Seti@Home optimized science apps and information
Optimized Seti@Home apps => Windows => Topic started by: Gus on 11 Aug 2007, 12:35:05 am
-
This is a brand new machine, so I don't know if these unrecoverable errors are coming from the hardware, software, or whatever. It's a core-2 quad (Q6600) right out of the box about 3 hours ago, a stock Dell computer, and I've never run anything on it but KWSN_2.4_SSE3-Core2_MB. It is not OC'ed nor modified in any way. It's cool in the house so temp should not be an issue. The box has displayed no errors so far except these.
Or they may just be bad wu's and mean nothing.
I received two of these errors so far, pretty close together, in about 2 hours of total crunch time. I'll attach a text file with the Messages Tab entries around the time of one of the errors, a URL to the two Results, and a URL to the computer.
I'm going to bed shortly so if you have any questions feel free to email me at obermege2 at comcast dot net and I'll reply in the morning. . . .
EDIT
OK, after watching this for a while it looks like they may be bad wu's. I've had several others and the names of them all begin with 03mr07aa.19893.3344.xx.x.xxx. I'll leave this post up just in case.
[attachment deleted by admin]
-
We'll take a look - right now, your result is the only one available for those WUs, so it's difficult to compare.
They do look like "can happen" mistakes.
Regards,
Simon.
-
One has now been returned OK by someone else. Also the most recent one is not the same vintage as the others. (Ignore the very first one, it was likely caused when I switched to the optimized app.)
It may indeed be a problem with the machine after all. I'll reseat the memory and run some diags. I've since set it to receive no new work. Hate to do that the way things are running but I don't want to trash a bunch of wu's.
Can anyone suggest good free diagnostics that really torture the machine?
(Later) I now see that it spit out a whole page of errors one after the other. Just reseated memory and restarted, we'll see.
-
Prime95 - CPU torture test
MemTest86+ - Memory test (runs at boot-time)
-
Thanks, I have 4 copies of Prime95 running now, I'll do the memtest after they've run a while.
-
Hi,
I have many WUs with same error. Bad WUs? The following is common:
1: Windows Vista
2:
>
Access Violation (0xc0000005)
- Paged Pool Usage -
QuotaPagedPoolUsage: 43920, QuotaPeakPagedPoolUsage: 43952
QuotaNonPagedPoolUsage: 2512, QuotaPeakNonPagedPoolUsage: 2512
<
Peak value exceeds Qouta, but this is not reflected in Windows evenlog..
Here are more WUs:
http://setiathome.berkeley.edu/result.php?resultid=588614096
http://setiathome.berkeley.edu/result.php?resultid=588541570
http://setiathome.berkeley.edu/result.php?resultid=588535840
http://setiathome.berkeley.edu/result.php?resultid=588535838
http://setiathome.berkeley.edu/result.php?resultid=588535820
http://setiathome.berkeley.edu/result.php?resultid=588530274
http://setiathome.berkeley.edu/result.php?resultid=588530252
http://setiathome.berkeley.edu/result.php?resultid=588466412
http://setiathome.berkeley.edu/result.php?resultid=588466410
http://setiathome.berkeley.edu/result.php?resultid=588466408
http://setiathome.berkeley.edu/result.php?resultid=588466406
http://setiathome.berkeley.edu/result.php?resultid=588466404
http://setiathome.berkeley.edu/result.php?resultid=588466402
http://setiathome.berkeley.edu/result.php?resultid=588466274
http://setiathome.berkeley.edu/result.php?resultid=588434143
http://setiathome.berkeley.edu/result.php?resultid=588434135
http://setiathome.berkeley.edu/result.php?resultid=588434099
http://setiathome.berkeley.edu/result.php?resultid=588434077
http://setiathome.berkeley.edu/result.php?resultid=588432479
http://setiathome.berkeley.edu/result.php?resultid=588432477
I will install r2.4 on another box to see if this can be reproduced on Windows XP.
Morten
-
@Morten-
Those are indeed the same errors I was getting. They always happened after running only a few seconds. Mine is running Vista Home Basic on a Q6600. (Darn I hate Vista, but that's another issue.)
@Simon-
I'm beginning to think there may be a problem with the discovery routine that the science app runs when a result first starts. What makes me think so is that I noticed on two occasions all four results that were in progress and had been running fine failed when I stopped and restarted BOINC. The beginning is the only time they fail. Once they get going they run fine unless Manager is stopped and restarted, then they may or may not fail immediately.
?????
