PrimeGrid
Please visit donation page to help the project cover running costs for this month

Join PrimeGrid

Returning Participants

Community

Leader Boards

Results

Other

drummers-lowrise

Advanced search

Message boards : Project Staging Area : upcoming problems k=27 Prime Search

Author Message
Michael BeckerProject donor
Avatar
Send message
Joined: 3 Apr 11
Posts: 40
ID: 93509
Credit: 205,777,194
RAC: 62,903
321 LLR Turquoise: Earned 5,000,000 credits (5,005,512)Cullen LLR Ruby: Earned 2,000,000 credits (2,748,467)ESP LLR Ruby: Earned 2,000,000 credits (2,801,533)Generalized Cullen/Woodall LLR Turquoise: Earned 5,000,000 credits (9,165,416)PPS LLR Turquoise: Earned 5,000,000 credits (5,005,087)PSP LLR Turquoise: Earned 5,000,000 credits (5,016,050)SoB LLR Ruby: Earned 2,000,000 credits (3,161,823)SR5 LLR Turquoise: Earned 5,000,000 credits (5,001,107)SGS LLR Ruby: Earned 2,000,000 credits (3,987,761)TRP LLR Turquoise: Earned 5,000,000 credits (5,109,367)Woodall LLR Ruby: Earned 2,000,000 credits (3,850,262)Cullen/Woodall Sieve (suspended) Ruby: Earned 2,000,000 credits (4,969,265)Generalized Cullen/Woodall Sieve Ruby: Earned 2,000,000 credits (2,000,094)PPS Sieve Emerald: Earned 50,000,000 credits (51,368,787)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Amethyst: Earned 1,000,000 credits (1,001,345)TRP Sieve (suspended) Ruby: Earned 2,000,000 credits (2,224,085)AP 26/27 Sapphire: Earned 20,000,000 credits (25,288,965)GFN Sapphire: Earned 20,000,000 credits (39,205,178)PSA Sapphire: Earned 20,000,000 credits (28,861,785)
Message 93241 - Posted: 17 Mar 2016 | 18:52:35 UTC

All my computer shows some error while computing 27*2^xxxxxxx-1
If they test 27*2^xxxxxxx+1 there are no errors reported.

[2016-03-16 23:58:01 MZ] 27: PRPNet server is version 5.4.0
k=27 Prime Search
Starting Lucas Lehmer Riesel prime test of 27*2^5930132-1
Using FFT length 336K, Pass1=448, Pass2=768
V1 = 5 ; Computing U0...done.
Iter: 4111249/5930131, ERROR: ROUND OFF (0.40625) > 0.4
Continuing from last save file.
Resuming LLR test of 27*2^5930132-1 at iteration 4057601 [68.42%]
Disregard last error. Result is reproducible and thus not a hardware problem.
For added safety, redoing iteration using a slower, more reliable method.
Continuing from last save file.
Resuming LLR test of 27*2^5930132-1 at iteration 4111249 [69.32%]
27*2^5930132-1 is not prime. LLR Res64: 3626CFDFF2AAC004 Time : 36873.613 sec.
[2016-03-17 10:18:49 MZ] 27: 27*2^5930132-1 is not prime. Residue 3626CFDFF2AAC004


Any idea?

Michael GoetzProject donor
Volunteer moderator
Project scientist
Avatar
Send message
Joined: 21 Jan 10
Posts: 9640
ID: 53948
Credit: 107,070,964
RAC: 95,795
321 LLR Amethyst: Earned 1,000,000 credits (1,169,719)Cullen LLR Amethyst: Earned 1,000,000 credits (1,157,331)ESP LLR Amethyst: Earned 1,000,000 credits (1,179,211)Generalized Cullen/Woodall LLR Amethyst: Earned 1,000,000 credits (1,330,821)PPS LLR Amethyst: Earned 1,000,000 credits (1,254,873)PSP LLR Ruby: Earned 2,000,000 credits (2,632,269)SoB LLR Ruby: Earned 2,000,000 credits (2,153,211)SR5 LLR Turquoise: Earned 5,000,000 credits (6,048,315)SGS LLR Amethyst: Earned 1,000,000 credits (1,680,461)TRP LLR Amethyst: Earned 1,000,000 credits (1,183,026)Woodall LLR Amethyst: Earned 1,000,000 credits (1,145,077)321 Sieve (suspended) Silver: Earned 100,000 credits (200,576)Cullen/Woodall Sieve (suspended) Ruby: Earned 2,000,000 credits (4,170,256)Generalized Cullen/Woodall Sieve Ruby: Earned 2,000,000 credits (2,085,723)PPS Sieve Jade: Earned 10,000,000 credits (18,175,834)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Amethyst: Earned 1,000,000 credits (1,035,522)TRP Sieve (suspended) Ruby: Earned 2,000,000 credits (2,051,121)AP 26/27 Turquoise: Earned 5,000,000 credits (5,881,239)GFN Sapphire: Earned 20,000,000 credits (42,507,589)PSA Jade: Earned 10,000,000 credits (10,028,791)
Message 93243 - Posted: 17 Mar 2016 | 19:42:01 UTC - in response to Message 93241.

All my computer shows some error while computing 27*2^xxxxxxx-1
If they test 27*2^xxxxxxx+1 there are no errors reported.

Any idea?


In this case, absolutely yes.

Last week, the PRPNet server reported that you had found a prime on the 27 port. The problem with PRPNet is that nothing there is double checked. When we checked the "prime", it was in fact composite.

