Welcome to MilkyWay@home

BOINC 6.10.6 released

Message boards : Number crunching : BOINC 6.10.6 released
Message board moderation

To post messages, you must log in.

1 · 2 · Next

AuthorMessage
Ironworker16
Avatar

Send message
Joined: 31 Jan 09
Posts: 31
Credit: 69,908,565
RAC: 0
Message 31154 - Posted: 19 Sep 2009, 0:54:45 UTC
Last modified: 19 Sep 2009, 1:10:07 UTC

I see 6.10.6 is posted I'm going to download and install.

Edit: Boinc Download

Installed up and running.
ID: 31154 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile arkayn
Avatar

Send message
Joined: 14 Feb 09
Posts: 999
Credit: 74,932,619
RAC: 0
Message 31155 - Posted: 19 Sep 2009, 1:16:09 UTC

I have it installed on my Dell with a 4550 and it seems to work fine.
ID: 31155 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Zanth
Avatar

Send message
Joined: 18 Feb 09
Posts: 158
Credit: 110,699,054
RAC: 0
Message 31161 - Posted: 19 Sep 2009, 6:06:43 UTC

No rampant work fetches?
ID: 31161 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Raimund Barbeln

Send message
Joined: 7 Oct 07
Posts: 25
Credit: 35,401,003
RAC: 5,711
Message 31165 - Posted: 19 Sep 2009, 7:56:19 UTC - in response to Message 31161.  

Seems to work fine for me.

MW ATI WUs are crunching FIFO again, Seti CUDA WUs too.
And ibercivis just downloaded only 1 WU when the other one finished.

So it looks to me as if 6.10.6 is finally a usable Version again.
When life gives you lemons, make lemonade!
ID: 31165 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Paul D. Buck

Send message
Joined: 12 Apr 08
Posts: 621
Credit: 161,934,067
RAC: 0
Message 31169 - Posted: 19 Sep 2009, 8:57:31 UTC

On other boards the indications are that the WF bug is fixed so you won't get continual downloads of work but that the preemption bug is still there where BOINC changes its mind about which tasks to run on CUDA cards.

Depending on which projects you run this may or may not be a significant issue for you.

UCB is finding issues with the way that they are selecting tasks to run so eventually they may stumble on the main cause and correct it ...

This may be more of an issue with SaH where there is a higher likelihood that a long list of tasks will be obtained at the same moment and that the selected to run tasks may not be the most appropriate ...

In a way I hate to keep pointing out I told them so, but not enough to mention that with today's "wide" systems there is almost no instances where a task is in such deadline trouble that it cannot wait for the next available processing element to finish the task it is working on...

I know that there are people out there with only a single GPU, but, the far more likely case at this moment are those like me that have more than one GPU in most of their systems. Thus back to my point ...in the days of single cores it made a lot of sense to micromanage the order in which tasks were processed. Most of my systems are 10 processing elements wide ... it is like a bank with one teller vs. a bank with 10 ... the line moves faster in the latter ...
ID: 31169 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile The Gas Giant
Avatar

Send message
Joined: 24 Dec 07
Posts: 1947
Credit: 240,884,648
RAC: 0
Message 31177 - Posted: 19 Sep 2009, 11:56:56 UTC
Last modified: 19 Sep 2009, 11:58:24 UTC

After watching it for an hour or so on and off, it appears to still put wu's into waiting mode, but only if it has run out of wu's to run (for example I'm running 6 across my 2 ATI cards, so I mean when it gets down to 5 running with no more in the queue). BOINC then gets another load and immediately it has the 6th to crunch it places that wu into running mode but doesn't progress it. As each subsequent wu is downloaded BOINC places the least progressed wu into waiting mode and starts the next wu, but it doesn't progress. I've seen it do this until only 1 or 2 wu are left actually running while 4 or 5 are trying to run (but aren't and have time accumulating against them) and all others are in waiting state. Suspending MW are restarting kicks things off again until BOINC next has a hard time downloading work.

