Assertion failed

Andi47 · 2009-08-01, 12:31

I just got this error when running aliqueit.exe (aliquot sequences), I assume that the error was generated by ECM:

Code:

Assertion failed: (__builtin_constant_p (1UL) && (1UL) == 0 ? ((F[i])->_mp_size < 0 ? -1 : (F[i])->_mp_size > 0) : __gmpz_cmp_ui (F[i],1UL)) == 0, file pm1fs2.c, line 1713

It was trying to find a factor of this c93:
500932227439480882534434656406251090340619826528894701445819119292053983744651640218766142539

Unfortunately, aliqueit.exe has already deleted the outputfile, so I can only assume that it has happened at one of these steps, most probably at one of the last ones:

Code:

[Aug 01 2009, 11:28:29] c93: running P-1 at B1=132e3...

[Aug 01 2009, 11:28:30] c93: running P+1 x3 at B1=132e3...

[Aug 01 2009, 11:28:34] c93: running 74 ecm curves at B1=11e3...

[Aug 01 2009, 11:28:53] c93: running P-1 at B1=6e5...

[Aug 01 2009, 11:28:56] c93: running P+1 x3 at B1=6e5...

[Aug 01 2009, 11:29:10] c93: running 214 ecm curves at B1=5e4...

[Aug 01 2009, 11:33:05] c93: running P-1 at B1=3e6...

[Aug 01 2009, 11:33:17] c93: running P+1 x3 at B1=3e6...

[Aug 01 2009, 11:34:18] c93: running 430 ecm curves at B1=25e4...

[Aug 01 2009, 14:10:29] c93: running P-1 at B1=12e6...

[Aug 01 2009, 14:11:11] c93: running P+1 x3 at B1=12e6...

I am using ECM 6.2.3 with GMP 4.3.0, compiled under minGW/msys. The PC is a Pentium 4, running Windows XP, 3 GB RAM available. (So I don't think that for example P+1 with B1=12e6 might exceed the available memory.)

akruppa · 2009-08-04, 12:37

The test that failed checks that the polynomial F(x) is monic. If it isn't, it means that something went wrong while recursively expanding F(x), but what exactly is hard to guess... a bug in the weighted NTT? In the code to expand F(x) from the factored set S? An out-of-memory condition that does not get handeled properly? There's quite a bit of code involved in building F(x). Without exact parameters to reproduce the failed assertion, there's not much I can do about it, I'm afraid.

Alex

Andi47 · 2009-08-04, 14:11

Aliqueit.exe is running the curves with the given B1. For ECM it takes default B2, for P-1 and p+1 it uses -b2scale 5.0 (I altered the aliqueit.ini file in a way that it uses that b2scale.)

For me, Out of memory seems very unlikely on a PC which is equipped with 3 GB RAM and an c93 input number.

As I said, unfortunately aliqueit.exe had deleted the logfile before I caught the error message.

(Idea: What do you think about adding the input parameters to the error messages?)

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
GMP-ECM 7.0-dev assertion failed!	Antonio	GMP-ECM	9	2015-03-10 08:21
Assertion failure in 6.4.2	bsquared	GMP-ECM	4	2013-03-01 15:52
Can an OS config be OK after RAM failed?	RickC	Hardware	8	2010-10-28 03:31
GMP-ECM Assertion	yoyo	GMP-ECM	8	2009-01-16 20:57
connection failed	musos	Information & Answers	1	2008-12-12 14:29

2009-08-04, 12:37	#2
akruppa "Nancy" Aug 2002 Alexandria 2,467 Posts	The test that failed checks that the polynomial F(x) is monic. If it isn't, it means that something went wrong while recursively expanding F(x), but what exactly is hard to guess... a bug in the weighted NTT? In the code to expand F(x) from the factored set S? An out-of-memory condition that does not get handeled properly? There's quite a bit of code involved in building F(x). Without exact parameters to reproduce the failed assertion, there's not much I can do about it, I'm afraid. Alex

2009-08-04, 14:11	#3
Andi47 Oct 2004 Austria 2×17×73 Posts	Aliqueit.exe is running the curves with the given B1. For ECM it takes default B2, for P-1 and p+1 it uses -b2scale 5.0 (I altered the aliqueit.ini file in a way that it uses that b2scale.) For me, Out of memory seems very unlikely on a PC which is equipped with 3 GB RAM and an c93 input number. As I said, unfortunately aliqueit.exe had deleted the logfile before I caught the error message. (Idea: What do you think about adding the input parameters to the error messages?)