-
Simon, See "Client error with Chicken on 64bit Vista" in the Number Crunching boards. The author is running the 32 bit client on Vista 64, someone else reported the same with 32-bit Vista. One is running a core 2 E6600, the other a core 2 T7400. The second person reports that 64-bit on 64-bit works fine.
The problem looks like it may be specific to the core2 32-bit app. If so, I'm sorry to offer problems, but relieved that my machine may be OK after all. Prime95 is running 4 threads clean for about 3 hours now. I'm going to try the "stock" MB app and see how that goes.
-
I've also had three (so far) unrecoverable errors with 2.4 - I've logged one in the thread Gus has just linked, but they took the database down just as I was going to post the other two! However, I disabled network access after the first one, and I've backed up the entire BOINC folder, so I should have WU / result / state files for the other two.
The first error was on a high AR linefeed WU, the next two are October deadline multibeams. I'll post links to the results on the SETI board when the server is back up.
This is on my octo, running 32-bit Vista Business, BOINC v5.10.13, and the Core 2 variant of Chicken 2.4 (for Xeon 5320). It hasn't shown any Windows error messages or other signs of distress, and crunching is continuing on all 8 cores, with a mixture of SETI, Einstein and CPDN. It's done about 20 WUs since I upgraded to 2.4 this morning, so these errors are the exception rather than the rule.
Richard Haselgrove
-
And one more error. I can't upload/report because the servers are still down, but I'll backup the data again before uploading/reporting.
The WUs affected are:
12my00aa.25379.26192.790908.3.204
03mr07ab.4943.6207.6.4.94
03mr07ab.4907.6207.5.4.101
03mr07ab.4544.11115.3.4.8
- I'll post WUID / RID when I can.
-
I have around ten unrecoverable errors as well since switching to 2.4.
I havn´t a long time, so it should something wrong with 2.4.
While i´m not able to report have to wait til the servers are up agai~.
Mike
-
Four more errors on Core 2 / Vista overnight (and they'll be the last - I'm now dry on that box).
No signs of distress on SSE3 / server boxes: unless anyone has any better suggestions, I think I'll switch the Vista box to the SSE3 compile when the servers come back up, and do a comparison run that way.
NB before anyone asks: this is a Dell Precision workstation, running at stock speed with plenty of cooling. And I upgraded to the A05 BIOS after the first reports of Core 2 errata, so I think it's unlikely (and highly co-incidental) if these errors were in any way platform-releated.
-
Okay,
Those were compiled using Visual Studio 2003 (as all apps until now), but Vista doesn't like VS 2003 anymore, it seems.
So, new apps compiled with VS 2005 SP1 (with Vista patches) will be uploaded shortly (today, if I can manage).
Regards,
Simon.
-
having vista32 and 2.4 app and no problems ...
-
Okay,
seeing as problems with 32-bit Windows apps on Vista have been reported, there will be new versions out shortly.
Those we`e compiled uwing Visual Studio 2003 (as all apps until now), but Vista doesn't like VS 2003 anymore, it seems.
So, new apps compiled with VS 2005 SP1 (with Vista patches) will be uploaded shortly (today, if I can manage).
Regards,
Simon.
Thanks - but don't feel you ne%d to r h, get a good night's sleep so you can think clearly!
I presume it would be helpful to run with th` new Cord 2 ' Vista when available, rather than SSE3: anything else I can do to help track it down?
Richard
-
Regards,
Simon.
Thanks - but don't feel you need to rush, get a good night's sleep so you can think clearly!
I presume it would be helpful to run tit` th` ne" Core 2 / Vista whef available, rather than SSE3: an{thing else I can do to help track it down?
Richard
Thanks indeed Simon. I agree completely with @ic `rd( no ush. Now that we h`ve work I'm running with the other app no problem. So, take your time an enjoy what's left of the weekend if you can.
When you do give it a pry if you 7ould like me to help test it don't hesitate to ask. My email address is listed above. Just attach the "+Dilåû t/ Intall".
Best Regards
Gus
-
my last post is not true today - have on all wu error - 1073741819 (0xc0000005)
Vista 32bit and 2.4 sse2AMD and 2.4SSE3 P4 patched too
-
this is in the stderror.txt
:
gstate.init() failed: -181
UNRECOGNIZED: measurementbenchmark_debug
UNRECOGNIZED: work_request_factor
UNRECOGNIZED: measurement_debug
UNRECOGNIZED: work_request_factor
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
UNRECOGNIZED: suspend_if_no_recent_input
i cannnot attach debugger to the code because it crashed immediatelly
but when i run it separately with -standalone (same workunit as before) then it is running without error ....
-
I saw Joe Segur's announcement of Crunch3r's new compilation for Vista.
