Validator problem?
log in

Advanced search

Message boards : Number crunching : Validator problem?

Previous · 1 · 2 · 3 · Next
Author Message
zombie67 [MM]
Avatar
Send message
Joined: 9 Mar 15
Posts: 15
Credit: 1,283,842
RAC: 0
Message 113 - Posted: 27 Mar 2015, 13:15:14 UTC

Can we please get an update on the validation situation?

It looks like tasks crunched by both windows and linux will not validate together. Which means if the the third machine is windows, the linux machine loses out. And vice versa. Until this gets resolved, can homogeneous redundancy please be turned on?

http://boinc.berkeley.edu/trac/wiki/HomogeneousRedundancy
____________
Dublin, California
Team: SETI.USA

Profile UBT - Chris
Send message
Joined: 19 Mar 15
Posts: 8
Credit: 3,921
RAC: 0
Message 114 - Posted: 28 Mar 2015, 0:21:54 UTC

Hi, Tim - no, I haven't, but I'll try it next time my laptop's on - after I've been to sleep... Then I'll let you know either here, or on the team forum if it has worked.
____________
Thanks, from UBT - Chris



zombie67 [MM]
Avatar
Send message
Joined: 9 Mar 15
Posts: 15
Credit: 1,283,842
RAC: 0
Message 115 - Posted: 28 Mar 2015, 14:29:38 UTC - in response to Message 113.

It looks like tasks crunched by both windows and linux will not validate together. Which means if the the third machine is windows, the linux machine loses out. And vice versa. Until this gets resolved, can homogeneous redundancy please be turned on?

http://boinc.berkeley.edu/trac/wiki/HomogeneousRedundancy



Bump. We are wasting many many thousands of tasks
____________
Dublin, California
Team: SETI.USA

Dr Who Fan
Avatar
Send message
Joined: 12 Mar 15
Posts: 18
Credit: 4,613
RAC: 0
Message 116 - Posted: 28 Mar 2015, 16:15:16 UTC - in response to Message 115.

Add to that tasks completed using Windows XP STILL DO NOT VALIDATE under new version 3.x
Like I said in my previous post... I am NOT willing to contribute to a project and NOT receive credit for work done.
____________

Dr Who Fan
Avatar
Send message
Joined: 12 Mar 15
Posts: 18
Credit: 4,613
RAC: 0
Message 121 - Posted: 30 Mar 2015, 19:15:14 UTC - in response to Message 116.

Add to that tasks completed using Windows XP STILL DO NOT VALIDATE under new version 3.x
Like I said in my previous post... I am NOT willing to contribute to a project and NOT receive credit for work done.

I would like an answer from the project OWNER/ADMIN on the status of
WHY NONE OF MY COMPLETED TASKS UNDER WINDOWS XP HAVE VALIDATED on host 179

____________

zombie67 [MM]
Avatar
Send message
Joined: 9 Mar 15
Posts: 15
Credit: 1,283,842
RAC: 0
Message 124 - Posted: 1 Apr 2015, 12:13:29 UTC - in response to Message 115.

It looks like tasks crunched by both windows and linux will not validate together. Which means if the the third machine is windows, the linux machine loses out. And vice versa. Until this gets resolved, can homogeneous redundancy please be turned on?

http://boinc.berkeley.edu/trac/wiki/HomogeneousRedundancy



Bump. We are wasting many many thousands of tasks


Bump again. Status please?
____________
Dublin, California
Team: SETI.USA

Profile UBT - Chris
Send message
Joined: 19 Mar 15
Posts: 8
Credit: 3,921
RAC: 0
Message 125 - Posted: 1 Apr 2015, 19:20:06 UTC

Mine have all recently validated... I think, and I am on Windows 8
____________
Thanks, from UBT - Chris



zombie67 [MM]
Avatar
Send message
Joined: 9 Mar 15
Posts: 15
Credit: 1,283,842
RAC: 0
Message 126 - Posted: 1 Apr 2015, 22:42:07 UTC - in response to Message 125.

Mine have all recently validated... I think, and I am on Windows 8


Not mine. Again, many THOUSANDS of tasks marked invalid. Across everyone, probably hundreds of thousands of tasks now.

