Message boards :
Number crunching :
New PARI error?
Message board moderation
Author | Message |
---|---|
Send message Joined: 28 Oct 11 Posts: 179 Credit: 223,608,902 RAC: 113,054 |
Task 5506164 Now starting the targeted Martinet search: |
Send message Joined: 8 Jul 11 Posts: 1323 Credit: 410,382,578 RAC: 243,966 |
Task 5506164 I have seen that before, although very infrequently. This is a hard one to track down, because I have never been able to replicate it - I bet the next user who gets it will finish it successfuly and if I run it stand alone it will also succeed. |
Send message Joined: 25 Feb 13 Posts: 216 Credit: 9,899,302 RAC: 0 |
Task 5506164 I hope the error is fund. I now that the newest PARI-version don´t run at Windows, because an compiling problem. I think we use PARI 2.5(6?), or? If we can´t solve that problem, we look to some alternatives, like MATLAB. But this is expensive... :( I hope the other WU would be computed succesfuly. not enough precomputed primes... |
Send message Joined: 8 Jul 11 Posts: 1323 Credit: 410,382,578 RAC: 243,966 |
Just to follow up on this... The next user who got this WU processed it successfully. I also had no problems running it in stand alone mode. So like I said, every time I have seen this error, I have been unable to duplicate it. This obviously makes it very hard to debug. The problem is not specific to a particular platform- I have seen it on both Windows and Linux. I suspect something might be happening in the host's environment that is leading to this error message (low memory, disk I/O, etc). Pari reserves a chunk of memory for it's workspace and I believe that is where the precomputed primes are placed- maybe this memory is somehow being clobbered??? |