:57 UTC - in response to Message 869954.

Oh, I didn't see your post, Bryan.


It's only my first post! :-p LOL! It's also my second account, but who's counting... :/

I didn't think to capture the traffic to see if it was already being compressed. Duh!

____________

Profile Bryan Price
Send message
Joined: 3 Apr 99
Posts: 3
Credit: 509,972
RAC: 0
United States
Message 870122 - Posted: 27 Feb 2009, 18:07:55 UTC - in response to Message 870016.

I just received a 1200+ hour Astropulse work unit (computer is a P4 2.65Mhz) isn't that kinda long ??


I got something like a 480 hour AP unit according to BOINC, but I should have it done in 80 hours (8 hours at 10%, 16 hours at 20%). I'm hoping it keeps me busy over the weekend. :)
____________

Ingleside
Volunteer developer
Send message
Joined: 4 Feb 03
Posts: 1546
Credit: 4,198,344
RAC: 18,944
Norway
Message 870422 - Posted: 28 Feb 2009, 14:07:31 UTC - in response to Message 870055.

Another thing that would have greatly helped keep that situation from spiraling out of control is what I believe would be a better way to do the quota system. Instead of doubling the quota for a good result returned, it should just be +2. It was pointed out that if you are on an 8-CPU system and you turn in 800+ bad tasks, it only takes eleven (11) good results to bring the quota back to 800. 4-cpu takes 10, 2-cpu only takes 8. Then there's the CUDA quotas that were thrown into the mix, as well, with the multiply factor for that. I think +2 instead of 2x would keep problem computers at bay very nicely. It doesn't even have to be +2.. it can be +5..just as long as it's not multiplication.

Actually, 7 good results in a row will give you full quota, regardless of single-core or 8-way system. The quota is always between 1 and 100, but is multiplied by #cpu then deciding to send work or not. This also means 99 errors in a row will give computers minimum quota, even if it's a 8-way system.



____________
"I make so many mistakes. But then just think of all the mistakes I don't make, although I might."

Cosmic_Ocean
Avatar
Send message
Joined: 23 Dec 00
Posts: 2248
Credit: 8,601,124
RAC: 4,319
United States
Message 870433 - Posted: 28 Feb 2009, 14:29:15 UTC - in response to Message 870422.

Another thing that would have greatly helped keep that situation from spiraling out of control is what I believe would be a better way to do the quota system. Instead of doubling the quota for a good result returned, it should just be +2. It was pointed out that if you are on an 8-CPU system and you turn in 800+ bad tasks, it only takes eleven (11) good results to bring the quota back to 800. 4-cpu takes 10, 2-cpu only takes 8. Then there's the CUDA quotas that were thrown into the mix, as well, with the multiply factor for that. I think +2 instead of 2x would keep problem computers at bay very nicely. It doesn't even have to be +2.. it can be +5..just as long as it's not multiplication.

Actually, 7 good results in a row will give you full quota, regardless of single-core or 8-way system. The quota is always between 1 and 100, but is multiplied by #cpu then deciding to send work or not. This also means 99 errors in a row will give computers minimum quota, even if it's a 8-way system.



That's right..I forgot the quota was just 1-100 period.. So yeah. I think it should take more than 8 good results to go from 1 to 100.
____________

Linux laptop uptime: 1484d 22h 42m
Ended due to UPS failure, found 14 hours after the fact

John McLeod VII
Volunteer developer
Volunteer tester
Avatar
Send message
Joined: 15 Jul 99
Posts: 24387
Credit: 519,750
RAC: 26
United States
Message 870784 - Posted: 1 Mar 2009, 4:42:57 UTC - in response to Message 870433.

Another thing that would have greatly helped keep that situation from spiraling out of control is what I believe would be a better way to do the quota system. Instead of doubling the quota for a good result returned, it should just be +2. It was pointed out that if you are on an 8-CPU system and you turn in 800+ bad tasks, it only takes eleven (11) good results to bring the quota back to 800. 4-cpu takes 10, 2-cpu only takes 8. Then there's the CUDA quotas that were thrown into the mix, as well, with the multiply factor for that. I think +2 instead of 2x would keep problem computers at bay very nicely. It doesn't even have to be +2.. it can be +5..just as long as it's not multiplication.

