20050815, 02:14  #1 
Aug 2004
New Zealand
2·5·23 Posts 
segfault with large input on Opteron
$ echo '100000!1'  ecm 1000
GMPECM 6.0.1 [powered by GMP 4.1.4] [ECM] Input number is 100000!1 (456574 digits) Using B1=1000, B2=44623, polynomial x^1, sigma=1288196876 Step 1 took 3245570ms Segmentation fault I also tried running this with v v turned on. This produced a few lines with bignums on it. The last few lines before the segfault are: B2'=65280 k=2 b2=32640 d=510 d2=1 dF=64, i0=1 Expected number of curves to find a factor of n digits: 20 25 30 35 40 45 50 55 60 65 3345 451878 1.1e+08 3.5e+10 inf inf inf inf inf inf Initializing tables of differences for F took 86618ms Computing roots of F took 320977ms Any ideas? The machine is a dual 1.8 GHz Opteron with 16GB of RAM. Last fiddled with by sean on 20050815 at 02:17 Reason: more info 
20050815, 07:55  #2 
Jun 2005
3×5 Posts 
Is GMP and ECM compiled using the stack or heap for number storage?

20050815, 09:59  #3 
"Nancy"
Aug 2002
Alexandria
2,467 Posts 
After "Computing roots of F" it starts to build the polynomial F where arithmetic with large polynomials starts. The KroneckerSchoenhage multiplication in GMPECM 6.0 uses a lot of temp space. As cjohnsto notes, the problem is probably that alloca() is used and it runs out of stack space.
Alex 
Thread Tools  
Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
msieve QS segfault when given large worktodo.ini  fivemack  Msieve  2  20111025 20:11 
Segfault in mprime v25.11 build 2  Graff  Software  9  20100104 21:51 
segfault  junky  NFSNET Discussion  0  20060706 03:25 
latest fedora core 3 kernels causes segfault  blackguard  Linux  1  20050706 01:05 
PRP3 segfault with big numbers.  Washuu  Software  6  20050407 17:08 