Downloaded the SSSE3 version for my Xeon 5320, and immediately got this error:
http://setiathome.berkeley.edu/result.php?resultid=598036420
Advise caution pending further investigation.
-
Howdy,
first, if you try to do a "bench" test ... do NOT add "-show_benchmarks"...
simply run it with "application_name -bench"...
second,
.NET run time libraries at least ver. 2.0 are required to run the application (3.0 is even better).
I did test the apps on Vista ultimate 64 (all apps except the SSSE3) and did a test run on Windows XP SP2 (SMP PIII 500MHz)... all worked fine for me.
HTH
-
1) Working in unzipped download directory: copied testWU_1 to work_unit.sah. Got stderr.txt:
Can't open init data file - running in standalone mode
Can't open init data file - running in standalone mode
------- [ benchmark ] --------
PwrSpectOnly--: 507003 x1.00 0 -- [ avg magnitude = 1.8306 (16)]
sse_GetPSO_npr: 526561 x0.96 0
sse_GetPSO_p32: 560840 x0.90 0
sse_GetPSO_p64: 485576 x1.04 0
sse_GetPSO_p128: 593404 x0.85 0
PwrSpectOnly--[sse_GetPSO_p64]: 485576 (chosen)
------------------------------
Transpose--: 5275459 x1.00 0 -- [ avg magnitude = 0.9994 (15)]
Transpose2: 3351264 x1.57 0
Transpose4: 2036594 x2.59 0
and a result.sah - do you need that?
2) I can see .NET 1.1 in "Programs and Features", but not yet .NET 2 or 3. Hold on while I check.
-
FWIW, i do have another "unconfirmed" idea. I'll spuspect that boinc CC 5.10.x is causing some trouble too.
However, i cannot verify that cuz i do not run S@H any more and before that i never used a CC 5.10.X ( those are all badly broken).
Though i think it's worth trying to run the app using boinc CC 5.8.16.
-
I couldn't see .NET 2 anywhere in 'Programs and Features' (successor to 'Add or Remove Programs'). So I downloaded dotnetfx.exe from http://www.microsoft.com/downloads/details.aspx?FamilyID=0856eacb-4362-4b0d-8edd-aab15c5e04f5&displaylang=en - it says Vista Business is supported, on that page.
When I ran it, I got Error: Product already installed This product is already installed as part of the operating system.
I don't want to test install BOINC 5.8.16 tonight - too late for that sort of thing! - and anyway I've got no WUs to test, download problems all day >:(
I'll leave SETI suspended overnight, but network enabled - see if I can get some WUs to test that way.
-
FWIW, i do have another "unconfirmed" idea. I'll spuspect that boinc CC 5.10.x is causing some trouble too.
However, i cannot verify that cuz i do not run S@H any more and before that i never used a CC 5.10.X ( those are all badly broken).
I've been running CC 5.10.13 for a while now, and seen no evidence of the "badly broken" of which you speak. Runs SETI stock, KSWN 2.2B, Einstein and CPDN just fine. Two possible cases of the service stopping unexpectedly, but nothing else. What's the problem you see?
-
Result http://setiathome.berkeley.edu/result.php?resultid=597123181 has downloaded while we've been chatting, so I've let it run. It's a short one (~40 minutes), and so far it's running OK - 25% in just under 10 minutes. At least we'll get something different to look at in std_err.txt
-
I'll back Crunch3r's comments on 5.10.13 as it would not work w/ x32 -doze 2.4 SSSE3 on my T7200.
WUs would randomly slow by 75% and take 2-2.5 times as long to complete as they should. I was getting longer times than a P-M 1.7 running stock 5.27. NO WAY that is correct. in other cases, threads would start and freeze leaving only 1 thread running.
FWIW, this "stable" BOINC version also froze-up running 2.3 development aps and 2.2B.
I've tried resets, uninstalls/reinstalls, defrag/restart etc.
Once I went back to 5.4.11, everything ran perfect.
I also had issues w/ 5.10.13 on T7200 iMactel and "reverted" to 5.4.9. Problems went away.
IMO, 5.10.13 has problems and isn't a stable release. Actually, I've run Betas w/ less issues in the past than this "stable" one.
Not ranting on BOINC, but this one version is just particularly bad. :o
-
Hi Simon,
I've been seeing the same errors as Richard. I'm running 2.4_Windows_x32_SSSE3 on Vista Home Premium. This is a Core 2 Duo (E6600). For me the WU's always blow up as soon as they start, either on a new WU or on one that was running normally and is being restarted by closing BOINC and restarting it.
After reading your suggestions in this thread I confirmed that .NET Framework 3.0 is installed on the system and I backloaded the CC to 5.8.16. It was running 5.10.13. But I am getting the same errors.
