Invalid Host Messaging

Message boards : Number crunching : Invalid Host Messaging
Message board moderation

To post messages, you must log in.

Previous · 1 . . . 6 · 7 · 8 · 9 · 10 · 11 · 12 . . . 22 · Next

AuthorMessage
rob smith Special Project $250 donor
Volunteer tester

Send message
Joined: 7 Mar 03
Posts: 16460
Credit: 339,211,489
RAC: 218,577
United Kingdom
Message 1654661 - Posted: 19 Mar 2015, 21:23:46 UTC

Another volunteer for the Hall of Infamy:
http://setiathome.berkeley.edu/show_user.php?userid=9995210

One of his crunchers has a problem with its GPU and is returning "invalid" results by the bucket load.
OM has been sent.
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?
ID: 1654661 · Report as offensive     Reply Quote
Werecow
Avatar

Send message
Joined: 13 Mar 05
Posts: 56
Credit: 3,557,819
RAC: 3,103
United States
Message 1660710 - Posted: 2 Apr 2015, 11:58:33 UTC

http://setiathome.berkeley.edu/results.php?hostid=6789552

Hundreds of results invalid/inconclusive, but 20 validating against hosts with the same problem.

PM sent 02 Apr 2015.
ID: 1660710 · Report as offensive     Reply Quote
rob smith Special Project $250 donor
Volunteer tester

Send message
Joined: 7 Mar 03
Posts: 16460
Credit: 339,211,489
RAC: 218,577
United Kingdom
Message 1666020 - Posted: 16 Apr 2015, 17:30:08 UTC

I wish to congratulate the anonymous owner of computer 6072583 on his/her record attempt...
State:
All (6221)
In progress (200)
Validation pending (924)
Validation inconclusive (1194)
Valid (425)
Invalid (3478)
Error (0)

Application:
All (6221)
AstroPulse v6 (3)
AstroPulse v7 (8)
SETI@home Enhanced (0)
SETI@home v7 (6210)


That is about a third of your returns are invalid, and still you stomp on without a care in the world. Why don't you turn your random generator off and save your fuel bill?
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?
ID: 1666020 · Report as offensive     Reply Quote
rob smith Special Project $250 donor
Volunteer tester

Send message
Joined: 7 Mar 03
Posts: 16460
Credit: 339,211,489
RAC: 218,577
United Kingdom
Message 1666021 - Posted: 16 Apr 2015, 17:35:08 UTC

And another anonymous hall of infamy candidate:
The owner of computer 5469669 whose contribution is:

State:
All (4576)
In progress (100)
Validation pending (735)
Validation inconclusive (913)
Valid (81)
Invalid (2747)
Error (0)

Application:
All (4576)
AstroPulse v6 (0)
AstroPulse v7 (1)
SETI@home Enhanced (0)
SETI@home v7 (4575)

Over 50% invalid....
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?
ID: 1666021 · Report as offensive     Reply Quote
Fractale2015
Volunteer tester

Send message
Joined: 28 Mar 15
Posts: 42
Credit: 260,222
RAC: 0
France
Message 1686100 - Posted: 30 May 2015, 18:25:26 UTC

Strozure (5359757)
NVIDIA GeForce 9600 GT (1024MB) driver: 337.88 OpenCL: 1.0

État: Tous (4494) · En cours (200) · Validation pending (1780) · Validation inconclusive (1498) · Valide (61) · Invalide (953) · Erreur (2)

---------
Steven L. Hill (6780580)
AMD AMD Radeon HD 6370D/6380G/6410D/6480G (Sumo) (512MB) driver: 1.4.1589 OpenCL: 1.1

État: Tous (1584) · En cours (61) · Validation pending (503) · Validation inconclusive (520) · Valide (28) · Invalide (472) · Erreur (0)


---------
Makoto.F (5025697)
NVIDIA GeForce 9800 GT (511MB) driver: 340.52 OpenCL: 1.0

État: Tous (3047) · En cours (8) · Validation pending (1354) · Validation inconclusive (947) · Valide (2) · Invalide (736) · Erreur (0)
ID: 1686100 · Report as offensive     Reply Quote
WezH
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 571
Credit: 66,120,754
RAC: 2,069
Finland
Message 1686225 - Posted: 31 May 2015, 7:48:09 UTC

Computer 7200981

PM sent

Laptop, ATI GPU making invalids.
"Please keep Your signature under four lines so Internet traffic doesn't go up too much"

- In 1992 when I had my first e-mail address -
ID: 1686225 · Report as offensive     Reply Quote
Fractale2015
Volunteer tester

Send message
Joined: 28 Mar 15
Posts: 42
Credit: 260,222
RAC: 0
France
Message 1687665 - Posted: 4 Jun 2015, 10:41:08 UTC - in response to Message 1686100.  

