upgrading to 7.0.56

Message boards : Number crunching : upgrading to 7.0.56
Message board moderation

To post messages, you must log in.

1 · 2 · 3 · Next

AuthorMessage
Profile trader
Volunteer tester

Send message
Joined: 25 Jun 00
Posts: 126
Credit: 4,968,173
RAC: 0
United States
Message 1347861 - Posted: 17 Mar 2013, 20:03:26 UTC

currently running 7.0.28. with opt app v0.40.

when i upgrade to the new app do i need to finish all task and install or do i need to uninstall and then install it. and will i need to reinstall the opt app again?

side question best nv driver version to use?
I RTFM and it was WYSIWYG then i found out it was a PEBKAC error
ID: 1347861 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14649
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1347884 - Posted: 17 Mar 2013, 20:38:58 UTC - in response to Message 1347861.  

It's best to finish all GPU tasks before upgrading.

You don't need to uninstall anything: you won't need to reinstall any optimised apps.

Drivers are a movable feast.
ID: 1347884 · Report as offensive
spitfire_mk_2
Avatar

Send message
Joined: 14 Apr 00
Posts: 563
Credit: 27,306,885
RAC: 0
United States
Message 1347914 - Posted: 17 Mar 2013, 22:19:34 UTC
Last modified: 17 Mar 2013, 22:22:53 UTC

Bad advice is bad.




ID: 1347914 · Report as offensive
Claggy
Volunteer tester

Send message
Joined: 5 Jul 99
Posts: 4654
Credit: 47,537,079
RAC: 4
United Kingdom
Message 1347933 - Posted: 18 Mar 2013, 0:47:58 UTC - in response to Message 1347914.  

Bad advice is bad.


It always says that for non-recommended Boinc versions, But Boinc 7.0.56 is a Release Candidate, and is somewhat more finely polished than Boinc 7.0.28,

Claggy
ID: 1347933 · Report as offensive
Profile Mr. Kevvy Crowdfunding Project Donor*Special Project $250 donor
Volunteer moderator
Volunteer tester
Avatar

Send message
Joined: 15 May 99
Posts: 3776
Credit: 1,114,826,392
RAC: 3,319
Canada
Message 1347937 - Posted: 18 Mar 2013, 1:05:17 UTC - in response to Message 1347885.  
Last modified: 18 Mar 2013, 1:06:04 UTC

Richard, what benefits would 7.0.56 give over 7.0.28?


There's a bug in 7.0.28 where it doesn't identify GPUs that are suitable for OpenCL AstroPulse. I have one box formerly on 7.0.28 that would consistently download an OpenCL unit, then report that there were no available compatible GPUs and abort it, then download it again, ad infinitum. 7.0.56 fixed this, so this is one reason to use it.

Also 7.0.56 seems to have fixed the bug in earlier RCs where it would abort CUDA units after a few seconds indicating that they were taking too long! So far, it seems good.
ID: 1347937 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14649
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1347939 - Posted: 18 Mar 2013, 1:07:57 UTC - in response to Message 1347914.  

On the other hand, some of us read that as USE FOR TESTING, and do our best to get the bugs out before the rest of you see it.

Sorry Chris - but a quick scan down BOINC 7 Change Log and news got me to well over 300 changes between 18 May 2012 (when v7.0.28 was released) and now. I didn't feel like writing a precis over the weekend, but I've tested pretty much every one of them (and reported a good few bugs, old and new, along the way). My judgement is that v7.0.56 is pretty close to ready, and I'd choose that one over the next couple: I actually persuaded David to tackle a bug report of mine from five years ago ([trac]#588[/trac]), so that'll be a couple more iterations before we get the bugs out of the fixes.

I see a lot of bad advice on these message boards, but in this case, I don't think the charge is justified.
ID: 1347939 · Report as offensive
Profile trader
Volunteer tester

Send message
Joined: 25 Jun 00
Posts: 126
Credit: 4,968,173
RAC: 0
United States
Message 1347972 - Posted: 18 Mar 2013, 4:41:25 UTC - in response to Message 1347939.  

richard

the card i am installing is a gtx680, what i am currently is onboard intel hd4000 video. with this information IYO do you think upgrading to the new version will be worth it?
I RTFM and it was WYSIWYG then i found out it was a PEBKAC error
ID: 1347972 · Report as offensive
Luther

Send message
Joined: 11 Mar 13
Posts: 12
Credit: 715,795
RAC: 0
United States
Message 1348018 - Posted: 18 Mar 2013, 8:45:44 UTC - in response to Message 1347972.  