Here (http://setiathome.berkeley.edu/result.php?resultid=595641500) is the WU, and I'm attaching the original result file which I managed to save.
Regards,
Gus
More Info -
- This is not the same machine that I reported at the beginning of this thread. That was a Q6600 with Vista Home Basic.
- I just realized that WinZip truncated the attached file name which ends with 4.5.11
- The machine has all available Windoz Updates installed.
- I'm going to bed now before I pass out.
[attachment deleted by admin]
-
Let one more run as a test.
http://setiathome.berkeley.edu/result.php?resultid=597127577 errored out under 3 seconds.
Suspending all SETI activity for the moment - I now have some spare WUs I can use for testing debug builds if you want.
-
Maybe I should have been deleting them - didn't know there was a limit. Try again.
I'll try SSE3 quickly, then must break for food.
-
Maybe I should have been deleting them - didn't know there was a limit. Try again.
I'll try SSE3 quickly, then must break for food.
@Richard - Any luck running SSE3?
@Crunch3r - You've probably considered this already, and if it is a silly suggestion please forgive a novice.
If I understand it correctly, when the app starts it tries several techniques on the data to determine the best routines to use to crunch it. Since (for me at least) these always fail within seconds of starting, might the problem be within that routine? In other words, might the app sometimes try an illegal operation as part of these tests? Just a thought.
BTW, I offered earlier and I'll do so again. I'm available for help in testing this if you would like. gobermeyer at comcast dot net.
-Regards
-
I've been trying the updated 2.4 SSSE3 app since earlier today, and so far 6 workunits have errored out within 2 or 3 seconds, and 4 workunits have been succesfully completed, uploaded, reported and credit was granted. This host (http://setiathome.ssl.berkeley.edu/show_host_detail.php?hostid=3700057) has Vista/32 Ultimate with a Core2Duo 6320 and Boinc 5.10.13......
[Edit:] Noticed a 35% decrease in crunching time in comparison to the stock client on an AR0.776008 workunit.
[Edit2:] Update: 7 compute errors, 10 successful, 3 pending.
[Edit3:] It seems that a workunit can error out in two or three seconds after a restart, even if it was running fine before the restart.
Greetings,
Sander
-
I'm just in the process of re-activating my test http://setiathome.berkeley.edu/show_host_detail.php?hostid=3728483 - currently running an SSE3 test build for Crunch3r.
I'm not sure if he's planning to work on it today, being Sunday and all that - we had a long session yesterday - but if you're willing to help with beta testing you could send him a private message.
-
I'm just in the process of re-activating my test http://setiathome.berkeley.edu/show_host_detail.php?hostid=3728483 - currently running an SSE3 test build for Crunch3r.
I'm not sure if he's planning to work on it today, being Sunday and all that - we had a long session yesterday - but if you're willing to help with beta testing you could send him a private message.
Howdy,
as you might know Richard Haselgrove did some stress test on his Xeon (see his link above) it seems to work well now,so i decided to release some new apps based on what was used for the test app.
I hope that those errors are fixed now, otherwise i'm running out of options now.
However if you want to test them ---> http://calbe.dw70.de/seti.html
Make sure you copy all files including the KWSN_DEBUG.pdb.
-
Great news - everything certainly seems to have gone very smoothly with today's build.
In that case, I'll drain down the test account over the next couple of hours, and swich to using the builds you've just linked under my normal host ID (2901600): that way, we'll also get some experience of BOINC v5.10.13, a larger cache (which means they'll validate quicker - though that's a relative term, at the moment), and co-running with other BOINC projects.
-
Running the latest KWSN_2.4V_MB_SSSE3.exe now... I'll post my findings when this host finishes some workunits. Thanks for the hard work! It's really appreciated.
Greetings,
Sander
[Update1:] So far so good: 7 results returned and credit granted. 0 errors. Restarted boinc service twice without errors.
-
Thanks Crunch3r. I've updated a Conroe/Vista Home Premium running 5.10.13 with the new app. It's currently running a few 22 minute WU's so we should get confirmation quickly.
Update. OK so far, two WU's started and returned, both have received credit. Two new WU's started, so far none have errored-out. The latest two will take longer.
-
OK. So far I am running 2.4V on the following-
SSE2 on a Willamette with Windows 2000
SSE2 on a Northwood with XP Pro
SSE3 on a Prescott with XP Pro
Two - SSSE3 on a Conroe (Duo) both with Vista Home Premium
SSSE3 on a Kentsfield (Quad) with Vista Home Basic
All are running the 5.10.13 CC.
