Updating for upcoming apps.....

Message boards : Number crunching : Updating for upcoming apps.....
Message board moderation

To post messages, you must log in.

Previous · 1 · 2 · 3 · 4 · 5 · 6 . . . 10 · Next

AuthorMessage
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14650
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1205615 - Posted: 14 Mar 2012, 0:47:13 UTC - in response to Message 1205563.  

I got one. Look at my shiny new toy.

Hey, I got my own shiny new toy. I can die happy.

Seriously, the flood of queries seems to have died down, so I'll clock off and get some sleep soon. Just a small lemsip first, I think, and then to bed.
ID: 1205615 · Report as offensive
Josef W. Segur
Volunteer developer
Volunteer tester

Send message
Joined: 30 Oct 99
Posts: 4504
Credit: 1,414,761
RAC: 0
United States
Message 1205639 - Posted: 14 Mar 2012, 2:08:19 UTC - in response to Message 1205557.  

NEWS before it happens:

Email from Eric:

The big issue is if they [optimised app users] return their results before any from the official clients have been returned, it's hard to tell what will happen. I'd recommend that people using optimized apps turn off astropulse v6 processing for a couple weeks. I'll post a news item to that effect.

The reason being, that they aren't certain until it happens whether opt app users will get any credit. There was a bug at Beta which experienced testers will know all too well - zero credit. It cleared itself without intervention after a couple of months (slow turnover at Beta), but nobody quite knows why.

You have been warned: crunch stock, or (possibly) crunch for zero credit.


Well, there aren't any opt V6 released, so what else than stock could be run?

You, for instance, could copy the Beta stock application and add appropriate entries in your app_info.xml so the servers would know you had an Astropulse v6 application. But it would still be an anonymous platform application as far as the servers were concerned, and subject to the server bug if it is active here. The real issue with the bug is not that YOU get zero credit if you make the choice to try that way, it's that you may be forcing zero credit on all your wingmates.

=====================

The list of applications in the preferences will henceforth be the same list of apps shown on the Applications page. But an app record in the database which has no actual applications is just a placeholder, so the SETI@home v7 choice won't do anything for now. Having all listed gives those who have a strong preference the ability to set their preferences before getting work they didn't want. In fact, anyone who is using settings which prohibit some work must opt in if they want the new work. The default (all applications) mode will of course include the new work.

I very much doubt that SETI@home v7 will be released here soon. There are some changes in the source code which have not yet been built and tested at SETI Beta. But I'll cross my fingers and hope I'm right, otherwise we'll have a real difficult task trying to reproduce a certain stock buglet in the optimized code.
                                                                  Joe
ID: 1205639 · Report as offensive
Cosmic_Ocean
Avatar

Send message
Joined: 23 Dec 00
Posts: 3027
Credit: 13,516,867
RAC: 13
United States
Message 1205644 - Posted: 14 Mar 2012, 2:31:43 UTC - in response to Message 1205615.  

I got one. Look at my shiny new toy.

Hey, I got my own shiny new toy. I can die happy.

Yay, I got one, too. Only took about 15 work requests. All the files needed were downloaded, as well (as expected, of course). Initial estimate is/was ~125 hours, but the ETA is dropping by ~5 seconds per second, so a rough early estimate is ~25 hours for stock on my system, when r409 was doing about 11 hours.
Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)
ID: 1205644 · Report as offensive
Profile Wiggo
Avatar

Send message
Joined: 24 Jan 00
Posts: 34744
Credit: 261,360,520
RAC: 489
Australia
Message 1205656 - Posted: 14 Mar 2012, 3:39:19 UTC - in response to Message 1205598.  

Eventual problem with AP 505 validation, now that we have 6.01 in the game. Just had one AP 505 validated, and the credit became appalingly low. Never seen this low credit for an AP before. Might this be some problems with the new validator?

wuid=937432877

I've copped that a few times now when partnered with ATi/AMD video cards, I feel your pain. :(

Cheers.
ID: 1205656 · Report as offensive
kittyman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Jul 00
Posts: 51468
Credit: 1,018,363,574
RAC: 1,004
United States
Message 1205682 - Posted: 14 Mar 2012, 6:04:19 UTC - in response to Message 1205502.  