[AF>Le_Pommier>MacGeneration.com] LWayne (6271783)
AMD Radeon HD 4670 (256MB) OpenCL: 1.0

État: Tous (487) · En cours (18) · Validation pending (200) · Validation inconclusive (180) · Valide (5) · Invalide (84) · Erreur (0)

PM sent
ID: 1687665 · Report as offensive     Reply Quote
Profile Ageless
Volunteer tester
Avatar

Send message
Joined: 9 Jun 99
Posts: 14565
Credit: 3,718,189
RAC: 923
Netherlands
Message 1689031 - Posted: 7 Jun 2015, 21:41:58 UTC

I just sent Fred a private message, because he's wasting work by the hundreds, as he's using an AMD GPU with Catalysts 11.11, where 11.12 are the minimum required. So sad people don't ever look at their system, to check what it's doing.
Jord

Ancient Astronaut Theorists can tell you that I do not help with tech questions via private message. Please use the forums for that.
ID: 1689031 · Report as offensive     Reply Quote
Profile ReiAyanami
Avatar

Send message
Joined: 6 Dec 05
Posts: 116
Credit: 186,347,670
RAC: 68,601
Japan
Message 1690211 - Posted: 11 Jun 2015, 18:27:39 UTC
Last modified: 11 Jun 2015, 18:34:07 UTC

This probably had been discussed before but I couldn't find the explanation for it. So, could someone please direct me to the right answer?
What I see is that one of my PCs started accumulating invalid more than 30. So I looked at the WU and found many look like the following:

http://setiathome.berkeley.edu/workunit.php?wuid=1816027316

Where only two WUs were returned and mine labeled invalid while the other one said inconclusive.
How can this be possible? I thought matching at least 2 results validate the calculation. So, until they receive the third result shouldn't the both of currently returned WUs be labeled inconclusive?
ID: 1690211 · Report as offensive     Reply Quote
Josef W. Segur
Volunteer developer
Volunteer tester

Send message
Joined: 30 Oct 99
Posts: 4504
Credit: 1,414,761
RAC: 0
United States
Message 1690248 - Posted: 11 Jun 2015, 20:47:33 UTC - in response to Message 1690211.  

This probably had been discussed before but I couldn't find the explanation for it. So, could someone please direct me to the right answer?
What I see is that one of my PCs started accumulating invalid more than 30. So I looked at the WU and found many look like the following:

http://setiathome.berkeley.edu/workunit.php?wuid=1816027316

Where only two WUs were returned and mine labeled invalid while the other one said inconclusive.
How can this be possible? I thought matching at least 2 results validate the calculation. So, until they receive the third result shouldn't the both of currently returned WUs be labeled inconclusive?

There's special code in the SaHv7 validator which instantly invalidates a result which ran full length but does not have a best_autocorr signal at the end. That is to be sure nobody is trying to process v7 tasks with a v6 application under anonymous platform.

Your task details for those invalidated cases indicate normal CUDA50 processing, so the suspicion is that somehow the uploaded result files have been truncated or corrupted. I've asked Eric Korpela to take a look at those if he has time.
                                                                  Joe
ID: 1690248 · Report as offensive     Reply Quote
Profile ReiAyanami
Avatar

Send message
Joined: 6 Dec 05
Posts: 116
Credit: 186,347,670
RAC: 68,601
Japan
Message 1690317 - Posted: 12 Jun 2015, 0:54:18 UTC