You need to install .56 for sli setups, the .28 wouldn't use more then one gpu for me.
ID: 1348018 · Report as offensive
Bob Giel
Volunteer tester

Send message
Joined: 11 Jan 04
Posts: 76
Credit: 5,419,128
RAC: 0
United States
Message 1348022 - Posted: 18 Mar 2013, 9:20:51 UTC

I upgraded to 7.0.56 from 7.0.26 and all GPU tasks have errored out with this message:

Name 02dc12ab.12106.27515.206158430213.10.77_1
Workunit 1189873574
Created 16 Mar 2013, 0:34:19 UTC
Sent 16 Mar 2013, 8:04:03 UTC
Received 18 Mar 2013, 3:53:58 UTC
Server state Over
Outcome Computation error
Client state Compute error
Exit status 197 (0xc5) EXIT_TIME_LIMIT_EXCEEDED
Computer ID 6212459
Report deadline 9 May 2013, 4:52:59 UTC
Run time 5,705.43
CPU time 154.07
Validate state Invalid
Credit 0.00
Application version SETI@home Enhanced
Anonymous platform (NVIDIA GPU)

Have reverted back to 7.0.26 and no longer see errors on GPU tasks.

ID: 1348022 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14649
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1348024 - Posted: 18 Mar 2013, 9:52:07 UTC

Oh dear. What a lot of scrabbling for answers.

@ Bob,
As has been posted several times over the last 48 hours, there is a problem with the step-transition between .28 and any later version. Unless you want to be locked into v7.0.28 for evermore, you're going to have to get over it sometime. The errors are just for work which was issued under the old version, but crunched under the new version. You can avoid them either by running down your cache first, or by resetting the project after upgrading.

@ Luther,
You've found one of the many improvements. You could run SLI (two identical GPUs) under the old version, but sometimes you had to force BOINC to use them both with a configuration file. You'll still have to do that with unmatched cards, but less often.

@ trader,
I see you've written nearly 40 posts (and I know at least one PM as well) since you re-started crunching after a four year break. I'd slow down a bit. As you're finding, the new BOINC and GPU crunching is very different to what you remember. Do a little more reading, and a little less writing, before you decide whether to upgrade BOINC. And start by reading the reply to Bob above.
ID: 1348024 · 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 1348027 - Posted: 18 Mar 2013, 10:07:31 UTC

I was going to reply but I see Richard has already done so.

To reiterate the recommended approach when upgrading from a earlier (as in 7.0.28) version is to finish off and report your GPU work first. Personally I usually run the machine dry before upgrading but that's just me.

I am running .56 on one machine and have .52 on a bunch of others. One of the advantages of the current version is support for the Intel HD graphics built into the newer i5 and i7 chips.
BOINC blog
ID: 1348027 · Report as offensive
Profile trader
Volunteer tester

Send message
Joined: 25 Jun 00
Posts: 126
Credit: 4,968,173
RAC: 0
United States
Message 1348029 - Posted: 18 Mar 2013, 10:12:21 UTC - in response to Message 1348024.  

Oh dear. What a lot of scrabbling for answers.

@ trader,
I see you've written nearly 40 posts (and I know at least one PM as well) since you re-started crunching after a four year break. I'd slow down a bit. As you're finding, the new BOINC and GPU crunching is very different to what you remember. Do a little more reading, and a little less writing, before you decide whether to upgrade BOINC. And start by reading the reply to Bob above.



WOW A WHOLE 40 POSTS.... OOPS 41
I RTFM and it was WYSIWYG then i found out it was a PEBKAC error
ID: 1348029 · Report as offensive
Profile TRuEQ & TuVaLu
Volunteer tester
Avatar

Send message
Joined: 4 Oct 99
Posts: 505
Credit: 69,523,653
RAC: 10
Sweden
Message 1348705 - Posted: 20 Mar 2013, 13:57:58 UTC

I recomend .56 over .28

ID: 1348705 · Report as offensive
Profile Fred E.
Volunteer tester

Send message
Joined: 22 Jul 99
Posts: 768
Credit: 24,140,697
RAC: 0
United States
Message 1348709 - Posted: 20 Mar 2013, 14:09:06 UTC

Several advantages have been mentioned for .56 vs. .28 - here's another. I'm running it because you can run multiple instances on the gpu without using anonymous platform. That helps on projects where anonymous platform is discouraged (like Beta, WCG, Milkyway), and would help those that are unable to install the lunatics apps here on main. This version is stable.
Another Fred
Support SETI@home when you search the Web with GoodSearch or shop online with GoodShop.
ID: 1348709 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14649
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1348711 - Posted: 20 Mar 2013, 14:10:25 UTC

And now there's a .57 to test. Installed OK, too early to say if there are any problems yet.
ID: 1348711 · Report as offensive
Profile Jord
Volunteer tester
Avatar

Send message
Joined: 9 Jun 99
Posts: 15184
Credit: 4,362,181
RAC: 3
Netherlands
Message 1348739 - Posted: 20 Mar 2013, 16:01:18 UTC - in response to Message 1348711.  

