Message boards :
Number crunching :
More Invalid wu's
Message board moderation
Previous · 1 · 2 · 3 · Next
Author | Message |
---|---|
Send message Joined: 26 Apr 08 Posts: 87 Credit: 64,801,496 RAC: 0 |
Invalid ps_sgr_210F5_3s_hiw_4147052_1245758465_0 on an old iMac here. Plus SETI Classic = 21,082 WUs |
Send message Joined: 26 Jul 08 Posts: 627 Credit: 94,940,203 RAC: 0 |
As I said already, the invalid WUs have nothing to do with the application you use. It can be the stock app for a Mac, you will still have some invalid WUs. It is simply the "plausibility check" of the result on the server that sometimes decides the result can't be true (even if it is completely correct). That is probably caused by some strange parameter combinations in a WU, which delivers an extremely bad fitness. You shouldn't worry about this. |
Send message Joined: 12 Nov 07 Posts: 2425 Credit: 524,164 RAC: 0 |
You shouldn't worry about this. But we need something to worry and complain about! ;-) Doesn't expecting the unexpected make the unexpected the expected? If it makes sense, DON'T do it. |
Send message Joined: 24 Dec 07 Posts: 1947 Credit: 240,884,648 RAC: 0 |
You shouldn't worry about this. The error rate I'm seeing adds up to about 2000 credits a day lost.....now that is something to complain about! |
Send message Joined: 2 Jan 08 Posts: 122 Credit: 69,480,206 RAC: 1,380 |
I have only noticed a few of these faulty work units. So far all the ones I have caught have been on my Pentium 4 Windows machine and each time the Work Unit in question has been a "ps_sgr_210F5_3s_hiw_" type. Instead of running for 1 1/2 to 2 hours as normal they run for up to 8 hours. Always finish and don't error out during whilst running, but I aborted one at 5 1/2 hours last night and another ran for over 8 hours after I had gone to bed, giving zero credit as classed as invalid. Have probably had more but work units don't hang round all that long. |
Send message Joined: 19 Mar 09 Posts: 27 Credit: 117,670,452 RAC: 0 |
I have also problems with some WU's. I don't mind the loss of credit, but when I get "display driver stopped responding" the astronomy_0.19_ATI_x64f.exe process starts eating 100% of CPU time! As I can't babysit the computers all the time, I have 100% CPU load for long times and BOINC not doing any work until I kill that process or restart BOINC. This is an example of it: http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=88488548 Using Win7 64bit, Cat 9.5, 3850. Often the WU's that jam up are not shown on the list, as after restart processing starts fromn 0% again and might go through the second time. Have tried to change f and w parameters, but just get low GPU utilization and problems still appear.. Current settings: <cmdline>n1 w1.7 f100</cmdline> Any chance of getting the ATI-app to notice it's jammed up and auto-kill itself? Any idea why I'm getting these problems? Oh and sometimes no credit is granted for succesfull WU's: http://milkyway.cs.rpi.edu/milkyway/workunit.php?wuid=86663177 |
Send message Joined: 6 Apr 08 Posts: 2018 Credit: 100,142,856 RAC: 0 |
|
Send message Joined: 2 Jan 08 Posts: 122 Credit: 69,480,206 RAC: 1,380 |
I have only noticed a few of these faulty work units. Just killed another one of these 210F5_3s work units at over 5 hours run time. Interesting that the reported result shows Zero CPU time not the 5 hours I have been running. My output on this P4 has dropped a lot recently and i now put it down to this same WU type that is taking so long to run and then gives Zero Credit for the effort, so I know I am getting more than the couple that I have been able to catch. |
Send message Joined: 12 Nov 07 Posts: 2425 Credit: 524,164 RAC: 0 |
huh finally had one, waste of 85 min. : ps_sgr_210F5_3s_hiw_6368068_1246063275_0 http://milkyway.cs.rpi.edu/milkyway/result.php?resultid=89593602 Since the error rate is higher with this bunch, couldn't the code for this series of wu's be checked if it's correct or too strict? Or has it? Doesn't expecting the unexpected make the unexpected the expected? If it makes sense, DON'T do it. |
Send message Joined: 17 Mar 08 Posts: 165 Credit: 410,228,216 RAC: 0 |
Did the work but no credit...HUH? DD, 90683000 Name ps_sgr_210F5_3s_hiw_7397059_1246214151_0 Workunit 88832145 Created 28 Jun 2009 18:35:52 UTC Sent 28 Jun 2009 18:37:09 UTC Received 28 Jun 2009 18:50:06 UTC Server state Over Outcome Success Client state Done Exit status 0 (0x0) Computer ID 84507 Report deadline 1 Jul 2009 18:37:09 UTC CPU time 211.5469 stderr out <core_client_version>6.6.36</core_client_version> <![CDATA[ <stderr_txt> Running Milkyway@home ATI GPU application version 0.19f by Gipsel CPU: AMD Athlon(tm) 64 X2 Dual Core Processor 3800+ (2 cores/threads) 2.16499 GHz (313ms) CAL Runtime: 1.3.145 Found 1 CAL device Device 0: ATI Radeon HD 3800 (RV670) 1024 MB local RAM (remote 28 MB cached + 256 MB uncached) GPU core clock: 759 MHz, memory clock: 504 MHz 320 shader units organized in 4 SIMDs with 16 VLIW units (5-issue), wavefront size 64 threads supporting double precision 3 WUs already running on GPU 0 No free GPU! Waiting ... 63.9688 seconds. Starting WU on GPU 0 main integral, 320 iterations predicted runtime per iteration is 473 ms (33.3333 ms are allowed), dividing each iteration in 15 parts borders of the domains at 0 112 216 320 432 536 640 752 856 960 1072 1176 1280 1392 1496 1600 Calculated about 9.89542e+012 floatingpoint ops on GPU, 1.23583e+008 on FPU. Approximate GPU time 211.547 seconds. probability calculation (stars) Calculated about 2.46469e+009 floatingpoint ops on FPU. WU completed. CPU time: 63.6719 seconds, GPU time: 211.547 seconds, wall clock time: 703.21 seconds, CPU frequency: 2.165 GHz </stderr_txt> ]]> Validate state Invalid Claimed credit 0.883484783514827 Granted credit 0 application version 0.19 |
