mersenneforum.org  

Go Back   mersenneforum.org > Factoring Projects > GMP-ECM

Reply
 
Thread Tools
Old 2013-01-21, 17:35   #23
ET_
Banned
 
ET_'s Avatar
 
"Luigi"
Aug 2002
Team Italia

476610 Posts
Default

Quote:
Originally Posted by akruppa View Post
The "$" in "$HAVE___GMPN_ADD_NC = 1" is a typo in main.c:184. It is harmless except for the stray "$" in the output.

Luigi: your config.log contains
Code:
configure:14122: checking for __gmpn_add_nc
configure:14122: gcc -std=gnu99 -o conftest -W -Wall -Wundef -pedantic -O2 -m64 -march=core2 -mtune=core2    conftest.c -lm -lm -lm -lm -lm  -lgmp >&5
/tmp/ccWHZdew.o: In function `main':
conftest.c:(.text.startup+0x7): undefined reference to `__gmpn_add_nc'
collect2: ld returned 1 exit status
configure:14122: $? = 1
configure: failed program was:
....
configure:14122: result: no
so the configure script should not set HAVE___GMPN_ADD_NC to 1. Edit: question already asked by WraithX. What happens when you explicitly link to a specific GMP library with the configure --with=gmp=/path/to/library option?
To be sincere, I did not try.
Having a working executale that was passing all checks was enough for me, but if you need an answer I still have the old environment and can do some tests...

Luigi
ET_ is offline   Reply With Quote
Old 2013-01-21, 18:01   #24
akruppa
 
akruppa's Avatar
 
"Nancy"
Aug 2002
Alexandria

2,467 Posts
Default

I'm relatively sure that is was a case of using different GMP versions during configure tests and when dynamically linking the resulting binary, as WraithX already suggested. This happens fairly often.

I checked and the __gmpn_add_nc() function still has the same signature as in 5.0.x, and still is not in the public interface, so not much should have changed wrt. our use of that function.
akruppa is offline   Reply With Quote
Old 2013-01-27, 11:16   #25
M0CZY
 
M0CZY's Avatar
 
May 2005
Brutal Police State, UK

2·59 Posts
Default

I have just compiled ECM-6.4.3 with GMP-5.1.0, and it completed properly with no errors, but when I ran 'make check' it stopped with the following error, on both Windows XP and Ubuntu 10.04.4

Code:
GMP-ECM 6.4.3 [configured with GMP 5.1.0] [P-1]
Input number is 234^997+997^234 (2363 digits)
Using B1=100, B2=492, polynomial x^1, x0=1055220884
Step 1 took 15ms
Step 2 took 1719ms
All P-1 tests are ok.
echo ""

./test.ecm ./ecm.exe
GMP-ECM 6.4.3 [configured with GMP 5.1.0] [ECM]
Input number is 2050449353925555290706354283 (28 digits)
Using B1=30, B2=0-2443992, polynomial x^1, sigma=7
Step 1 took 0ms
Step 2 took 47ms
********** Factor found in step 2: 30210181
Found probable prime factor of  8 digits: 30210181
Probable prime cofactor 67872792749091946543 has 20 digits
GMP-ECM 6.4.3 [configured with GMP 5.1.0] [ECM]
Input number is 2050449353925555290706354283 (28 digits)
Using B1=30, B2=0-2443992, polynomial x^1, sigma=7
Step 1 took 0ms
Step 2 took 47ms
********** Factor found in step 2: 30210181
Found probable prime factor of  8 digits: 30210181
Probable prime cofactor 67872792749091946543 has 20 digits
GMP-ECM 6.4.3 [configured with GMP 5.1.0] [ECM]
Input number is 212252637915375215854013140804296246361 (39 digits)
Using B1=63421, B2=1822795201-1822795212, polynomial x^1, sigma=781683988
Step 1 took 204ms
Step 2 took 0ms
############### ERROR ###############
Expected return code 8 but got 0
make: *** [check] Error 1
Does this error mean that the compiled ECM files shouldn't be used?
M0CZY is offline   Reply With Quote
Old 2013-01-27, 13:13   #26
akruppa
 
akruppa's Avatar
 
"Nancy"
Aug 2002
Alexandria

2,467 Posts
Default

That looks like a bug. Don't use that binary for factoring work, but perhaps keep it around in case I can't reproduce the bug here.
akruppa is offline   Reply With Quote
Old 2013-01-27, 21:29   #27
Dubslow
Basketry That Evening!
 