Thank you, Joe.
I'm using Lunatics_Win64_v0.42.
ID: 1690317 · Report as offensive     Reply Quote
Profile jason_gee
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 24 Nov 06
Posts: 7489
Credit: 91,093,184
RAC: 0
Australia
Message 1690537 - Posted: 12 Jun 2015, 16:44:02 UTC - in response to Message 1690248.  
Last modified: 12 Jun 2015, 16:45:10 UTC

[Edit:] reply meant in response to message after Joe's, rather than to Joe's

That's a strange one I'll be watching the emails on. Differential diagnostic techniques might apply. Could you post screenshots of DPC latency checker runs at Idle and full crunch with shorties ? Just eliminates a swathe of possible system issues rather unknowable by the application.
"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.
ID: 1690537 · Report as offensive     Reply Quote
Josef W. Segur
Volunteer developer
Volunteer tester

Send message
Joined: 30 Oct 99
Posts: 4504
Credit: 1,414,761
RAC: 0
United States
Message 1690560 - Posted: 12 Jun 2015, 18:41:15 UTC - in response to Message 1690317.  

Thank you, Joe.
I'm using Lunatics_Win64_v0.42.

That's fine, the x41zc CUDA apps haven't changed.

Eric did check some of those results. The surprise is that there was no truncation or corruption, the result files were clean but simply didn't have the required best_autocorr (which should have been produced the very first time the autocorr search was performed, and typically updated by later searches). Given the millions of tasks which have been done by x41zc without previously showing that issue AFAIK, it's a rare puzzle.

Since jason_gee is the primary authority on those CUDA apps, I'll probably not contribute anything further.
                                                                   Joe
ID: 1690560 · Report as offensive     Reply Quote
Profile ReiAyanami
Avatar

Send message
Joined: 6 Dec 05
Posts: 116
Credit: 186,347,670
RAC: 68,601
Japan
Message 1691613 - Posted: 15 Jun 2015, 12:38:17 UTC
Last modified: 15 Jun 2015, 12:44:05 UTC

Sorry for the delay I was away from home.
I started DPC Latency checker while all the calculations suspended,
After 30 seconds, I started GPU calculations and kept it running for over one minutes. GPU usage was between 95 and 99% according to nVidia software.
Here's what's said:
Test Interval: 1000
Current Latency: 139
Absolute Maximum:166
I hope I used the software right (I'm not sure what I measured...)
Thanks.
ID: 1691613 · Report as offensive     Reply Quote
Profile jason_gee
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 24 Nov 06
Posts: 7489
Credit: 91,093,184
RAC: 0
Australia
Message 1691650 - Posted: 15 Jun 2015, 14:27:03 UTC - in response to Message 1691613.  
Last modified: 15 Jun 2015, 14:48:17 UTC

Yep, looks fine, as long as there were no big spikes (which would have shown in the maximum, if they had happened)

[Edit] in addition to the below items, I just noticed the systems are hyperthreaded ? it's a long shot, but simply freeing a (virtual) CPU core (if not already, sorry if I missed it) and/or raising the process priority of the application might influence those symptoms. Sing out if you'd need directions on that. If it helps, It would indicate pretty high system pressure IMO.[/ Edit]

That [DPC latency check] just eliminates a whole bunch of possibilities to do with system drivers of all sorts, PCIe transactions don't see any particularly bad delays, and that there are no weird power saving things going on that could make the CPU miss events ( e.g. Windows 7 default CPU processor management throttling down the CPU speed to 5%, hidden in advanced ::S ).

All that means is that whatever's causing that data to be missing, isn't because of a lot of basics, and more exploration is needed.

Is the storage drive Boinc crunches on an SSD, and if so what brand/model ? Also I see they're Intel CPUs, are you running Intel Chipset RAID or AHCI mode ?

These are just more things that combined with some boinc/api nuances could show something out of the ordinary. some other probing basics on various lines could range from reseating cards in the PCIe slot, and power connectors, checking temperatures/frequencies and voltages. Even though there is no direct evidence I see of an issue there, something minor could be affecting another part of the system, so doesn't hurt for the easy checks.

failing the easy stuff out of the way, does use of the special commode build available at: http://www.jgopt.org/download.html change the symptoms ? Many could help putting that in if you need more details.
"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.
ID: 1691650 · Report as offensive     Reply Quote
Profile ReiAyanami
Avatar

Send message
Joined: 6 Dec 05
Posts: 116
Credit: 186,347,670
RAC: 68,601
Japan
Message 1691756 - Posted: 15 Jun 2015, 17:17:20 UTC
Last modified: 15 Jun 2015, 17:18:36 UTC

Jason, thank you for taking time.
SSD I'm using is Corsair CSSD-F120GB2 and this is the only storage device on the system.
By virtual CPU, do you mean 4 core processor running 8 threads?