Send message Joined: 12 Nov 07 Posts: 2425 Credit: 524,164 RAC: 0 |
Can I suggest that a post be made on the homepage that mentions this problem. People might read that. Doesn't expecting the unexpected make the unexpected the expected? If it makes sense, DON'T do it. |
Send message Joined: 24 Dec 07 Posts: 1947 Credit: 240,884,648 RAC: 0 |
Cluster Physik said.... As I said already, the invalid WUs have nothing to do with the application you use. It can be the stock app for a Mac, you will still have some invalid WUs. It is simply the "plausibility check" of the result on the server that sometimes decides the result can't be true (even if it is completely correct). That is probably caused by some strange parameter combinations in a WU, which delivers an extremely bad fitness. You shouldn't worry about this. |
Send message Joined: 12 Nov 07 Posts: 2425 Credit: 524,164 RAC: 0 |
Another one, forgot to abort first. Task ID 92773155 Name ps_sgr_210F5_3s_hiw_9406080_1246479241_0 Workunit 90841166 |
Send message Joined: 26 Dec 07 Posts: 41 Credit: 2,582,082 RAC: 0 |
I dispute Cluster Physik's 'you shouldn't worry about this'. I only crunch during the eight-hour working day and these invalid tasks can run for almost two hours. It peeves me mightily when 25% of my crunch time is discarded for zero credit. It might not be a worry for those with GPU where only a few minutes worth of crunch is discarded but we don't all have that option. And another thing. All of these 3s tasks run 20-40% longer than 2s tasks. For no extra credit. |
Send message Joined: 26 Jul 08 Posts: 627 Credit: 94,940,203 RAC: 0 |
And another thing. All of these 3s tasks run 20-40% longer than 2s tasks. For no extra credit. Of course they give about a third more credits than the 2s WUs! 2s: 55.52 credits 3s: 74.24 credits 74.24/55.52 - 1 = 33.7% more |
Send message Joined: 6 Apr 08 Posts: 2018 Credit: 100,142,856 RAC: 0 |
I dispute Cluster Physik's 'you shouldn't worry about this'. I only crunch during the eight-hour working day and these invalid tasks can run for almost two hours. It peeves me mightily when 25% of my crunch time is discarded for zero credit. It might not be a worry for those with GPU where only a few minutes worth of crunch is discarded but we don't all have that option. I run both GPU and CPU-only on some boxes and see I have some WUs returning zero credit and appreciate the disappoitment when PCs have been running on 'wasted' WUs. However, as much as I want to amass as many credits as possible because it looks good and I like to see that progress in my science contributions when I check my stats - as much as all that, this is still an alpha project and participation still comes with all the server downtimes, problematic WUs and all the other glitches and gremlins that come with it. Fortunately we have gremlins like Cluster Physik to help us out ;) |
Send message Joined: 26 Dec 07 Posts: 41 Credit: 2,582,082 RAC: 0 |
Of course they give about a third more credits than the 2s WUs! With an exclamation mark, no less. OK, scratch that. Correct to read: 'Some 3s, while not invalidating, run 20–40% longer than others'. On my CPU, since The Great Credit Adjustment of February (or whenever it was), I get about 50cr/hr. Except for quite a few of the recent 3s which seem to run over time and deliver 35-40cr/hr. Not that I am getting all bent out of shape over it. It was just the 'there is no problem' assertion earlier in this thread that I felt like swatting. |
Send message Joined: 21 Mar 09 Posts: 15 Credit: 1,545,913 RAC: 0 |
Still getting frequent 0 credit wu's (see previous posts) the latest is http://milkyway.cs.rpi.edu/milkyway/workunit.php?wuid=92341139 . Is this issue going to be adressed? |
Send message Joined: 21 Aug 08 Posts: 625 Credit: 558,425 RAC: 0 |
Out of curiosity, has anyone noticed any of these invalid tasks showing up on an AMD system AND said system is NOT using the GPU app? I hadn't seen any on my Athlon64 3700+, but I attached my P4 yesterday and already have one. Additionally, everyone in this thread who has mentioned getting them has an Intel system. Further, Conan has so far only mentioned that the issue has only been on his P4... Only Dan Morris has mentioned an AMD, but he is using the GPU app on that one... |
Send message Joined: 21 Aug 08 Posts: 625 Credit: 558,425 RAC: 0 |
Out of curiosity, has anyone noticed any of these invalid tasks showing up on an AMD system AND said system is NOT using the GPU app? I hadn't seen any on my Athlon64 3700+, but I attached my P4 yesterday and already have one. Additionally, everyone in this thread who has mentioned getting them has an Intel system. Further, Conan has so far only mentioned that the issue has only been on his P4... Only Dan Morris has mentioned an AMD, but he is using the GPU app on that one... Answering my own question: I finally did see one on my AMD... The one that failed was: ps_sgr_210F5_3s_hiw_14747556_1247129544 Valid just before it was : ps_sgr_210F5_3s_hiw_14747555_1247129544 Maybe that will help someone narrow down the issue... |
©2024 Astroinformatics Group