Bottom line: your computer is malfunctioning and producing errors in the calculations. Because you're running on PRPNet, you see no indication of this because there's typically no check done on the results the computer returns. If the computer was running BOINC tasks, you would have seen immediately that many of the results were marked inconclusive or invalid. But on PRPNet, you're flying blind and typically neither you nor we would ever know there's anything wrong.

In this case, however, the false prime makes it obvious what's happening. Base 2 Reisel numbers, i.e., k*2^n-1, are prone to reporting false primes when the calculation has errors because errors sometimes produce a zero as the final result, and the algorithm used on these numbers returns a zero when the number is prime. So errors on this type of number sometimes cause false prime reports.

So, yeah, the computer is not working correctly. Overclocking, overheating, bad memory, etc., etc., etc. All the usual suspects. Something's wrong, but it's hard to say what it is without testing.
____________

Please do not PM me with support questions. They will usually go unanswered. Ask on the forums instead. Thank you!

Michael BeckerProject donor
Avatar
Send message
Joined: 3 Apr 11
Posts: 40
ID: 93509
Credit: 205,777,194
RAC: 62,903
321 LLR Turquoise: Earned 5,000,000 credits (5,005,512)Cullen LLR Ruby: Earned 2,000,000 credits (2,748,467)ESP LLR Ruby: Earned 2,000,000 credits (2,801,533)Generalized Cullen/Woodall LLR Turquoise: Earned 5,000,000 credits (9,165,416)PPS LLR Turquoise: Earned 5,000,000 credits (5,005,087)PSP LLR Turquoise: Earned 5,000,000 credits (5,016,050)SoB LLR Ruby: Earned 2,000,000 credits (3,161,823)SR5 LLR Turquoise: Earned 5,000,000 credits (5,001,107)SGS LLR Ruby: Earned 2,000,000 credits (3,987,761)TRP LLR Turquoise: Earned 5,000,000 credits (5,109,367)Woodall LLR Ruby: Earned 2,000,000 credits (3,850,262)Cullen/Woodall Sieve (suspended) Ruby: Earned 2,000,000 credits (4,969,265)Generalized Cullen/Woodall Sieve Ruby: Earned 2,000,000 credits (2,000,094)PPS Sieve Emerald: Earned 50,000,000 credits (51,368,787)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Amethyst: Earned 1,000,000 credits (1,001,345)TRP Sieve (suspended) Ruby: Earned 2,000,000 credits (2,224,085)AP 26/27 Sapphire: Earned 20,000,000 credits (25,288,965)GFN Sapphire: Earned 20,000,000 credits (39,205,178)PSA Sapphire: Earned 20,000,000 credits (28,861,785)
Message 93245 - Posted: 17 Mar 2016 | 20:26:00 UTC - in response to Message 93243.

i can believe that on of my computer have a problem, but not all my computer.
One i7-6700k, one i7-2600k and one i5-M480 and they all show errors while processing 27*2^xxxxxxx-1, and no errors while processing 27*2^xxxxxxx+1

Michael GoetzProject donor
Volunteer moderator
Project scientist
Avatar
Send message
Joined: 21 Jan 10
Posts: 9640
ID: 53948
Credit: 107,070,964
RAC: 95,795
321 LLR Amethyst: Earned 1,000,000 credits (1,169,719)Cullen LLR Amethyst: Earned 1,000,000 credits (1,157,331)ESP LLR Amethyst: Earned 1,000,000 credits (1,179,211)Generalized Cullen/Woodall LLR Amethyst: Earned 1,000,000 credits (1,330,821)PPS LLR Amethyst: Earned 1,000,000 credits (1,254,873)PSP LLR Ruby: Earned 2,000,000 credits (2,632,269)SoB LLR Ruby: Earned 2,000,000 credits (2,153,211)SR5 LLR Turquoise: Earned 5,000,000 credits (6,048,315)SGS LLR Amethyst: Earned 1,000,000 credits (1,680,461)TRP LLR Amethyst: Earned 1,000,000 credits (1,183,026)Woodall LLR Amethyst: Earned 1,000,000 credits (1,145,077)321 Sieve (suspended) Silver: Earned 100,000 credits (200,576)Cullen/Woodall Sieve (suspended) Ruby: Earned 2,000,000 credits (4,170,256)Generalized Cullen/Woodall Sieve Ruby: Earned 2,000,000 credits (2,085,723)PPS Sieve Jade: Earned 10,000,000 credits (18,175,834)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Amethyst: Earned 1,000,000 credits (1,035,522)TRP Sieve (suspended) Ruby: Earned 2,000,000 credits (2,051,121)AP 26/27 Turquoise: Earned 5,000,000 credits (5,881,239)GFN Sapphire: Earned 20,000,000 credits (42,507,589)PSA Jade: Earned 10,000,000 credits (10,028,791)
Message 93253 - Posted: 18 Mar 2016 | 0:04:12 UTC - in response to Message 93245.

i can believe that on of my computer have a problem, but not all my computer.
One i7-6700k, one i7-2600k and one i5-M480 and they all show errors while processing 27*2^xxxxxxx-1, and no errors while processing 27*2^xxxxxxx+1


Mia culpa; I didn't read your post carefully enough. I was so pleased to be able to actually let you know your computer needed attention that I didn't read much further.

Okay, multiple computers...

First of all the one having actual problems that you need to look into is the one using PRPNet with the machine id of "i7-6700". If you have more than one machine using that ID, it's the one that processed 27*2^5916818-1. If nothing else, I would put that machine on one of the BOINC projects that use double checking, i.e., don't use PRPNet, and don't run SGS, PPS, or PPSE. You want all of that machine's tasks checked by another computer so you can get an idea of how often it's producing errors.