Nice but if they where smart I guess they should have removed the AP v5 app.

There will still be reissues running about for some time. I suppose that still gives folks the option of accepting them, or if they wish to receive the new apps only.

Is there really any left? Didn't know that.

Prolly not, John....LOL.
My bad, I was thinking 505, not 5....
V5 probably could go, if there were any still lurking, they could be sent out to folks accepting 'other' than selected tasks.

My, but my thread got busy whilst I was away at work.....
No wonder, what with AP splitting back up again.
"Freedom is just Chaos, with better lighting." Alan Dean Foster

ID: 1205682 · Report as offensive
kittyman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Jul 00
Posts: 51468
Credit: 1,018,363,574
RAC: 1,004
United States
Message 1205691 - Posted: 14 Mar 2012, 6:37:16 UTC
Last modified: 14 Mar 2012, 6:38:38 UTC

Sooooo, and I don't prefer AP, but run it if sent when nothing else is available....
Am I to understand that all AP work split from this point is V6?
If so, I should not be getting any more AP, other than perhaps a few reissues, as I am opti, and have no V6 app.

Is the download size of a V6 WU the same as a 505? And, has, as one poster alluded to, the crunch time been increased to the point that AP V6 will be a boon rather than a bane to bandwidth usage? (Crunch time VS download size).

The kitties wanna know....LOL.
"Freedom is just Chaos, with better lighting." Alan Dean Foster

ID: 1205691 · Report as offensive
Dave Stegner
Volunteer tester
Avatar

Send message
Joined: 20 Oct 04
Posts: 540
Credit: 65,583,328
RAC: 27
United States
Message 1205693 - Posted: 14 Mar 2012, 6:51:02 UTC

Mark,

The one's I have from Beta are 8,196K, same as 5.05.


Dave

ID: 1205693 · Report as offensive
kittyman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Jul 00
Posts: 51468
Credit: 1,018,363,574
RAC: 1,004
United States
Message 1205694 - Posted: 14 Mar 2012, 7:04:53 UTC - in response to Message 1205693.  
Last modified: 14 Mar 2012, 7:05:33 UTC

Mark,

The one's I have from Beta are 8,196K, same as 5.05.


OK, any difference in the crunch times noted yet?
"Freedom is just Chaos, with better lighting." Alan Dean Foster

ID: 1205694 · Report as offensive
Dave Stegner
Volunteer tester
Avatar

Send message
Joined: 20 Oct 04
Posts: 540
Credit: 65,583,328
RAC: 27
United States
Message 1205698 - Posted: 14 Mar 2012, 7:19:33 UTC


My machines do CPU with opti apps in about 17-18 hours.

Only stock is available yet, they take 48 hours or so.

Don't think I can answer your question based on that.

Not sure what the change from 5.05 to 6.0 is.

I have read in other threads that 6.01 is not really diff from 6.0, only that is is passing info to validator that can identify early exit.

Quote from Josef in Beta:
/quote
To add some detail, the changes from version 6.00 to version 6.01 of Astropulse v6 do not affect the science processing. What is being tested now is the Validator ability to tell BOINC when the run time of a task should not be included in the averages. That depends on 6.01 reporting the fraction_done as well as the fraction_blanked which earlier versions already reported.

The fraction_done is specifically needed when a task finishes early because there have been 30 repetitive AND 30 single pulses found. Those cases are rare and unpredictable.
Joe
quote/




Dave

ID: 1205698 · Report as offensive
kittyman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Jul 00
Posts: 51468
Credit: 1,018,363,574
RAC: 1,004
United States
Message 1205699 - Posted: 14 Mar 2012, 7:27:04 UTC - in response to Message 1205698.  

OK...if true, then V6 will not be any better than 505 in terms of bandwidth usage.

However, the change you are noting is, as I understand it, the last step before limits can be raised since the Boinc code boondoggle that made them necessary.