Well, it won't fix the crash my client had the other night, as that one is as much a mystery as how the connection to the forums continued with the network being disabled. ;-)
ID: 1348739 · Report as offensive
TBar
Volunteer tester

Send message
Joined: 22 May 99
Posts: 5204
Credit: 840,779,836
RAC: 2,768
United States
Message 1348780 - Posted: 20 Mar 2013, 18:09:13 UTC - in response to Message 1348739.  

Do you know if 7.0.57 has this fix in it?
I don't see it listed yet, BOINC 7 Change Log and news
I find it interesting the first item missing from the Error reports is the Time...
ID: 1348780 · Report as offensive
Profile Jord
Volunteer tester
Avatar

Send message
Joined: 9 Jun 99
Posts: 15184
Credit: 4,362,181
RAC: 3
Netherlands
Message 1348792 - Posted: 20 Mar 2013, 18:33:37 UTC - in response to Message 1348780.  
Last modified: 20 Mar 2013, 18:40:28 UTC

I'm waiting for the Macintosh version before I put up the change log.
Rom just emailed that he'll release a new version tonight. So it may all be .58

The fix you see there, is for applications, not for the client. The BOINC API is used in compiling applications, not the client. See http://boinc.berkeley.edu/trac/wiki/SourceCodeGit#Sourcecoderoadmap and for the actual fix https://github.com/matszpk/native-boinc-for-android/commit/8c8ccabee7b6b3f34c2bae0f2d72a49f2d8107da
ID: 1348792 · Report as offensive
TBar
Volunteer tester

Send message
Joined: 22 May 99
Posts: 5204
Credit: 840,779,836
RAC: 2,768
United States
Message 1348799 - Posted: 20 Mar 2013, 18:50:12 UTC - in response to Message 1348792.  

So, it appears you are saying the applications will have to be recompiled to incorporate this fix?
ID: 1348799 · Report as offensive
Claggy
Volunteer tester

Send message
Joined: 5 Jul 99
Posts: 4654
Credit: 47,537,079
RAC: 4
United Kingdom
Message 1348840 - Posted: 20 Mar 2013, 19:41:10 UTC - in response to Message 1348780.  
Last modified: 20 Mar 2013, 19:46:27 UTC

Do you know if 7.0.57 has this fix in it?
I don't see it listed yet, BOINC 7 Change Log and news
I find it interesting the first item missing from the Error reports is the Time...

You're comparing Apples to Pairs, the NativeBoinc tree has changes applied to Boinc 6.10.58, Boinc 6.12.38 and 7.0.36 making them new versions of those client versions,

While on the official version tree once a change has been applied, the new Boinc version is given a new version number, Here Boinc 6.10.58 is always Boinc 6.10.58, Boinc 6.12.38 is always Boinc 6.12.38 and Boinc 7.0.36 is always Boinc 7.0.36

On the Offical tree there already has been changes to the client & api at Boinc 7.0.38:

client and API: improve the way an app checks for the death of the client.

Old: heartbeat mechanism
Problem: if the client is blocked for > 30 secs (e.g. because it takes a long time to write the state file, or because it's stopped in a debugger) then apps exit. This is bad if the app doesn't checkpoint and has been running for a long time.

New: the client passes its PID to the app.
The app periodically (10 sec) checks that the process still exists.

Notes:
For backward compatibility (e.g. new API w/ old client, or vice versa) the client still sends heartbeats, and the API checks heartbeats if the client doesn't pass a PID.
The new mechanism works only if the client's PID isn't assigned to a new process within 10 secs of the client exiting. Windows 2000 reuses PIDs immediately, so check for Win2K and don't use this mechanism if so.

TODO: For Unix multithread apps, critical sections aren't currently being enforced. Need to fix this by masking signals.


and it's own fix:

•API: fix Unix bug when checking if client is alive based on PID.


Can't use waitpid() here; works only for children.
Use kill(pid, 0) instead.


The next question is, Is NativeBoinc 6.10.58/6.12.38/7.0.36 up to date with these changes? the question after that is: Why isn't the author of NativeBoinc particapating in the boinc_dev list and getting his changes applied to the Official Boinc?
Note: these NativeBoinc changes could end up being Android only, and be of no concern to Desktop users.

Claggy
ID: 1348840 · Report as offensive
1 · 2 · 3 · Next

Message boards : Number crunching : upgrading to 7.0.56


 
©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.