Now that that's out of the way,

Iter: 4111249/5930131, ERROR: ROUND OFF (0.40625) > 0.4
Continuing from last save file.
Resuming LLR test of 27*2^5930132-1 at iteration 4057601 [68.42%]
Disregard last error. Result is reproducible and thus not a hardware problem.
For added safety, redoing iteration using a slower, more reliable method.


tl;dr: That's normal LLR activity. Nothing to be concerned about.

LLR tries to use the smallest FFT that it can because a smaller FFT is faster. Choosing the FFT size isn't an exact science, however, and sometimes it guesses wrong. If it uses an FFT that's too big, you never know about it. It could have used a faster FFT, but neither the program nor you will ever know. If it chooses an FFT that's too small, the internal error detection sees the round off errors getting too large.

When that happens, LLR needs to figure out if it simply picked an FFT that's too small, or if the computer is having hardware problems. It will back up to the previous checkpoint and restart from there (the calculation was still good at that point.) If the error doesn't occur again, i.e., it's not reproducible, then the computer has a hardware problem. The calculation might still complete -- and even complete correctly (this time), but the computer isn't working right. On the other hand, if the exact same error happens again, then the computer is working and the FFT is just too small. LLR then adapts its algorithm to account for this and continues on to a successful conclusion.
____________

Please do not PM me with support questions. They will usually go unanswered. Ask on the forums instead. Thank you!

Michael BeckerProject donor
Avatar
Send message
Joined: 3 Apr 11
Posts: 40
ID: 93509
Credit: 205,777,194
RAC: 62,903
321 LLR Turquoise: Earned 5,000,000 credits (5,005,512)Cullen LLR Ruby: Earned 2,000,000 credits (2,748,467)ESP LLR Ruby: Earned 2,000,000 credits (2,801,533)Generalized Cullen/Woodall LLR Turquoise: Earned 5,000,000 credits (9,165,416)PPS LLR Turquoise: Earned 5,000,000 credits (5,005,087)PSP LLR Turquoise: Earned 5,000,000 credits (5,016,050)SoB LLR Ruby: Earned 2,000,000 credits (3,161,823)SR5 LLR Turquoise: Earned 5,000,000 credits (5,001,107)SGS LLR Ruby: Earned 2,000,000 credits (3,987,761)TRP LLR Turquoise: Earned 5,000,000 credits (5,109,367)Woodall LLR Ruby: Earned 2,000,000 credits (3,850,262)Cullen/Woodall Sieve (suspended) Ruby: Earned 2,000,000 credits (4,969,265)Generalized Cullen/Woodall Sieve Ruby: Earned 2,000,000 credits (2,000,094)PPS Sieve Emerald: Earned 50,000,000 credits (51,368,787)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Amethyst: Earned 1,000,000 credits (1,001,345)TRP Sieve (suspended) Ruby: Earned 2,000,000 credits (2,224,085)AP 26/27 Sapphire: Earned 20,000,000 credits (25,288,965)GFN Sapphire: Earned 20,000,000 credits (39,205,178)PSA Sapphire: Earned 20,000,000 credits (28,861,785)
Message 93271 - Posted: 18 Mar 2016 | 18:25:28 UTC - in response to Message 93253.

Thank you for the information,
i didn't know the thing with the fft-size, and that the program sometimes need to switch to a larger fft.
My machine 'i7-6700' crunch SGS without problems, but bigger wu's (SOB and 27) are invalid.
Maybe there is really a memory problem, and i will perform a RAM-test as soon as possible.
If there are errors detected, i will try to adjust the timings.

best regards

mackerelProject donor
Volunteer tester
Avatar
Send message
Joined: 2 Oct 08
Posts: 1681
ID: 29980
Credit: 192,590,292
RAC: 193,022
321 LLR Turquoise: Earned 5,000,000 credits (5,191,713)Cullen LLR Ruby: Earned 2,000,000 credits (4,619,611)ESP LLR Ruby: Earned 2,000,000 credits (3,725,865)Generalized Cullen/Woodall LLR Ruby: Earned 2,000,000 credits (3,220,326)PPS LLR Sapphire: Earned 20,000,000 credits (26,374,738)PSP LLR Jade: Earned 10,000,000 credits (10,128,604)SoB LLR Ruby: Earned 2,000,000 credits (2,643,659)SR5 LLR Jade: Earned 10,000,000 credits (12,262,110)SGS LLR Turquoise: Earned 5,000,000 credits (6,702,766)TPS LLR (retired) Bronze: Earned 10,000 credits (34,130)TRP LLR Turquoise: Earned 5,000,000 credits (7,400,028)Woodall LLR Ruby: Earned 2,000,000 credits (2,471,793)321 Sieve (suspended) Silver: Earned 100,000 credits (203,937)Cullen/Woodall Sieve (suspended) Turquoise: Earned 5,000,000 credits (5,383,853)Generalized Cullen/Woodall Sieve Sapphire: Earned 20,000,000 credits (20,070,475)PPS Sieve Sapphire: Earned 20,000,000 credits (39,662,287)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Ruby: Earned 2,000,000 credits (2,293,882)TRP Sieve (suspended) Turquoise: Earned 5,000,000 credits (5,012,757)AP 26/27 Jade: Earned 10,000,000 credits (12,231,866)GFN Sapphire: Earned 20,000,000 credits (20,017,852)PSA Ruby: Earned 2,000,000 credits (2,939,755)
Message 93272 - Posted: 18 Mar 2016 | 18:42:07 UTC - in response to Message 93271.

My machine 'i7-6700' crunch SGS without problems, but bigger wu's (SOB and 27) are invalid.
Maybe there is really a memory problem, and i will perform a RAM-test as soon as possible.
If there are errors detected, i will try to adjust the timings.


