The PARI stack overflows !

Message boards : Number crunching : The PARI stack overflows !
Message board moderation

To post messages, you must log in.

AuthorMessage
Richard Haselgrove

Send message
Joined: 28 Oct 11
Posts: 179
Credit: 222,898,150
RAC: 128,996
Message 683 - Posted: 8 Jul 2012, 14:49:40 UTC

Eric,

You might like to have a look at workunit 1022542. 7 people have attempted it so far, all ending with the same error. I've got copy number eight - any parameter I can tweak to give it the extra stack space it's asking for?

N2 = 25.
*** the PARI stack overflows !
current stack size: 150000000 (143.051 Mbytes)
[hint] you can increase GP stack with allocatemem()

*** Error in the PARI system. End of program
ID: 683 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Eric Driver
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 8 Jul 11
Posts: 1320
Credit: 408,748,050
RAC: 257,145
Message 684 - Posted: 8 Jul 2012, 21:00:55 UTC - in response to Message 683.  

Thanks for reporting. Your best bet at this time is to abort that wu (I also just cancelled it so it wont be re-sent).

I noticed this problem about a week ago. It seems to happen with about 1 out of every 1000 wus. As it turns out, increasing the stack size doesn't help. I've tracked the problem down to the factoring algorithm. If I use a more robust algorithm it fixes the problem but the wus take about 2 to 3 times longer to complete. Right now I'd rather put up with a few bad wus than take the hit in processing time.

So my strategy in the short term is to try and find these bad wus early and cancel them before they bounce around between too many hosts. In the meantime, I am trying to find a better work around. And once I understand the problem better I will report it to the folks at pari.
ID: 684 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Richard Haselgrove

Send message
Joined: 28 Oct 11
Posts: 179
Credit: 222,898,150
RAC: 128,996
Message 685 - Posted: 8 Jul 2012, 22:45:25 UTC - in response to Message 684.  

Thanks for reporting. Your best bet at this time is to abort that wu (I also just cancelled it so it wont be re-sent).

Done. I'll let you know if I see any others with high replication numbers.
ID: 685 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
ritterm
Avatar

Send message
Joined: 15 Apr 12
Posts: 3
Credit: 1,032,575
RAC: 0
Message 721 - Posted: 2 Oct 2012, 11:49:43 UTC

I'm not sure if you're looking to have these reported, but I just got the same error with WU 1321700. I was victim #6 and it's been sent to a 9th host.
ID: 721 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Eric Driver
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 8 Jul 11
Posts: 1320
Credit: 408,748,050
RAC: 257,145
Message 722 - Posted: 2 Oct 2012, 15:01:43 UTC - in response to Message 721.  

I'm not sure if you're looking to have these reported, but I just got the same error with WU 1321700. I was victim #6 and it's been sent to a 9th host.


Yes, thanks for reporting this. It's been a while since I've seen one of these errors.
ID: 722 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote

Message boards : Number crunching : The PARI stack overflows !


Main page · Your account · Message boards


Copyright © 2024 Arizona State University