Dubslow's Avatar
 
"Bunslow the Bold"
Jun 2011
40<A<43 -89<O<-88

3·29·83 Posts
Default

The timings for each step are very bizarre compared to the previous non-errors.
Dubslow is offline   Reply With Quote
Old 2013-01-27, 21:51   #28
Ralf Recker
 
Ralf Recker's Avatar
 
Oct 2010

191 Posts
Default

Building gmp-ecm 6.4.3 with gmp 5.1.0 fails. Compiling works but 'make check' fails. This issue was fixed by Paul Zimmermann at the end of 2012 in the svn trunk.

Code:
[mpmod.c] fix to deal with change of semantics of the internal GMP functions
          mpn_redc_1 and mpn_redc_2

Last fiddled with by Ralf Recker on 2013-01-27 at 21:57
Ralf Recker is offline   Reply With Quote
Old 2013-02-09, 14:24   #29
M0CZY
 
M0CZY's Avatar
 
May 2005
Brutal Police State, UK

7616 Posts
Default

Quote:
Building gmp-ecm 6.4.3 with gmp 5.1.0 fails. Compiling works but 'make check' fails.
Is a copy of ECM that does work going to be posted onto the download page?
M0CZY is offline   Reply With Quote
Old 2013-02-15, 21:57   #30
PhilF
 
PhilF's Avatar
 
Feb 2005
Colorado

2·257 Posts
Default

I get the exact same error on 'make check' as the previous poster, at the exact same spot, on 2 different machines. This is with GMP 5.1.1 and ECM 6.4.3.

I don't know if the code is OK but it is only a problem with the check, or if I should go to a previous version of GMP, or if I should go to a previous version of ECM, or both.

Code:
Input number is 212252637915375215854013140804296246361 (39 digits)
Using B1=63421, B2=1822795201-1822795212, polynomial x^1, sigma=781683988
Step 1 took 204ms
Step 2 took 0ms
############### ERROR ###############
Expected return code 8 but got 0
make: *** [check] Error 1

Last fiddled with by PhilF on 2013-02-15 at 21:59
PhilF is offline   Reply With Quote
Old 2013-02-17, 01:16   #31
PhilF
 
PhilF's Avatar
 
Feb 2005
Colorado

51410 Posts
Default

I can confirm that installing and using the older GMP 5.0.1 fixed the 'make check' problem I was having.
PhilF is offline   Reply With Quote
Old 2013-02-18, 15:20   #32
PhilF
 
PhilF's Avatar
 
Feb 2005
Colorado

2×257 Posts
Default

I can also confirm that ECM 6.4.3 / GMP 5.1.1 combo compiles and checks fine on a 64-bit Core 2 machine w/64-bit Debian linux.

The failures I was having was on 32-bit P4 machines (Northwood core).
PhilF is offline   Reply With Quote
Old 2013-02-18, 16:33   #33
akruppa
 
akruppa's Avatar
 
"Nancy"
Aug 2002
Alexandria

2,467 Posts
Default

The bug is due to GMP 5.1.x using a different semantic for __gmp_redc_{12}() functions than GMP 5.0.x and MPIR did, and GMP-ECM 6.4.{0123} expects the old kind. A GMP-ECM 6.4.4 release is currently being prepared, which fixes this incompatibility.

Last fiddled with by akruppa on 2013-02-18 at 16:52 Reason: GMP 5.0.x *did* have the functions, with the old semantics
akruppa is offline   Reply With Quote
Reply

Thread Tools


Similar Threads
Thread Thread Starter Forum Replies Last Post
Error Prone Machines PageFault Data 17 2012-04-10 01:40
Scalability of Glucas on large machines: A future //ed prime 95 ? T.Rex Software 29 2006-12-13 21:40
Start and Stop Prime 95 on Large Groups of Windows XP Machines MarcGetty Software 3 2006-03-07 07:54
Putting prime 95 on a large number of machines moo Software 10 2004-12-15 13:25
Team_Prime_Rib error-prone machines GP2 Data 10 2003-10-05 18:34

All times are UTC. The time now is 05:17.

Mon Sep 28 05:17:53 UTC 2020 up 18 days, 2:28, 0 users, load averages: 1.44, 1.46, 1.42

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2020, Jelsoft Enterprises Ltd.

This forum has received and complied with 0 (zero) government requests for information.

Permission is granted to copy, distribute and/or modify this document under the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software Foundation.
A copy of the license is included in the FAQ.