That would be consistent with memory related problems. SGS units are small enough to sit inside the processor cache so wont really hit ram, whereas bigger units will need to access ram throughout.

I had assorted problems with my i7-6700k and other Skylake processors early on, so I have been there. I can try to help as needed. It mostly boils down to getting the latest mobo bios, not necessarily picking all the most aggressive settings, and a bit of luck especially with higher speed modules.

Michael, is there some checking rate even with small units to ensure the machine doesn't go bad? E.g. is every xxxx units double checked even on a trusted machine?

Michael BeckerProject donor
Avatar
Send message
Joined: 3 Apr 11
Posts: 40
ID: 93509
Credit: 205,777,194
RAC: 62,903
321 LLR Turquoise: Earned 5,000,000 credits (5,005,512)Cullen LLR Ruby: Earned 2,000,000 credits (2,748,467)ESP LLR Ruby: Earned 2,000,000 credits (2,801,533)Generalized Cullen/Woodall LLR Turquoise: Earned 5,000,000 credits (9,165,416)PPS LLR Turquoise: Earned 5,000,000 credits (5,005,087)PSP LLR Turquoise: Earned 5,000,000 credits (5,016,050)SoB LLR Ruby: Earned 2,000,000 credits (3,161,823)SR5 LLR Turquoise: Earned 5,000,000 credits (5,001,107)SGS LLR Ruby: Earned 2,000,000 credits (3,987,761)TRP LLR Turquoise: Earned 5,000,000 credits (5,109,367)Woodall LLR Ruby: Earned 2,000,000 credits (3,850,262)Cullen/Woodall Sieve (suspended) Ruby: Earned 2,000,000 credits (4,969,265)Generalized Cullen/Woodall Sieve Ruby: Earned 2,000,000 credits (2,000,094)PPS Sieve Emerald: Earned 50,000,000 credits (51,368,787)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Amethyst: Earned 1,000,000 credits (1,001,345)TRP Sieve (suspended) Ruby: Earned 2,000,000 credits (2,224,085)AP 26/27 Sapphire: Earned 20,000,000 credits (25,288,965)GFN Sapphire: Earned 20,000,000 credits (39,205,178)PSA Sapphire: Earned 20,000,000 credits (28,861,785)
Message 93273 - Posted: 18 Mar 2016 | 19:52:49 UTC - in response to Message 93272.
Last modified: 18 Mar 2016 | 20:35:54 UTC

Michael, is there some checking rate even with small units to ensure the machine doesn't go bad? E.g. is every xxxx units double checked even on a trusted machine?


this machine: hostid=505857
crunching since Dec.30.2015, no invalid SGS, 3? invalid PPS, and 2 invalid SOB (+1 i expect)
Mobo is Gigabyte Z170-HD3P, updated to latest bios version F5 last sunday.
Yesterday i set Memory Settings to 'enhanced stability, and i also try to perform the RAM-test by using a Ubuntu Live USB-stick, but i can't enter the bootmenu because of the Grub-settings. Press and hold <shift key> didn't help, Live-linux will start automatically. I will modify grub when i have a bit more time.

i keep you informed

edit:
Memory is: G.Skill RipJaws V DDR4-3200 DIMM CL16-16-16-36

Profile Rafael
Volunteer tester
Avatar
Send message
Joined: 22 Oct 14
Posts: 733
ID: 370496
Credit: 166,388,542
RAC: 540,333
321 LLR Ruby: Earned 2,000,000 credits (4,015,911)Cullen LLR Ruby: Earned 2,000,000 credits (2,267,883)ESP LLR Ruby: Earned 2,000,000 credits (2,331,371)Generalized Cullen/Woodall LLR Ruby: Earned 2,000,000 credits (2,349,241)PPS LLR Ruby: Earned 2,000,000 credits (2,039,006)PSP LLR Ruby: Earned 2,000,000 credits (2,443,859)SoB LLR Ruby: Earned 2,000,000 credits (2,348,456)SR5 LLR Ruby: Earned 2,000,000 credits (2,297,757)SGS LLR Ruby: Earned 2,000,000 credits (2,897,962)TRP LLR Ruby: Earned 2,000,000 credits (3,023,786)Woodall LLR Ruby: Earned 2,000,000 credits (2,234,383)Generalized Cullen/Woodall Sieve Ruby: Earned 2,000,000 credits (3,291,694)PPS Sieve Turquoise: Earned 5,000,000 credits (5,201,453)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Ruby: Earned 2,000,000 credits (2,000,053)TRP Sieve (suspended) Ruby: Earned 2,000,000 credits (2,030,160)AP 26/27 Ruby: Earned 2,000,000 credits (2,433,886)GFN Sapphire: Earned 20,000,000 credits (21,211,034)PSA Double Bronze: Earned 100,000,000 credits (101,970,647)
Message 93274 - Posted: 18 Mar 2016 | 21:11:25 UTC - in response to Message 93273.

Memory is: G.Skill RipJaws V DDR4-3200 DIMM CL16-16-16-36


Mine takes 17-18-18-38 at 1.38v to get 3200 stable (or at least I think so, as everything up to SoB, which I just started testing now, showed no problems). Now, my kit is only rated for 3000 15-17-17-36, but still....

