mprime 257 bug?

Hello
when I'm running mrpime257 on my computer, this program occurs segmentation fault.

[root@localhost Desktop]# cd ~/mprime
[root@localhost mprime]# cd x86
[root@localhost x86]# ./mprime -d
[Main thread Nov 13 01:26] Mersenne number primality test program version 25.7
[Main thread Nov 13 01:26] Starting workers.
[Worker #1 Nov 13 01:26] Worker starting
[Worker #2 Nov 13 01:26] Worker starting
[Worker #2 Nov 13 01:26] Setting affinity to run worker on logical CPU #1
[Worker #1 Nov 13 01:26] Setting affinity to run worker on logical CPU #0
[Worker #2 Nov 13 01:26] Optimal P-1 factoring of M47176159 using up to 512MB of memory.
[Worker #2 Nov 13 01:26] Assuming no factors below 2^68 and 2 primality tests saved if a factor is found.
[Worker #1 Nov 13 01:26] Optimal P-1 factoring of M47176123 using up to 512MB of memory.
[Worker #1 Nov 13 01:26] Assuming no factors below 2^68 and 2 primality tests saved if a factor is found.
[Worker #2 Nov 13 01:26] Optimal bounds are B1=565000, B2=14831250
[Worker #2 Nov 13 01:26] Chance of finding a factor is an estimated 6.43%
[Worker #1 Nov 13 01:26] Optimal bounds are B1=565000, B2=14831250
[Worker #1 Nov 13 01:26] Chance of finding a factor is an estimated 6.43%
[Worker #2 Nov 13 01:27] Using FFT length 2560K
[Worker #1 Nov 13 01:27] Using FFT length 2560K
[Worker #2 Nov 13 01:27] M47176159 stage 1 is 23.57% complete.
[Worker #1 Nov 13 01:27] M47176123 stage 1 is 20.06% complete.
[Worker #2 Nov 13 01:43] M47176159 stage 1 is 24.80% complete. Time: 1006.630 sec.
[Worker #1 Nov 13 01:44] M47176123 stage 1 is 21.28% complete. Time: 1049.582 sec.
[Worker #2 Nov 13 02:00] M47176159 stage 1 is 26.02% complete. Time: 1016.429 sec.
[Worker #1 Nov 13 02:01] M47176123 stage 1 is 22.51% complete. Time: 1046.348 sec.
[Worker #2 Nov 13 02:17] M47176159 stage 1 is 27.25% complete. Time: 1016.582 sec.
[Worker #1 Nov 13 02:18] M47176123 stage 1 is 23.74% complete. Time: 1017.952 sec.
[Worker #1 Nov 13 02:32] SUMOUT error occurred.
[Worker #1 Nov 13 02:32] Waiting five minutes before restarting.
[Worker #2 Nov 13 02:33] M47176159 stage 1 is 28.48% complete. Time: 969.902 sec.
[Worker #1 Nov 13 02:37] Using FFT length 2560K
[Worker #1 Nov 13 02:37] M47176123 stage 1 is 24.32% complete.
Segmentation fault
[root@localhost x86]#

Is this program bug?

[Prime95]

A segmentation fault is definitely a program bug. It looks like after the 5 minute wait, there is a bug in the restart code. I'll look into it.

Now, why did you get a SUMOUT error. This is probably a hardware error. If you are overclocking try reducing the overclock. Or run memtest or a torture test to see if you have a problem.

[ckdo]

Originally Posted by Prime95

A segmentation fault is definitely a program bug.

Speaking of which, this may be old news:

[Worker #1 Aug 14 01:16] Stage 2 GCD complete. Time: 246.293 sec.                                                       
[Worker #1 Aug 14 01:16] P-1 found a factor in stage #2, B1=410000, B2=11070000.                                        
[Worker #1 Aug 14 01:16] M34400389 has a factor: 9541285631068389553                                                    
Segmentation fault (core dumped)

Background: P-1 found a 64 bit factor for an exponent which was supposedly trial factored to 68 bits.

[Prime95]

The segmentation fault is of interest - do you know why it crashed? Was it communicating with the server? Firing up the next item in worktodo.txt?

The missed factor happens occasionally. It was missed quite a while ago. It might be a hardware error on the factoring machine or maybe a software problem - we'd have to rerun trial factoring from 2^63 to 2^64. Any volunteers?

[ckdo]

Originally Posted by Prime95

The segmentation fault is of interest - do you know why it crashed? Was it communicating with the server? Firing up the next item in worktodo.txt?

The missed factor happens occasionally. It was missed quite a while ago. It might be a hardware error on the factoring machine or maybe a software problem - we'd have to rerun trial factoring from 2^63 to 2^64. Any volunteers?

My impression at the time was that it crashed when trying to send the result in. The prime.spl entry had been created already but the worktodo.txt entry not yet removed (or the new version not yet written to disk). When restarted, mprime would start the same exponent again. My original post with more log detail is here.

As for the missed factor, the worktodo.txt line was inserted manually. The TF limit was simply too high, it should have been 63. The factor has been reported properly, and a TF-63-64 verification run was done as well already (which found the factor).