ADMIN: Please comment.
____________
Dublin, California
Team: SETI.USA

Profile Daniel
Project administrator
Send message
Joined: 5 Mar 15
Posts: 73
Credit: 162,134
RAC: 0
Message 130 - Posted: 2 Apr 2015, 4:06:11 UTC - in response to Message 126.

The problem is affecting 1/3 of work units (8 thousand); see an example below. With an "in" file with a single line "171116789,500", the program will check that prime number and others up to 500 higher. The output file "out" should look like the one below but only with entries in that range for instance 171117181, not numbers like 140737447089616 or 7067552. We are not able to reproduce the bug on our computers although we are going to try with another one. It would help if someone who can reproduce the error could debug our program which is a short and simple one. To compile on Windows, use VC++ 10, replace in the boinc code replace samples/example_app/uc2.cpp with https://github.com/hmonroe/vdw/blob/master/VDWDaniel/src/vdwnumbersBOINC.cpp and compile.

7.4.25

171116789,500
140737447089616,7067552,7067440,7067456,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,171117181,
0,0,0,0,0,0,171116977,171117181,171117031,
0,0,0,0,171116809,171117031,171117161,171117109,0,
0,0,0,171117061,171117229,0,0,0,0,
0,0,0,171117181,0,0,0,0,0,
0,0,171117161,171117071,171116893,0,0,0,0,
0,0,171117229,0,0,0,0,0,0,
0,0,171117181,0,0,0,0,0,0,
0,171117229,0,0,0,0,0,0,0,
0,171117127,0,0,0,0,0,0,0,
0,171116839,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,
12:52:39 (28903): called boinc_finish(0)

daveandton
Send message
Joined: 29 Mar 15
Posts: 1
Credit: 0
RAC: 0
Message 131 - Posted: 2 Apr 2015, 9:56:53 UTC

Got to sort out the site if you want people to contribute. Looked not very carefully and attached with a Linux computer. wasted 3 days till I realised that they are not working because they are Microsoft WU's wasting my time. NOT IMPRESSED

zombie67 [MM]
Avatar
Send message
Joined: 9 Mar 15
Posts: 15
Credit: 1,283,842
RAC: 0
Message 132 - Posted: 2 Apr 2015, 12:13:49 UTC

Hi Daniel,

Until this particular issue is resolved, can you please turn on homogenous redundancy? This will at least prevent a linux machine trying to validate against a windows machine, resulting in a race with the winner going to a random third machine.

http://www.123numbers.org/workunit.php?wuid=549647

FWIW, the only difference here is the 5th line of the stderr. For windows machines:

0,0,0,0,0,0,0,0,0,

For linux machines:

140733343532288,7067552,7067440,7067456,0,0,0,0,0,

Which is correct in this example?
____________
Dublin, California
Team: SETI.USA

Profile Daniel
Project administrator
Send message
Joined: 5 Mar 15
Posts: 73
Credit: 162,134
RAC: 0
Message 133 - Posted: 2 Apr 2015, 15:01:55 UTC - in response to Message 132.

The first line should be all zeroes.

Homogeneous redundancy helps with numerical discrepancies. The problem is a bug that affects some Windows and Linux machines but not ours. We do not know in advance which machines. We need to fix that bug but do not have a machine that produces it.

http://boinc.berkeley.edu/trac/wiki/HomogeneousRedundancy

Profile UBT - Chris
Send message
Joined: 19 Mar 15
Posts: 8
Credit: 3,921
RAC: 0
Message 146 - Posted: 14 Apr 2015, 14:36:25 UTC

It looks like it's been corrected now - my Windows 7 laptop is crunching tasks on this project again :D
____________
Thanks, from UBT - Chris



Dr Who Fan
Avatar
Send message
Joined: 12 Mar 15
Posts: 18
Credit: 4,613
RAC: 0
Message 148 - Posted: 21 Apr 2015, 20:27:45 UTC - in response to Message 121.

ADMIN: What is the status/progress with the Windows XP VALIDATION PROBLEMS?
____________

Conan
Send message
Joined: 10 Mar 15
Posts: 8
Credit: 48,848
RAC: 0
Message 153 - Posted: 24 Apr 2015, 22:44:04 UTC

Nothing has changed here for me, despite the recent upgraded applications, version 300.00 is no different to previous versions.