I can't leave it running overnight - so I'm going back to 6.4.7.
ID: 31177 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile (_KoDAk_)

Send message
Joined: 13 Jul 08
Posts: 33
Credit: 21,285,010
RAC: 0
Message 31210 - Posted: 19 Sep 2009, 20:37:01 UTC

HELP
6106 Requesting new tasks for GPU ONLy GPU
and so all projects
how to solve the request CPU
ID: 31210 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
zombie67 [MM]
Avatar

Send message
Joined: 29 Aug 07
Posts: 115
Credit: 501,600,404
RAC: 4,799
Message 31222 - Posted: 20 Sep 2009, 2:42:20 UTC - in response to Message 31161.  
Last modified: 20 Sep 2009, 2:42:43 UTC

No rampant work fetches?


With 6.10.5, I had 5 of 17 machines with that problem. With 6.10.6, I have one machine with that problem. So it's still not 100% fixed.

ID: 31222 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Paul D. Buck

Send message
Joined: 12 Apr 08
Posts: 621
Credit: 161,934,067
RAC: 0
Message 31337 - Posted: 22 Sep 2009, 8:30:07 UTC

For what it is worth I did try 6.10.6 and got the immediate start and trash tasks problem on the second cycle of MW tasks (sorry about that), the good news is that it only trashed 6 tasks and I hope I had on the flags they are going to ask for (it is always one way to discredit a report ... but you did not log what we wanted to see ...)

Anyway, I posted and maybe ... maybe ...
ID: 31337 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile arkayn
Avatar

Send message
Joined: 14 Feb 09
Posts: 999
Credit: 74,932,619
RAC: 0
Message 31343 - Posted: 22 Sep 2009, 12:35:55 UTC - in response to Message 31337.  

I just took my Quad back to 6.10.3 because it started doing the same thing to MW units, hope they fix the problem soon.
ID: 31343 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Paul D. Buck

Send message
Joined: 12 Apr 08
Posts: 621
Credit: 161,934,067
RAC: 0
Message 31347 - Posted: 22 Sep 2009, 16:45:21 UTC - in response to Message 31343.  

I just took my Quad back to 6.10.3 because it started doing the same thing to MW units, hope they fix the problem soon.

Not likely, to this point they are telling me that I don't know what I am doing so therefor there is no problem.
ID: 31347 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Paul D. Buck

Send message
Joined: 12 Apr 08
Posts: 621
Credit: 161,934,067
RAC: 0
Message 31348 - Posted: 22 Sep 2009, 16:47:51 UTC

Call for information from someone that is running 6.10.6 ...

Can you look at your tasks and tell me if they are being reported as 6.10.6 or 6.10.3 ...

I just looked at Arkayn's list and his error tasks are showing 6.10.3 in the report log ... I would like some more checking to see if there is a version ID issue with 6.10.6 incorrectly reporting as 6.10.3 ...
ID: 31348 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
XB-STX

Send message
Joined: 9 Aug 08
Posts: 18
Credit: 56,863,533
RAC: 0
Message 31349 - Posted: 22 Sep 2009, 16:53:37 UTC - in response to Message 31348.  

6.10.6 does indeed report correctly:

<core_client_version>6.10.6</core_client_version>
ID: 31349 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
BarryAZ

Send message
Joined: 1 Sep 08
Posts: 520
Credit: 302,524,931
RAC: 15
Message 31352 - Posted: 22 Sep 2009, 18:07:02 UTC - in response to Message 31155.  

One area where the 6.10.6 (and all previous 6.10 back to 6.6.36) does NOT seem capable of handling is the distinction between ATI GPU, CUDA GPU and CPU tasks to download.

I don't have any high end graphics cards (ie none with double precision support), so for Milkyway, at the project side, I've configured for no GPU support. The dumb BOINC client insists on looking for GPU tasks, and when, surprise, surprise, it doesn't find them it has succeeded in setting the project clock for its 60 second or 90 second retry time out. OK fine -- but then 60 seconds or 90 seconds later, AGAIN it goes looking for GPU tasks.

You would think that having the BOINC client check for the project configuration at the client site (or at the workstation since it does of course pick up things like resource share settings), and THEN go looking for the correct work units would be something the developers could put in place.

Since I have configurations with single precision GPU's -- which work on GPUGrid and Collatz, I can't simply ignore the evil 6.x client and run 5.4.5, instead I'm left railing at the developers from a distance....


I have it installed on my Dell with a 4550 and it seems to work fine.