Michael GoetzProject donor
Volunteer moderator
Project scientist
Avatar
Send message
Joined: 21 Jan 10
Posts: 9640
ID: 53948
Credit: 107,070,964
RAC: 95,795
321 LLR Amethyst: Earned 1,000,000 credits (1,169,719)Cullen LLR Amethyst: Earned 1,000,000 credits (1,157,331)ESP LLR Amethyst: Earned 1,000,000 credits (1,179,211)Generalized Cullen/Woodall LLR Amethyst: Earned 1,000,000 credits (1,330,821)PPS LLR Amethyst: Earned 1,000,000 credits (1,254,873)PSP LLR Ruby: Earned 2,000,000 credits (2,632,269)SoB LLR Ruby: Earned 2,000,000 credits (2,153,211)SR5 LLR Turquoise: Earned 5,000,000 credits (6,048,315)SGS LLR Amethyst: Earned 1,000,000 credits (1,680,461)TRP LLR Amethyst: Earned 1,000,000 credits (1,183,026)Woodall LLR Amethyst: Earned 1,000,000 credits (1,145,077)321 Sieve (suspended) Silver: Earned 100,000 credits (200,576)Cullen/Woodall Sieve (suspended) Ruby: Earned 2,000,000 credits (4,170,256)Generalized Cullen/Woodall Sieve Ruby: Earned 2,000,000 credits (2,085,723)PPS Sieve Jade: Earned 10,000,000 credits (18,175,834)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Amethyst: Earned 1,000,000 credits (1,035,522)TRP Sieve (suspended) Ruby: Earned 2,000,000 credits (2,051,121)AP 26/27 Turquoise: Earned 5,000,000 credits (5,881,239)GFN Sapphire: Earned 20,000,000 credits (42,507,589)PSA Jade: Earned 10,000,000 credits (10,028,791)
Message 93277 - Posted: 18 Mar 2016 | 21:40:36 UTC - in response to Message 93272.
Last modified: 18 Mar 2016 | 21:41:21 UTC

Michael, is there some checking rate even with small units to ensure the machine doesn't go bad? E.g. is every xxxx units double checked even on a trusted machine?


On SGS, PPS, and PPSE, no. And nothing at all is double checked on PRPNet.
____________

Please do not PM me with support questions. They will usually go unanswered. Ask on the forums instead. Thank you!

Michael BeckerProject donor
Avatar
Send message
Joined: 3 Apr 11
Posts: 40
ID: 93509
Credit: 205,777,194
RAC: 62,903
321 LLR Turquoise: Earned 5,000,000 credits (5,005,512)Cullen LLR Ruby: Earned 2,000,000 credits (2,748,467)ESP LLR Ruby: Earned 2,000,000 credits (2,801,533)Generalized Cullen/Woodall LLR Turquoise: Earned 5,000,000 credits (9,165,416)PPS LLR Turquoise: Earned 5,000,000 credits (5,005,087)PSP LLR Turquoise: Earned 5,000,000 credits (5,016,050)SoB LLR Ruby: Earned 2,000,000 credits (3,161,823)SR5 LLR Turquoise: Earned 5,000,000 credits (5,001,107)SGS LLR Ruby: Earned 2,000,000 credits (3,987,761)TRP LLR Turquoise: Earned 5,000,000 credits (5,109,367)Woodall LLR Ruby: Earned 2,000,000 credits (3,850,262)Cullen/Woodall Sieve (suspended) Ruby: Earned 2,000,000 credits (4,969,265)Generalized Cullen/Woodall Sieve Ruby: Earned 2,000,000 credits (2,000,094)PPS Sieve Emerald: Earned 50,000,000 credits (51,368,787)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Amethyst: Earned 1,000,000 credits (1,001,345)TRP Sieve (suspended) Ruby: Earned 2,000,000 credits (2,224,085)AP 26/27 Sapphire: Earned 20,000,000 credits (25,288,965)GFN Sapphire: Earned 20,000,000 credits (39,205,178)PSA Sapphire: Earned 20,000,000 credits (28,861,785)
Message 93284 - Posted: 18 Mar 2016 | 23:06:54 UTC - in response to Message 93277.

Michael, is there some checking rate even with small units to ensure the machine doesn't go bad? E.g. is every xxxx units double checked even on a trusted machine?

My computer 'i7-6700' on PRPNet is also hostid=505857 on Primegrid. On Primegrid nearly everything is positive checked, only three PPS and two SOB were invalid. So i know anything went wrong with this machine.
I will crunch big wu's only if i can't detect problems with my Hardware.
Maybe the bios update has fixed a problem.

Just finished MemTest86, no problems detected.
Then i checked Bios settings, an saw that Ram-clock was 2133 (to slow ?)
Changed clock to 3200 and re run MemTest86, again no problems detected.
Now i perform a torture test with prime95, it runs 1hour without problem, everything 'passed!'. I let it run the whole night and see whats up tomorrow morning.
Coretemperature is max 78C. Coolers are clean.SSD is 25C and Case is 28C. Nothing to hot!?
I've downloaded the newest version of MemTest86, and will repeat the test.