Linux 64 Bit does not validate against anything (possibly works with another Linux host but I have not found any). I have seen one host of zombie67 with over 4,800 processed work units but only 10 points total credit as he must of managed to snag another Linux host and get credit for just 1 work unit, 1 out of 4,800 is a pretty dismal success rate.

Windows XP 32 Bit does not validate with anything either, makes no difference if it is against Linux - doesn't work, Windows 64 Bit - doesn't work.

The applications have always run OK on my computers but they just don't validate.

I will give it another break, but it is looking like a Windows 64 Bit only project at the moment.

Conan

Profile JayPi
Send message
Joined: 29 Mar 15
Posts: 1
Credit: 1,000,608
RAC: 0
Message 154 - Posted: 26 Apr 2015, 9:15:06 UTC

I have different Computers running 123numbers. Only one computer have problems with this project. 0 of 1000 WUs where validated. It is an Intel i7-3770K with Win 7 Professional. Other Computers with either i7-3770K or the same operating system are running fine. It seems that the installed software is responsible for this, for example the Windows SDK or somewhat else.

Andrew Dicker
Send message
Joined: 18 Mar 15
Posts: 2
Credit: 0
RAC: 0
Message 155 - Posted: 28 Apr 2015, 23:10:23 UTC

Win 7 64-bit here, kept up to date to within a fortnight or so with win updates. No 3rd party antivirus. Nothing valid.
http://www.123numbers.org/results.php?hostid=248

den777
Send message
Joined: 7 May 15
Posts: 1
Credit: 7,773
RAC: 0
Message 161 - Posted: 8 May 2015, 15:52:11 UTC

I waisted time with invalid results on my host too.

Okaaay. I downloaded https://github.com/hmonroe/vdw/blob/master/VDWDaniel/src/VDWDaniel.cpp and compiled it with VS 2013 as 64-bit app, and it produces adequate results in output file (without crazy big numbers out of range). I only added zeroing bestPrimes array and wrote my own max() function in code (it's not available in newest cstdlib) to make it working.

I tried to build 32-bit version, but it always failing while allocating ~2Gb of memory. I'm not a skilled C++ programmer and don't know what to do here.

So what can we do?
I think the root of evil is either 32-bitness or old VC++ of original app.

Profile Daniel
Project administrator
Send message
Joined: 5 Mar 15
Posts: 73
Credit: 162,134
RAC: 0
Message 216 - Posted: 7 Sep 2015, 14:36:11 UTC - in response to Message 161.

Version 31 seems to have fixed the computational error because it uses half as much memory. The problem with old version was that each copy of the program running took up about 2 gigabytes of RAM, given the size of primes that are being checked. Now each copy takes up 1 gigabyte. To check a prime near 250 million, the program has to create an array with 250 million entries.
____________
Daniel Monroe
vdwnumbers.org Project Administrator

Mumps [MM]
Send message
Joined: 3 Oct 15
Posts: 3
Credit: 7,344,612
RAC: 0
Message 235 - Posted: 4 Oct 2015, 14:48:13 UTC - in response to Message 124.
Last modified: 4 Oct 2015, 14:53:48 UTC

It looks like tasks crunched by both windows and linux will not validate together. Which means if the the third machine is windows, the linux machine loses out. And vice versa. Until this gets resolved, can homogeneous redundancy please be turned on?

http://boinc.berkeley.edu/trac/wiki/HomogeneousRedundancy



Bump. We are wasting many many thousands of tasks


Bump again. Status please?

Bumping yet again. Two days ago I reattached about 2 dozen Linux hosts and still not a single validation against a Windows host. Still seems that Linux has that first line that's not zeroed out when compared to Windows. As an example:

http://www.123numbers.org/workunit.php?wuid=1054942

linux wrote:
<stderr_txt>
331282559,250
140735943467744,7067552,7067440,7067456,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,
windows wrote:
<stderr_txt>
331282559,250
0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,

Previous · 1 · 2 · 3 · Next
Post to thread

Message boards : Number crunching : Validator problem?


Main page · Your account · Message boards


Code and content created by Daniel Monroe © 2019.