I hope that is a positive sign, and the limits can be raised or lifted once the proper functioning of the new AP validator and app have been verified.
"Freedom is just Chaos, with better lighting." Alan Dean Foster

ID: 1205699 · Report as offensive
Dave Stegner
Volunteer tester
Avatar

Send message
Joined: 20 Oct 04
Posts: 540
Credit: 65,583,328
RAC: 27
United States
Message 1205700 - Posted: 14 Mar 2012, 7:29:07 UTC

That would be nice.


Dave

ID: 1205700 · Report as offensive
kittyman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Jul 00
Posts: 51468
Credit: 1,018,363,574
RAC: 1,004
United States
Message 1205701 - Posted: 14 Mar 2012, 7:31:17 UTC - in response to Message 1205700.  
Last modified: 14 Mar 2012, 7:32:03 UTC

That would be nice.


Would sure make the kitties sleep more comfortably....LOL.
And you know how much kitties luv their sleepy time.
"Freedom is just Chaos, with better lighting." Alan Dean Foster

ID: 1205701 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14650
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1205718 - Posted: 14 Mar 2012, 9:36:21 UTC - in response to Message 1205699.  

OK...if true, then V6 will not be any better than 505 in terms of bandwidth usage.

However, the change you are noting is, as I understand it, the last step before limits can be raised since the Boinc code boondoggle that made them necessary.

I hope that is a positive sign, and the limits can be raised or lifted once the proper functioning of the new AP validator and app have been verified.

Yes, from what I understand of the changes, the normal download size and runtime of AP_v6 is still the same. However, they have been working on the splitters to try and reduce the number of '100% blanked' and '30/30 exit' tasks that get sent out. Overall, that should reduce the bandwidth usage - particularly, in those bursts when a tape with the stuck data bit gets split and all tasks are uncrunchable.

And yes, the hope is very much that this will eventually allow the project to lift the limits. It is, however, the first (but most difficult) step rather than the last. The next stage - once this one has been tested, of course - will see runtime estimates and DCF return to more stable values: and then finally, there will be no need for the limits any more - we hope.
ID: 1205718 · Report as offensive
Profile cliff
Avatar

Send message
Joined: 16 Dec 07
Posts: 625
Credit: 3,590,440
RAC: 0
United Kingdom
Message 1205726 - Posted: 14 Mar 2012, 10:14:29 UTC - in response to Message 1205597.  

Hi Richard,
OK, as long as I know that in fact there is no v7 as such right now and can ignore it:-)

Regards,
Cliff,
Been there, Done that, Still no damm T shirt!
ID: 1205726 · Report as offensive
MarkJ Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 17 Feb 08
Posts: 1139
Credit: 80,854,192
RAC: 5
Australia
Message 1205733 - Posted: 14 Mar 2012, 10:54:39 UTC
Last modified: 14 Mar 2012, 10:57:38 UTC

Re: app versions

Multi-beam v7
As I understand it there is no optimized CPU multi-beam app at the moment, but there was one being tested I thought by Raistmer. The current x41g cuda multi-beam is considered "v7 ready". Not sure about OpenCL multi-beam app.

Astropulse v6
I gather there is only the stock app and there isn't an optimized v6 app

Could one of the more knowledgeable people (Richard, Joe, Jason or Raistmer) correct me if I am wrong please.
BOINC blog
ID: 1205733 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14650
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1205735 - Posted: 14 Mar 2012, 11:26:27 UTC - in response to Message 1205733.  

Re: app versions

Multi-beam v7
As I understand it there is no optimized CPU multi-beam app at the moment, but there was one being tested I thought by Raistmer. The current x41g cuda multi-beam is considered "v7 ready". Not sure about OpenCL multi-beam app.

Astropulse v6
I gather there is only the stock app and there isn't an optimized v6 app

Could one of the more knowledgeable people (Richard, Joe, Jason or Raistmer) correct me if I am wrong please.

As it happens, development and testing work on Multi-beam v7 is proceeding in synch on both the stock and optimised versions of the application. But the changes are bigger in this app, and it's nowhere near ready for prime-time yet. When the stock v7 is released here (it hasn't been yet), optimised apps will follow a few days later, as usual.