So far there has been NO sign of trouble. I've had a few "normal" -9 overflows but no sign of the dreaded -1073741819 error. All -9's that I've checked, about 6 or 7, have matched my wingman so they were valid.
Great Job ! to Crunch3r and thanks Richard for helping to rush this through.
Regards,
Gus
-
Well done, guys! 8)
...i decided to release some new apps based on what was used for the test app. I hope that those errors are fixed now...
Crunch3r, what's the difference between the 'good' build and the failing one? Only different compile options, or indeed the selection code, run each time at the beginning?
Peter
-
yo well done hurry to crunching on vistas
-
Well done, guys! 8)
...i decided to release some new apps based on what was used for the test app. I hope that those errors are fixed now...
Crunch3r, what's the difference between the 'good' build and the failing one? Only different compile options, or indeed the selection code, run each time at the beginning?
Peter
The main difference is that the failing apps were based on the 32 bit code base and the 'good' builds are based on the x64 code. Besides that ICC 10 and IPP 5.2 were used to compile the apps.
HTH
-
This the report I keep getting. Where did I go wrong?
Thanx
Dennis
Kompooter User :-\
<?xml version="1.0" encoding="UTF-16"?>
<DATABASE>
<EXE NAME="KWSN_2.4_SSE2-AMD_MB.exe" FILTER="GRABMI_FILTER_PRIVACY">
<MATCHING_FILE NAME="KWSN_2.4_SSE2-AMD_MB.exe" SIZE="1723904" CHECKSUM="0x9345E91" BIN_FILE_VERSION="2.4.0.0" BIN_PRODUCT_VERSION="5.0.2.7" PRODUCT_VERSION="5, 0, 2, 7" FILE_DESCRIPTION="Compiled with ICC 9.1 and IPP 5.1" COMPANY_NAME="Lunatics.at" PRODUCT_NAME="SETI@Home Enhanced Worker" FILE_VERSION="2, 4, 0, 0" ORIGINAL_FILENAME="KWSN_2.4_SSE2-AMD_MB.exe" INTERNAL_NAME="seti_boinc" LEGAL_COPYRIGHT="Copyright 2007, Regents University of California" VERFILEDATEHI="0x0" VERFILEDATELO="0x0" VERFILEOS="0x4" VERFILETYPE="0x1" MODULE_TYPE="WIN32" PE_CHECKSUM="0x0" LINKER_VERSION="0x0" UPTO_BIN_FILE_VERSION="2.4.0.0" UPTO_BIN_PRODUCT_VERSION="5.0.2.7" LINK_DATE="08/10/2007 21:51:57" UPTO_LINK_DATE="08/10/2007 21:51:57" VER_LANGUAGE="English (United States) [0x409]" />
</EXE>
<EXE NAME="kernel32.dll" FILTER="GRABMI_FILTER_THISFILEONLY">
<MATCHING_FILE NAME="kernel32.dll" SIZE="930816" CHECKSUM="0xF60887C3" BIN_FILE_VERSION="5.1.2600.1560" BIN_PRODUCT_VERSION="5.1.2600.1560" PRODUCT_VERSION="5.1.2600.1560" FILE_DESCRIPTION="Windows NT BASE API Client DLL" COMPANY_NAME="Microsoft Corporation" PRODUCT_NAME="Microsoft® Windows® Operating System" FILE_VERSION="5.1.2600.1560 (xpsp2_gdr.040517-1325)" ORIGINAL_FILENAME="kernel32" INTERNAL_NAME="kernel32" LEGAL_COPYRIGHT="© Microsoft Corporation. All rights reserved." VERFILEDATEHI="0x0" VERFILEDATELO="0x0" VERFILEOS="0x40004" VERFILETYPE="0x2" MODULE_TYPE="WIN32" PE_CHECKSUM="0xEC3A9" LINKER_VERSION="0x50001" UPTO_BIN_FILE_VERSION="5.1.2600.1560" UPTO_BIN_PRODUCT_VERSION="5.1.2600.1560" LINK_DATE="06/17/2004 17:58:35" UPTO_LINK_DATE="06/17/2004 17:58:35" VER_LANGUAGE="English (United States) [0x409]" />
</EXE>
</DATABASE>
-
This the report I keep getting. Where did I go wrong?
...
Hmm, the version of kernel32.dll indicates you're using Windows XP SP2 and we haven't had any other reports of difficulty on that OS.
I see you chose the SSE2 AMD version. Could it be that you have an Athlon XP system, for which you need the SSE version? If that's not it, please post more details about your system, and maybe a link to the host page at SETI@home.
Joe