ID: 31352 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Paul D. Buck

Send message
Joined: 12 Apr 08
Posts: 621
Credit: 161,934,067
RAC: 0
Message 31353 - Posted: 22 Sep 2009, 19:20:19 UTC - in response to Message 31352.  

One area where the 6.10.6 (and all previous 6.10 back to 6.6.36) does NOT seem capable of handling is the distinction between ATI GPU, CUDA GPU and CPU tasks to download.

Which was the point we tried to make with UCB... sadly JM VII sided with with them so the argument was lost.

My main concern is that server side scalability is never considered and there is that old straw and camel problem ... of course I cannot prove that the extra pings are not helping MW's server and the idiots in charge of the asylum use whatever argument is convient to avoid making changes (or at least that is how it seems to me)...

We cannot do x because that would increase the server hits and the server and connections to the database are such a burden ... we don't need to do y because hits on the server are trivial and exponential back off will blah blah blah ...
ID: 31353 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
BarryAZ

Send message
Joined: 1 Sep 08
Posts: 520
Credit: 302,524,931
RAC: 15
Message 31357 - Posted: 22 Sep 2009, 22:17:10 UTC - in response to Message 31353.  

It isn't that they are averse to changes, rather they are averse to changes they don't have as part of their agenda. It is analogous to 'keep the government out my medicare and social security' rants we hear way too much of these days.



My main concern is that server side scalability is never considered and there is that old straw and camel problem ... of course I cannot prove that the extra pings are not helping MW's server and the idiots in charge of the asylum use whatever argument is convient to avoid making changes (or at least that is how it seems to me)...

We cannot do x because that would increase the server hits and the server and connections to the database are such a burden ... we don't need to do y because hits on the server are trivial and exponential back off will blah blah blah ...


ID: 31357 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Misfit
Avatar

Send message
Joined: 27 Aug 07
Posts: 915
Credit: 1,503,319
RAC: 0
Message 31365 - Posted: 23 Sep 2009, 3:51:12 UTC - in response to Message 31353.  

One area where the 6.10.6 (and all previous 6.10 back to 6.6.36) does NOT seem capable of handling is the distinction between ATI GPU, CUDA GPU and CPU tasks to download.

Which was the point we tried to make with UCB... sadly JM VII sided with with them so the argument was lost.

No surprise. Siding with UCB is one of his specialties.
me@rescam.org
ID: 31365 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Paul D. Buck

Send message
Joined: 12 Apr 08
Posts: 621
Credit: 161,934,067
RAC: 0
Message 31366 - Posted: 23 Sep 2009, 3:58:47 UTC - in response to Message 31365.  

One area where the 6.10.6 (and all previous 6.10 back to 6.6.36) does NOT seem capable of handling is the distinction between ATI GPU, CUDA GPU and CPU tasks to download.

Which was the point we tried to make with UCB... sadly JM VII sided with with them so the argument was lost.

No surprise. Siding with UCB is one of his specialties.

Um, not totally ... though he does not disagree too long or too hard ... which I think is somewhat sad ...

I also think it was a bad day when he stopped being the primary developer on the Resource Scheduler ...
ID: 31366 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Emanuel

Send message
Joined: 18 Nov 07
Posts: 280
Credit: 2,442,757
RAC: 0
Message 31379 - Posted: 23 Sep 2009, 15:51:48 UTC - in response to Message 31177.  

6.10.7 is out for Windows x86 and x64; from the check-in notes:
- fix bug that caused unstarted coproc jobs to preempt ones already running.
- fix preemption bug, this time fer sure!

ID: 31379 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Paul D. Buck

Send message
Joined: 12 Apr 08
Posts: 621
Credit: 161,934,067
RAC: 0
Message 31383 - Posted: 23 Sep 2009, 18:03:05 UTC - in response to Message 31379.  

6.10.7 is out for Windows x86 and x64; from the check-in notes:
- fix bug that caused unstarted coproc jobs to preempt ones already running.
- fix preemption bug, this time fer sure!


The link above is bad, it should be this.

Now I have to go hunt down the release notes ...
ID: 31383 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
1 · 2 · Next

Message boards : Number crunching : BOINC 6.10.6 released

©2024 Astroinformatics Group