ardoProject donor
Avatar
Send message
Joined: 12 Dec 10
Posts: 168
ID: 76659
Credit: 1,688,014,588
RAC: 0
321 LLR Jade: Earned 10,000,000 credits (10,013,825)Cullen LLR Jade: Earned 10,000,000 credits (10,019,431)ESP LLR Emerald: Earned 50,000,000 credits (50,206,403)PPS LLR Jade: Earned 10,000,000 credits (10,012,367)PSP LLR Sapphire: Earned 20,000,000 credits (47,049,813)SoB LLR Sapphire: Earned 20,000,000 credits (32,579,834)SR5 LLR Jade: Earned 10,000,000 credits (10,029,121)SGS LLR Jade: Earned 10,000,000 credits (10,012,386)TRP LLR Jade: Earned 10,000,000 credits (10,081,730)Woodall LLR Jade: Earned 10,000,000 credits (10,028,972)321 Sieve (suspended) Ruby: Earned 2,000,000 credits (2,090,586)Cullen/Woodall Sieve (suspended) Double Bronze: Earned 100,000,000 credits (141,181,073)Generalized Cullen/Woodall Sieve Ruby: Earned 2,000,000 credits (2,055,101)PPS Sieve Double Silver: Earned 200,000,000 credits (200,169,481)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Emerald: Earned 50,000,000 credits (50,015,203)TRP Sieve (suspended) Sapphire: Earned 20,000,000 credits (20,245,280)AP 26/27 Jade: Earned 10,000,000 credits (10,002,382)GFN Emerald: Earned 50,000,000 credits (60,038,478)PSA Double Amethyst: Earned 1,000,000,000 credits (1,002,183,124)
Message 93285 - Posted: 18 Mar 2016 | 23:14:30 UTC

I would suggest running "stressapptest" as it does a lot better job at detecting memory issues.
____________
Badge score: 2*5 + 8*7 + 3*8 + 3*9 + 1*10 + 1*11 + 1*13 = 151

mackerelProject donor
Volunteer tester
Avatar
Send message
Joined: 2 Oct 08
Posts: 1681
ID: 29980
Credit: 192,590,292
RAC: 193,022
321 LLR Turquoise: Earned 5,000,000 credits (5,191,713)Cullen LLR Ruby: Earned 2,000,000 credits (4,619,611)ESP LLR Ruby: Earned 2,000,000 credits (3,725,865)Generalized Cullen/Woodall LLR Ruby: Earned 2,000,000 credits (3,220,326)PPS LLR Sapphire: Earned 20,000,000 credits (26,374,738)PSP LLR Jade: Earned 10,000,000 credits (10,128,604)SoB LLR Ruby: Earned 2,000,000 credits (2,643,659)SR5 LLR Jade: Earned 10,000,000 credits (12,262,110)SGS LLR Turquoise: Earned 5,000,000 credits (6,702,766)TPS LLR (retired) Bronze: Earned 10,000 credits (34,130)TRP LLR Turquoise: Earned 5,000,000 credits (7,400,028)Woodall LLR Ruby: Earned 2,000,000 credits (2,471,793)321 Sieve (suspended) Silver: Earned 100,000 credits (203,937)Cullen/Woodall Sieve (suspended) Turquoise: Earned 5,000,000 credits (5,383,853)Generalized Cullen/Woodall Sieve Sapphire: Earned 20,000,000 credits (20,070,475)PPS Sieve Sapphire: Earned 20,000,000 credits (39,662,287)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Ruby: Earned 2,000,000 credits (2,293,882)TRP Sieve (suspended) Turquoise: Earned 5,000,000 credits (5,012,757)AP 26/27 Jade: Earned 10,000,000 credits (12,231,866)GFN Sapphire: Earned 20,000,000 credits (20,017,852)PSA Ruby: Earned 2,000,000 credits (2,939,755)
Message 93290 - Posted: 18 Mar 2016 | 23:51:25 UTC - in response to Message 93273.

Oops, didn't notice earlier there were two active Michaels on this thread. Apologies if that caused any confusion.

On SGS, PPS, and PPSE, no. And nothing at all is double checked on PRPNet.


If I understood previous statements correctly, and that they still apply, there is double checking until the computer gets trusted as error free. So once the machine achieves this status, there is no checking at all on the small units? So if it were to go bad, primes could be missed? Alternatively, errors could only be detected when running bigger units?

Memory is: G.Skill RipJaws V DDR4-3200 DIMM CL16-16-16-36


That might be the same ram I'm using, although can't be sure without exact part number.

Let me see if I got this straight, when the ram was running at 2133, there were errors in units? I'd be surprised if ram, even if rated at higher speed, couldn't run at 2133 stably unless something was seriously wrong with it. Given this, I'd look towards the CPU as the possible problem area. When running 4 cores active, what clock and voltage is it using? There are various tools that could be used. I like HWINFO64 for windows. My possible concern here is that motherboard manufacturers like to do a little turbo-overclocking. For example, the i7-6700k is 4 GHz base, 4.2 GHz turbo when single core is active. A lot of motherboards will run it at 4.2 GHz all cores active for example, and jack up the voltage. It is pretty safe, but not really 100% guaranteed. If this is happening, I'd suggest turning off turbo (it might be called enhanced turbo or something else too) and make sure it is run at 4 GHz default voltages and see if that helps.

Having said all that, Prime95 isn't that different from many of the prime finding workloads here, so if Prime95 looks ok, prime finding here should be ok.