Actually, 7 good results in a row will give you full quota, regardless of single-core or 8-way system. The quota is always between 1 and 100, but is multiplied by #cpu then deciding to send work or not. This also means 99 errors in a row will give computers minimum quota, even if it's a 8-way system.



That's right..I forgot the quota was just 1-100 period.. So yeah. I think it should take more than 8 good results to go from 1 to 100.

The following sequence if repeated will also never generate a minimum quota.

49 errors followed by 1 good result.
____________


BOINC WIKI

msattlerProject donor
Volunteer tester
Avatar
Send message
Joined: 9 Jul 00
Posts: 38925
Credit: 579,128,332
RAC: 512,110
United States
Message 870859 - Posted: 1 Mar 2009, 8:19:15 UTC - in response to Message 870784.

Another thing that would have greatly helped keep that situation from spiraling out of control is what I believe would be a better way to do the quota system. Instead of doubling the quota for a good result returned, it should just be +2. It was pointed out that if you are on an 8-CPU system and you turn in 800+ bad tasks, it only takes eleven (11) good results to bring the quota back to 800. 4-cpu takes 10, 2-cpu only takes 8. Then there's the CUDA quotas that were thrown into the mix, as well, with the multiply factor for that. I think +2 instead of 2x would keep problem computers at bay very nicely. It doesn't even have to be +2.. it can be +5..just as long as it's not multiplication.

Actually, 7 good results in a row will give you full quota, regardless of single-core or 8-way system. The quota is always between 1 and 100, but is multiplied by #cpu then deciding to send work or not. This also means 99 errors in a row will give computers minimum quota, even if it's a 8-way system.



That's right..I forgot the quota was just 1-100 period.. So yeah. I think it should take more than 8 good results to go from 1 to 100.

The following sequence if repeated will also never generate a minimum quota.

49 errors followed by 1 good result.

99 bottles of beer on the wall?
____________
*********************************************
Embrace your inner kitty...ya know ya wanna!

I have met a few friends in my life.
Most were cats.

Richard HaselgroveProject donor
Volunteer tester
Send message
Joined: 4 Jul 99
Posts: 8466
Credit: 48,993,349
RAC: 73,894
United Kingdom
Message 870879 - Posted: 1 Mar 2009, 10:01:29 UTC - in response to Message 870784.

The following sequence if repeated will also never generate a minimum quota.

49 errors followed by 1 good result.

And that's close to what we saw last weekend. 49 errored Astropulse results, followed by one good MB result, means the host can continue indefinitely - even if it is trashing every task from one of the applications.

That defeats the purpose of quota: BOINC needs to apply it per application.

msattlerProject donor
Volunteer tester
Avatar
Send message
Joined: 9 Jul 00
Posts: 38925
Credit: 579,128,332
RAC: 512,110
United States
Message 870882 - Posted: 1 Mar 2009, 10:09:03 UTC - in response to Message 870879.

The following sequence if repeated will also never generate a minimum quota.

49 errors followed by 1 good result.

And that's close to what we saw last weekend. 49 errored Astropulse results, followed by one good MB result, means the host can continue indefinitely - even if it is trashing every task from one of the applications.

That defeats the purpose of quota: BOINC needs to apply it per application.

And maybe pull back just a little on the ramp-up of the quota..I am in favor of it recovering quickly.......but even half as much as it is now would be enough to heal a broken host in short order.
____________
*********************************************
Embrace your inner kitty...ya know ya wanna!

I have met a few friends in my life.
Most were cats.

John McLeod VII
Volunteer developer
Volunteer tester
Avatar
Send message
Joined: 15 Jul 99
Posts: 24387
Credit: 519,750
RAC: 26
United States
Message 871230 - Posted: 2 Mar 2009, 5:10:23 UTC - in response to Message 870882.

The following sequence if repeated will also never generate a minimum quota.

49 errors followed by 1 good result.

And that's close to what we saw last weekend. 49 errored Astropulse results, followed by one good MB result, means the host can continue indefinitely - even if it is trashing every task from one of the applications.

That defeats the purpose of quota: BOINC needs to apply it per application.

And maybe pull back just a little on the ramp-up of the quota..I am in favor of it recovering quickly.......but even half as much as it is now would be enough to heal a broken host in short order.

How about +2 rather than *2 for the recovery. (+1 was too slow, and *2 is too fast).
____________


BOINC WIKI

Previous · 1 · 2

Message boards : Technical News : Mirror (Feb 26 2009)

Copyright © 2014 University of California