Just curious: VALIDATE error for a specific WU
21 Feb 2013 15:52:40 UTC
Topic 196817
(moderation:
Hi!
I've noticed something in a BRP4 'Arecibo' WU, I completed and which ended in a 'VALIDATE error', something, what till now (around 3 years) has never happened to me, or at least I did not notice.
The ID of WU in question is:
148698799 (If needed, I have also screenshots, where the whole name of 'target' is and also a lot of other info...);
What seemed strange to me, was the fact, that ALL 6 wingman assigned to this task ended it as same as me, with 'VALIDATE error' (one of them had also a CPU 'version' of the task) and then the WU in question was canceled.
I think, we also had (at least 2 of us) different error codes (mine was 129:10....01, for the other I'm not sure, but it could be 131:....).
At least but not last, I have experienced 'VALIDATE errors' before, but very roghly ONLY in 0,001% of GPU tasks I completed.
Also, all the WUs I received before and after that, were VALID or are waiting for validation.
Was there maybe a reason in the WU itself?
D
P.S.
Thank You for the advice(s) I was given in the thread 'A big increase in VALIDATION time'.
I gave a very bad example and much of that what I was experiencing was a result of a bad habit of mine:
'Get a 'bunch' of WUs (200-), and only AFTER completing them, get new ones.'
A VERY BAD practice, which I (at least partly) immediately abandoned.
“A little knowledge is a dangerous thing. So is a lot.” - Albert EINSTEIN
Just curious: VALIDATE error for a specific WU
)
why is getting a Batch of WU's and finishing them and then getting another batch a bad idea ?
This is what I was planing to do .
Fluxcore: If you are
)
Fluxcore:
If you are worried about points or credit and the waiting time for them to validate it would be best to set in preferences for Computing preferences under Network usage Maintain enough work for an additional xx days to a number like 3 or 5 days. This is because it can and does take time for others to get the work and then turn it back in. Alot of us here have 3 to 5 days and more of work cache to work on. Like getting 200 task at a time.
If you don't really care about the time for work to get validated changing the days of work don't really matter. I have set work for my desktop to keep enough work for 5 additional days and I update it myself from time to time. My Pending credit so far stays around 7k. I don't care about the credit points. I only care about getting work done with out errors. We get 14 days to get the work done and turned back in I start my work for CPU with about 10 days or so left to turn it back in. I keep about 175 task in my client ready to work and update to get new work twice day and sometimes more. The GPU work is much faster and this is why I keep updating so I don't have to run out of work or even get close to it.
It's not a bad idea just make sure you don't get too much work where you can't get it turned back in before the deadline. Some users stress to much on credit. Just let it work is my advice and don't stress about credit only stress if something goes wrong like no work or most to all work fails.
PC setup MSI-970A-G46 AMD FX-8350 8 core OC'd 4.45GHz 16GB ram PC3-10700 Geforce GTX 650Ti Windows 7 x64 Einstein@Home
I realy dont care about
)
I realy dont care about points I just want to work this computer to the max without errors . I do like to see the results just to see if i have this computer set up properly ,but it is hard to compare with other computer set-ups because no 2 are the same . So I realy dont know if i have it setup properly .
The points thing is kind of my way of seeing if the set up I have done is the best I can get out of it . I may OC it as well but I dont know if that is worth it , I have the cooling capacity to OC but I would rather have no errors .
From time to time (less than
)
From time to time (less than one beam per week on average; we process >150 beams per day) we have a bad "beam" in the Arecibo data. The workunits generated from such a beam are processed technically ok by the application, but the result is regarded as garbage by the validator, resulting in validate errors. We try to cancel these workunits when we spot them before too much computing time is wasted, but sometimes we are too busy to watch and some workunits collect up to 20 validate error results before they error out automatically.
BM
BM
I've many viidate errors on
)
I've many viidate errors on my ATI tasks.