Michael GoetzProject donor
Volunteer moderator
Project scientist
Avatar
Send message
Joined: 21 Jan 10
Posts: 9640
ID: 53948
Credit: 107,070,964
RAC: 95,795
321 LLR Amethyst: Earned 1,000,000 credits (1,169,719)Cullen LLR Amethyst: Earned 1,000,000 credits (1,157,331)ESP LLR Amethyst: Earned 1,000,000 credits (1,179,211)Generalized Cullen/Woodall LLR Amethyst: Earned 1,000,000 credits (1,330,821)PPS LLR Amethyst: Earned 1,000,000 credits (1,254,873)PSP LLR Ruby: Earned 2,000,000 credits (2,632,269)SoB LLR Ruby: Earned 2,000,000 credits (2,153,211)SR5 LLR Turquoise: Earned 5,000,000 credits (6,048,315)SGS LLR Amethyst: Earned 1,000,000 credits (1,680,461)TRP LLR Amethyst: Earned 1,000,000 credits (1,183,026)Woodall LLR Amethyst: Earned 1,000,000 credits (1,145,077)321 Sieve (suspended) Silver: Earned 100,000 credits (200,576)Cullen/Woodall Sieve (suspended) Ruby: Earned 2,000,000 credits (4,170,256)Generalized Cullen/Woodall Sieve Ruby: Earned 2,000,000 credits (2,085,723)PPS Sieve Jade: Earned 10,000,000 credits (18,175,834)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Amethyst: Earned 1,000,000 credits (1,035,522)TRP Sieve (suspended) Ruby: Earned 2,000,000 credits (2,051,121)AP 26/27 Turquoise: Earned 5,000,000 credits (5,881,239)GFN Sapphire: Earned 20,000,000 credits (42,507,589)PSA Jade: Earned 10,000,000 credits (10,028,791)
Message 93292 - Posted: 19 Mar 2016 | 0:10:03 UTC - in response to Message 93290.

So if it were to go bad, primes could be missed?


Duh.

It's a near certainty primes have been, and will continue to be missed.

With SGS, PPS, and PPSE, sometimes you'll be the wingman to someone else who must have double checking because the host isn't trusted, so not all of your tasks will go unchecked. But if the computer is only making errors occasionally, it could go unnoticed for an extended period of time.

I'll be brutally honest: any group, project, or individual that doesn't double check their work is doing this for the thrill of finding primes and is only paying lip service to "wanting to do it right."

Whether the double check happens simultaneously, or later on, everything should be double checked. (Although it's VERY advantageous to participants if the double checking occurs simultaneously because you get immediate feedback when a computer goes bad. Finding out 10 years later that your computer was having trouble -- you know, the computer you tossed in the trash 8 years ago -- does nobody any good.)

The only reason we don't require SGS, PPS, and PPSE to be double checked is because historically PG hasn't done that and a lot of people like running without double checking so that almost always they'll be the prime finder. Every now and then we have internal discussions about how nobody wants to talk about the elephant in the room and come to grips with the fact that we really should just bite the bullet and make it mandatory on all the projects.

With PRPNet, double checking is problematic, which is why we don't use it. It's also part of the reason that so many PRPNet ports have been moved to BOINC. (The much more important reason is that BOINC will bring much more participation, of course.)

____________

Please do not PM me with support questions. They will usually go unanswered. Ask on the forums instead. Thank you!

mackerelProject donor
Volunteer tester
Avatar
Send message
Joined: 2 Oct 08
Posts: 1681
ID: 29980
Credit: 192,590,292
RAC: 193,022
321 LLR Turquoise: Earned 5,000,000 credits (5,191,713)Cullen LLR Ruby: Earned 2,000,000 credits (4,619,611)ESP LLR Ruby: Earned 2,000,000 credits (3,725,865)Generalized Cullen/Woodall LLR Ruby: Earned 2,000,000 credits (3,220,326)PPS LLR Sapphire: Earned 20,000,000 credits (26,374,738)PSP LLR Jade: Earned 10,000,000 credits (10,128,604)SoB LLR Ruby: Earned 2,000,000 credits (2,643,659)SR5 LLR Jade: Earned 10,000,000 credits (12,262,110)SGS LLR Turquoise: Earned 5,000,000 credits (6,702,766)TPS LLR (retired) Bronze: Earned 10,000 credits (34,130)TRP LLR Turquoise: Earned 5,000,000 credits (7,400,028)Woodall LLR Ruby: Earned 2,000,000 credits (2,471,793)321 Sieve (suspended) Silver: Earned 100,000 credits (203,937)Cullen/Woodall Sieve (suspended) Turquoise: Earned 5,000,000 credits (5,383,853)Generalized Cullen/Woodall Sieve Sapphire: Earned 20,000,000 credits (20,070,475)PPS Sieve Sapphire: Earned 20,000,000 credits (39,662,287)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Ruby: Earned 2,000,000 credits (2,293,882)TRP Sieve (suspended) Turquoise: Earned 5,000,000 credits (5,012,757)AP 26/27 Jade: Earned 10,000,000 credits (12,231,866)GFN Sapphire: Earned 20,000,000 credits (20,017,852)PSA Ruby: Earned 2,000,000 credits (2,939,755)
Message 93298 - Posted: 19 Mar 2016 | 1:30:01 UTC - in response to Message 93292.

Thanks for the explanation. I'm not personally bothered by double-check or not, but for some reason I thought adaptive replication (if that's the right name) was activated. I thought that would do selective double checking along the way hence my question on how often that happens. Or is that the thing as described anyway?

If I'm not mistaken, if we go back some years this wasn't always the case and double checking was always active. I'm not really sure on this, but I was more often the wingman on my early days and it was some time before I heard of the zero cache trick.

As an idea, this is likely outside the scope of boinc, but may be applicable to PRPnet: I like the effectively two project approach to 1st pass and double check. That is, there's a queue of work for 1st past testing people can choose from, and a 2nd queue for double check. This was the arrangement at rieselsieve while that was still around. 1st pass drew more participants hoping to find that big new prime, so the double check was usually much further behind. But that meant smaller units. During that project I actually found my 1st big prime in 1st pass work, and later I got a smaller but not insignificant one in double check work.