Much the same applies to Astropulse v6, except the changes are smaller, and development work on this overtook v7. The stock apps are out: optimised versions were developed in parallel at the same time, and are in final testing.

This is the first major application release at SETI since CreditNew was introduced two years ago. We've seen odd behaviour at Beta, with zero credit being awarded to applications running under anonymous platform. For that reason, we're going to be very cautious about releasing the optimised applications, following Eric's advice on the front page.

There will be a few mine-canaries running optimised apps here, to probe whether the credit bug is actually as serious as we think it is. If you meet a mine canary and get bitten by it, with zero credit for a stock task, please accept our apologies in advance - but I hope you understand why we're doing it.
ID: 1205735 · Report as offensive
LadyL
Volunteer tester
Avatar

Send message
Joined: 14 Sep 11
Posts: 1679
Credit: 5,230,097
RAC: 0
Message 1205736 - Posted: 14 Mar 2012, 11:33:43 UTC - in response to Message 1205733.  
Last modified: 14 Mar 2012, 11:38:48 UTC

Re: app versions

Multi-beam v7
As I understand it there is no optimized CPU multi-beam app at the moment, but there was one being tested I thought by Raistmer. The current x41g cuda multi-beam is considered "v7 ready". Not sure about OpenCL multi-beam app.

Astropulse v6
I gather there is only the stock app and there isn't an optimized v6 app

Could one of the more knowledgeable people (Richard, Joe, Jason or Raistmer) correct me if I am wrong please.


optimised MB7 apps are ready to go (more or less, we're still testing) CPU, CUDA (x41g or x41u), ATI OpenCL (r455).
edit: that's presuming Eric doesn't do more changes, in which case we obviously have to update as well.
Until Kepler is out and somebody here got their hands on one, we cannot be certain if x41g/u works on them.

optimised AP v6 are in beta testing on Beta. [or will be when I can get round to put up the AVX CPU app]
An updated installer is in preparation, but will not be released until we have established that credit will be granted.
Depending on the credit picture we may prerelease the apps, should we still need more beta testing, prior to installer release.
Until then only very cautious Alpha will occur on main, we may be prepared to get no credit, but we are not going to let our wingman suffer any more than necessary to establish whether or not credit granting is working.

Richard beat me to it.
I'm not the Pope. I don't speak Ex Cathedra!
ID: 1205736 · Report as offensive
Claggy
Volunteer tester

Send message
Joined: 5 Jul 99
Posts: 4654
Credit: 47,537,079
RAC: 4
United Kingdom
Message 1205743 - Posted: 14 Mar 2012, 12:15:29 UTC - in response to Message 1205694.  

Mark,

The one's I have from Beta are 8,196K, same as 5.05.


OK, any difference in the crunch times noted yet?

From my Benches of Stock AP v6 6.01, it is a little bit faster than the Stock v505 5.05 app, but not much,

Claggy
ID: 1205743 · Report as offensive
MarkJ Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 17 Feb 08
Posts: 1139
Credit: 80,854,192
RAC: 5
Australia
Message 1205745 - Posted: 14 Mar 2012, 12:41:02 UTC

Thanks for the replies both Richard and LadyL.

While I am not concerned with credit I am sure some users are. I wanted to get an idea what apps I need to update which is why I asked about the versions. Any idea on timeframes for the app roll out?
BOINC blog
ID: 1205745 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14650
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1205752 - Posted: 14 Mar 2012, 13:18:23 UTC - in response to Message 1205745.  

Thanks for the replies both Richard and LadyL.

While I am not concerned with credit I am sure some users are. I wanted to get an idea what apps I need to update which is why I asked about the versions. Any idea on timeframes for the app roll out?

Eric is suggesting a two week evaluation period, just to be sure. I hope it might be quicker than that, but we won't know until the mine canaries report back.
ID: 1205752 · Report as offensive
Previous · 1 · 2 · 3 · 4 · 5 · 6 . . . 10 · Next

Message boards : Number crunching : Updating for upcoming apps.....


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.