Going back to my original question, the reason that some of my returned WUs are labeled Invalid is due to a part of data is missing from those? If so, I didn't know there was a internal mechanism like that existed.
Thanks again.
ID: 1691756 · Report as offensive     Reply Quote
Profile jason_gee
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 24 Nov 06
Posts: 7489
Credit: 91,093,184
RAC: 0
Australia
Message 1691785 - Posted: 15 Jun 2015, 18:16:47 UTC - in response to Message 1691756.  

Yeah, Intel Hyperthreading, 4 physical cores running 8 threads (etc). Someone will have to chime in for correct options by Boinc client version (as I use an old client), but setting % of processors to use out of the 8 threads should enable you to free one or two.

For the Cuda application process priority, there is a file called ***mbcude.cfg (it is named slightly differently depending on the stock application you process with), which can be edited with notepad or similar plantext editor. There is an example line to uncomment for processpriority, and a suitable setting could be 'normal'

Yes, I was in on Joe's email correspondance with Project staff, which indicated the results were missing a chunk of information, though oddly the result file had its proper closing tags. That amounts to a rare and interesting mystery, as opposed to the known application faults and Boinc mechanism quirks, because it is something that should have happened before if it was application related.

Should freeing a (virtual) core or two to feed the GPU, raising the process priority a little to make sure there are not weird delays going on, and/or switching to special builds all fail to change the symptoms in any way, then we have a genuine headscratcher :)
"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.
ID: 1691785 · Report as offensive     Reply Quote
Profile betreger Project Donor
Avatar

Send message
Joined: 29 Jun 99
Posts: 8251
Credit: 19,749,114
RAC: 13,173
United States
Message 1694047 - Posted: 20 Jun 2015, 19:19:00 UTC
Last modified: 20 Jun 2015, 19:19:26 UTC

I politely sent a PM to user that was doing this
State: All (1476) · In progress (12) · Validation pending (590) · Validation inconclusive (592) · Valid (20) · Invalid (260) · Error (2)

After a couple of cordial messages back and forth he has decided to quit the project rather than asking for help.
His response is
I guess I'll just continue with Enigma & drop out of SETI. Too bad, 'cause I've been part of this since 1999.

ID: 1694047 · Report as offensive     Reply Quote
Fractale2015
Volunteer tester

Send message
Joined: 28 Mar 15
Posts: 42
Credit: 260,222
RAC: 0
France
Message 1694641 - Posted: 22 Jun 2015, 17:35:25 UTC - in response to Message 1694047.  

Odinlore II (7504170)
AMD AMD Radeon HD 6200/6300/7200/7300 series (Wrestler) (384MB) driver: 1.4.1589 OpenCL: 1.1

État: Tous (2036) · En cours (3) · Validation pending (606) · Validation inconclusive (936) · Valide (21) · Invalide (470) · Erreur (0)

PM sent
ID: 1694641 · Report as offensive     Reply Quote
Fractale2015
Volunteer tester

Send message
Joined: 28 Mar 15
Posts: 42
Credit: 260,222
RAC: 0
France
Message 1694685 - Posted: 22 Jun 2015, 20:27:31 UTC - in response to Message 1694641.  

Garry Webb (6789552)
AMD AMD Radeon HD 6520G/6530D/6550D/6620G (SuperSumo) (512MB) driver: 1.4.1589 OpenCL: 1.1

État: Tous (1182) · En cours (13) · Validation pending (366) · Validation inconclusive (494) · Valide (13) · Invalide (296) · Erreur (0)

PM sent
ID: 1694685 · Report as offensive     Reply Quote
Previous · 1 . . . 6 · 7 · 8 · 9 · 10 · 11 · 12 . . . 22 · Next

Message boards : Number crunching : Invalid Host Messaging


 
©2018 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.