p2030.20121018.G176.46-02.82.C.b3s0g0.00000_3192_1 149543554 3462777 20 Feb 2013 8:35:17 UTC 24 Feb 2013 10:00:53 UTC Validate error 2,748.04 298.29 2.05 --- Binary Radio Pulsar Search (Arecibo) v1.34 (opencl-ati)
p2030.20121018.G176.46-02.82.C.b3s0g0.00000_2808_0 149543464 3462777 20 Feb 2013 8:35:17 UTC 20 Feb 2013 9:21:59 UTC Validate error 2,767.05 246.90 1.71 --- Binary Radio Pulsar Search (Arecibo) v1.34 (opencl-ati)
p2030.20121018.G176.46-02.82.C.b6s0g0.00000_3504_0 149541421 3462777 20 Feb 2013 8:35:18 UTC 23 Feb 2013 22:26:09 UTC Validate error 2,744.63 244.59 1.68 --- Binary Radio Pulsar Search (Arecibo) v1.34 (opencl-ati)
p2030.20121018.G176.89-02.16.C.b6s0g0.00000_1848_1 149540914 3462777 20 Feb 2013 8:35:18 UTC 20 Feb 2013 21:26:01 UTC Validate error 2,777.47 196.83 1.36 --- Binary Radio Pulsar Search (Arecibo) v1.34 (opencl-ati)
p2030.20121018.G193.25-04.73.N.b1s0g0.00000_3624_1 149532794 3462777 20 Feb 2013 8:35:17 UTC 20 Feb 2013 20:39:46 UTC Validate error 2,654.53 199.01 1.38 --- Binary Radio Pulsar Search (Arecibo) v1.34 (opencl-ati)
p2030.20121018.G193.38-04.51.N.b2s0g0.00000_3120_0 149518908 3462777 20 Feb 2013 8:35:18 UTC 20 Feb 2013 22:11:52 UTC Validate error 2,749.12 228.18 1.58 --- Binary Radio Pulsar Search (Arecibo) v1.34 (opencl-ati)
p2030.20120507.G43.98-01.62.S.b5s0g0.00000_3664_1 148860756 3462777 15 Feb 2013 11:20:00 UTC 15 Feb 2013 18:21:40 UTC Validate error 2,753.11 239.62 1.69 --- Binary Radio Pulsar Search (Arecibo) v1.34 (opencl-ati)
p2030.20120507.G43.98-01.62.S.b1s0g0.00000_2448_0 148857580 3462777 15 Feb 2013 11:20:00 UTC 15 Feb 2013 12:52:07 UTC Validate error 2,728.07 242.00 1.71 --- Binary Radio Pulsar Search (Arecibo) v1.34 (opencl-ati)
p2030.20120507.G60.95-00.49.N.b1s0g0.00000_1256_1 148847615 3462777 15 Feb 2013 11:20:00 UTC 15 Feb 2013 14:31:39 UTC Validate error 2,741.77 243.41 1.72 --- Binary Radio Pulsar Search (Arecibo) v1.34 (opencl-ati)
p2030.20120507.G60.69-00.03.N.b3s0g0.00000_3160_1 148845052 3462777 15 Feb 2013 11:20:00 UTC 15 Feb 2013 23:16:30 UTC Validate error 2,836.95 250.80 1.77 --- Binary Radio Pulsar Search (Arecibo) v1.34 (opencl-ati)
p2030.20120517.G43.49-00.69.C.b4s0g0.00000_736_0 148814583 3462777 15 Feb 2013 11:20:00 UTC 15 Feb 2013 21:42:35 UTC Validate error 2,760.51 241.80 1.71 --- Binary Radio Pulsar Search (Arecibo) v1.34 (opencl-ati)
Greetings
RE: I've many viidate
)
http://einsteinathome.org/host/3462777/tasks&offset=0&show_names=1&state=4&appid=0
That is your link to your task should anyone want to look at them.
Now I looked threw afew not too many but the few I did I see either only your task failed while others didn't or something else.
Many things can cause work to fail like high temps. or overclocking and even drivers just to list afew. Newer drivers seem to be buggy as what works for me may not work for you "driver version". You will need to check your settings, temps, drivers and maybe afew other things to try to pin point why they are failing.
PC setup MSI-970A-G46 AMD FX-8350 8 core OC'd 4.45GHz 16GB ram PC3-10700 Geforce GTX 650Ti Windows 7 x64 Einstein@Home
RE: I've many viidate
)
second to that. both my ATI machines produce 10-12 invalid WUs per day.
i even try do downclock CPU and GPU below stock speed but nothing improved.
http://einsteinathome.org/host/6617227/tasks&offset=0&show_names=1&state=4&appid=0 and http://einsteinathome.org/host/6572526/tasks&offset=0&show_names=1&state=4&appid=0
what is the reason?
RE: RE: I've many viidate
)
Maybe they haven't quite got the 79?? card software just right yet here at Einstein? Hva you checked other 79?? users and see how they are doing? Also how many units are your running at one time? Lower it to just two, or even one unit, and see if that fixes the problem. What does the software gpu-z say about gpu load, fan speed and card temps? You can get the latest gpu-z here:
http://www.techpowerup.com/downloads/2198/TechPowerUp_GPU-Z_v0.6.7.html
I´ve picked up a task which
)
I´ve picked up a task which has 2 times a validate error 1 time error while computing 1 is in progress and 1 is unsent perhaps someone should throw an eye on it.
http://einsteinathome.org/workunit/150860898
RE: I´ve picked up a task
)
Already canceled by the project!
As to why see Bernd's message earlier in this thread.