JimBProject donor
Volunteer moderator
Project developer
Project scientist
Send message
Joined: 4 Aug 11
Posts: 647
ID: 107307
Credit: 452,819,188
RAC: 455,164
321 LLR Ruby: Earned 2,000,000 credits (2,014,631)Cullen LLR Ruby: Earned 2,000,000 credits (2,041,185)ESP LLR Ruby: Earned 2,000,000 credits (2,060,837)Generalized Cullen/Woodall LLR Ruby: Earned 2,000,000 credits (2,006,814)PPS LLR Ruby: Earned 2,000,000 credits (2,052,020)PSP LLR Ruby: Earned 2,000,000 credits (3,617,961)SoB LLR Turquoise: Earned 5,000,000 credits (8,423,909)SR5 LLR Turquoise: Earned 5,000,000 credits (6,328,868)SGS LLR Ruby: Earned 2,000,000 credits (2,000,455)TRP LLR Ruby: Earned 2,000,000 credits (2,068,214)Woodall LLR Ruby: Earned 2,000,000 credits (2,082,879)Cullen/Woodall Sieve (suspended) Ruby: Earned 2,000,000 credits (4,002,919)Generalized Cullen/Woodall Sieve Turquoise: Earned 5,000,000 credits (6,177,238)PPS Sieve Sapphire: Earned 20,000,000 credits (27,326,793)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Ruby: Earned 2,000,000 credits (2,341,676)TRP Sieve (suspended) Ruby: Earned 2,000,000 credits (2,070,804)AP 26/27 Ruby: Earned 2,000,000 credits (2,587,520)GFN Sapphire: Earned 20,000,000 credits (26,012,537)PSA Double Silver: Earned 200,000,000 credits (347,601,926)
Message 93300 - Posted: 19 Mar 2016 | 1:50:57 UTC

Yes, adaptive replication is on for SGS, PPS and PPSE. But once a computer passes muster and is getting mostly unreplicated jobs, it may take a while for subsequent errors to be noticed.

I don't believe we were ever doublechecking those three subprojects. In fact in the early days I think adaptive replication might have been used for conjectures. That's why we've had times when we doublechecked everything that wasn't DC'd before. Other DCs were done when moving subprojects from PRPNet to BOINC. Today all conjecture subprojects always require two matching results.

Michael GoetzProject donor
Volunteer moderator
Project scientist
Avatar
Send message
Joined: 21 Jan 10
Posts: 9640
ID: 53948
Credit: 107,070,964
RAC: 95,795
321 LLR Amethyst: Earned 1,000,000 credits (1,169,719)Cullen LLR Amethyst: Earned 1,000,000 credits (1,157,331)ESP LLR Amethyst: Earned 1,000,000 credits (1,179,211)Generalized Cullen/Woodall LLR Amethyst: Earned 1,000,000 credits (1,330,821)PPS LLR Amethyst: Earned 1,000,000 credits (1,254,873)PSP LLR Ruby: Earned 2,000,000 credits (2,632,269)SoB LLR Ruby: Earned 2,000,000 credits (2,153,211)SR5 LLR Turquoise: Earned 5,000,000 credits (6,048,315)SGS LLR Amethyst: Earned 1,000,000 credits (1,680,461)TRP LLR Amethyst: Earned 1,000,000 credits (1,183,026)Woodall LLR Amethyst: Earned 1,000,000 credits (1,145,077)321 Sieve (suspended) Silver: Earned 100,000 credits (200,576)Cullen/Woodall Sieve (suspended) Ruby: Earned 2,000,000 credits (4,170,256)Generalized Cullen/Woodall Sieve Ruby: Earned 2,000,000 credits (2,085,723)PPS Sieve Jade: Earned 10,000,000 credits (18,175,834)Sierpinski (ESP/PSP/SoB) Sieve (suspended) Amethyst: Earned 1,000,000 credits (1,035,522)TRP Sieve (suspended) Ruby: Earned 2,000,000 credits (2,051,121)AP 26/27 Turquoise: Earned 5,000,000 credits (5,881,239)GFN Sapphire: Earned 20,000,000 credits (42,507,589)PSA Jade: Earned 10,000,000 credits (10,028,791)
Message 93307 - Posted: 19 Mar 2016 | 11:44:50 UTC - in response to Message 93298.

If I'm not mistaken, if we go back some years this wasn't always the case and double checking was always active. I'm not really sure on this, but I was more often the wingman on my early days and it was some time before I heard of the zero cache trick.


You've got it backwards, as Jim stated. More stuff used to NOT be double checked. Jim and I are responsible for more of the projects being double checked than in the past.

If you're specifically running PPS so you'll always be the prime finder, and you miss a prime because of a computer problem, well, that's your prime that's lost. It bothers us, true, but it's a bigger loss for the participant.

In a conjecture project, however, it's a totally different story. Consider a hypothetical project where k=123456 still needs to be eliminated. Let's say there are actually two primes for this k, at n = 123,456 and the second at n=1,234,567,890. If the first prime is missed by a faulty computer, a ridiculous amount of unnecessary computation will be needed to find that second prime at such a high N.

All of the conjectures at PrimeGrid are now double checked. That wasn't always the case, and it's certainly not the case (or at least inconsistent) with many of the others out there. (NOTE: For PSP and SoB, however, the work done on those projects outside of PrimeGrid isn't fully double checked yet.)

In recent years, as we've moved projects from elsewhere onto BOINC, we've done full double checks of prior work. No missed primes were found in the conjectures, but we did find 8 overlooked GFN primes.
____________

Please do not PM me with support questions. They will usually go unanswered. Ask on the forums instead. Thank you!

Post to thread

Message boards : Project Staging Area : upcoming problems k=27 Prime Search

[Return to PrimeGrid main page]
Copyright © 2005 - 2017 Rytis Slatkevičius (contact) and PrimeGrid community. Server load 1.69, 2.96, 3.35
Generated 23 Nov 2017 | 9:17:03 UTC