Ubuntu troubles with cm builds - Page 2

EdH · 2022-11-30, 13:47

I originally used the repository libmpc-dev:

Code:

$ sudo apt list libmpc-dev
Listing... Done
libmpc-dev/focal,now 1.1.0-1 amd64 [installed]
libmpc-dev/focal 1.1.0-1 i386

and mpc:

Code:

$ sudo apt list mpc
Listing... Done
mpc/focal,now 0.33-1 amd64 [installed]

and that machine is the only one that is "almost" running fine. It did crash for a particular candidate. This is the only machine that creates Primo certificates. Oddly, it never complained about mpc being too old.

I've been trying all available (mpc mpfrcx and cm) packages from multiprecision.org and mpfr from mpfr.org. The compiled ecpp does run, but if it isn't going to write Primo certificates and its check program complains (and it failed on one candidate that Primo worked fine with), then I don't intend to use it.

The mpfr I retrieved is 4.1.1, while mpc wants >=4.1.0. I've considered trying mpfr-4.1.0 just to see if mpc likes that version.

To Chris' OS point: I currently have over three dozen Ubuntu machines. Although Ubunu is constantly studying me to see how much more annoyance I will tolerate with its new anti-features and added steps for everything, I don't see me moving to another OS. But, I do have a few with Debian already on them. I suppose I should give one of them a try for test completeness.

paulunderwood · 2022-11-30, 14:12

Quote:

Originally Posted by EdH

I've been trying all available (mpc mpfrcx and cm) packages from multiprecision.org and mpfr from mpfr.org. The compiled ecpp does run, but if it isn't going to write Primo certificates and its check program complains (and it failed on one candidate that Primo worked fine with), then I don't intend to use it.

It might seem like a silly question: Are you using the file flag -f ?

EdH · 2022-11-30, 15:06

Yes. Here is an example from the one machine that creates Primo certs:

Code:

$ bash ecpp -n '(3^1071*798+1)/140401818061117' -v -f test
CM: include 0.4.1dev, lib 0.4.1dev
GMP: include 6.2.1, lib 6.2.1
MPFR: include 4.0.2, lib 4.0.2
MPC: include 1.1.0, lib 1.1.0
MPFRCX: include 0.6.3, lib 0.6.3
PARI: include 2.11.2, lib 2.11.2
--- Time for primality test:       0.0 (0.0)
Could not open file 'test.cert1' for reading.
Writing to 'test.cert1'.
-- Time for class numbers up to Dmax=689728: 0.1 (0.1)
. . .
--- Time for first ECPP step, depth 52:  5.0 (5.0)
Could not open file 'test.cert2' for reading.
Writing to 'test.cert2'.
-- Time for 1661 bits (discriminant -280, invariant d, parameters 5_7_1_1_5_7): 0.1
. . .
--- Time for second ECPP step: 5.8 (5.9)
Writing to 'test'.
Writing to 'test.primo'.
--- Total time for ECPP:       10.9 (10.9)
$ ls test*
test  test.cert1  test.cert2  test.primo

and one of the others that will run:

Code:

$ bash ecpp -n '(3^1071*798+1)/140401818061117' -v -f test
GMP: include 6.2.0, lib 6.2.0
MPFR: include 4.0.2, lib 4.0.2
MPC: include 1.1.0, lib 1.1.0
MPFRCX: include 0.6.3, lib 0.6.3
PARI: include 2.11.2, lib 2.11.2
Could not open file 'test.cert1' for reading.
Writing to 'test.cert1'.
-- Time for class numbers up to Dmax=689728: 0.1 (0.1)
. . .
--- Time for first ECPP step, depth 45:  6.5 (6.5)
Could not open file 'test.cert2' for reading.
Writing to 'test.cert2'.
-- Time for 1661 bits (discriminant -4472, invariant d, parameters 2_13_2_2_2_13): 0.3
. . .
--- Time for second ECPP step: 7.6 (7.7)
Writing to 'test'.
--- Total time for ECPP:       14.1 (14.1)
$ ls test*
test  test.cert1  test.cert2

But many of my machines fail due to the libmpfrcx.so.1 error, all of the compiled mpc fail its check. I've also tried a Debian machine which errored for mpc and failed for libmpfrcx.so.1.

At this point I'm quite confused with which machines are which in regards to repo vs. manual and where each has failed. I would like to know why cm ./configure finds libmpfrcx, but fails its check. Some machines find it and I can't determine the difference.

paulunderwood · 2022-11-30, 15:11

The next question is: are you you using the same version of cm? Here is what I am using:

Code:

ecpp -v                                                          
CM: include 0.4.1dev, lib 0.4.1dev
GMP: include 6.2.1, lib 6.2.1
MPFR: include 4.1.0, lib 4.1.0
MPC: include 1.2.1, lib 1.2.1
MPFRCX: include 0.6.3, lib 0.6.3
PARI: include 2.13.4, lib 2.13.4
***** Error: cm_ecpp called with composite number.

PS. In notice "CM" is not listed on the machine that does not produce a primo file for you, Is this because you did not run sudo make install?

EdH · 2022-11-30, 15:30

Now, this is interesting! I thought I got them both from the exact same link on the multiprecision.org site, but they are clearly different:

Code:

$ bash ecpp -v
CM: include 0.4.1dev, lib 0.4.1dev
GMP: include 6.2.1, lib 6.2.1
MPFR: include 4.0.2, lib 4.0.2
MPC: include 1.1.0, lib 1.1.0
MPFRCX: include 0.6.3, lib 0.6.3
PARI: include 2.11.2, lib 2.11.2
***** Error: cm_ecpp called with composite number.
$ ls ../../cm*
../../cm-0.4.1dev-41c4bce.tar.gz

Code:

$ bash ecpp -v
GMP: include 6.2.0, lib 6.2.0
MPFR: include 4.0.2, lib 4.0.2
MPC: include 1.1.0, lib 1.1.0
MPFRCX: include 0.6.3, lib 0.6.3
PARI: include 2.11.2, lib 2.11.2
--- Time for first ECPP step, depth 0:  0.0 (0.0)
--- Time for second ECPP step: 0.0 (0.0)
--- Total time for ECPP:       0.0 (0.0)
$ ls ../../cm*
../../cm-0.4.0.tar.gz

Was there a link provided in one of the threads that I might have used for the first machine?

paulunderwood · 2022-11-30, 15:35

Quote:

Originally Posted by EdH

Was there a link provided in one of the threads that I might have used for the first machine?

Yes. Later versions produce a primo file, although you have rename it with a ".out" extension and chage it to "n=$...." inside the primo file, to get factoDB to accept it. There are scripts using sed to do this in another thread. Or of course you can alter the source and recompile and re-install.

EdH · 2022-11-30, 15:41

No. I ran install and I just ran it again with no change. But how did I get the dev version, 0.4.1dev? I don't see that on the site.

I'm going to try that on a totally separate machine and see what happens with it.

EdH · 2022-11-30, 15:47

I found it! The original download was from the news section of the CM pages. All my other downloads were from the CM download page "Latest version" which is only 0.4.0. I will step back for a moment and then renew my efforts. I still need to see if the other issues cause trouble.

Thank you very much for helping me sort this part out.

paulunderwood · 2022-11-30, 15:52

Quote:

Originally Posted by EdH

No. I ran install and I just ran it again with no change. But how did I get the dev version, 0.4.1dev? I don't see that on the site.

I'm going to try that on a totally separate machine and see what happens with it.

You might have to sudo make uninstall firstly. Let us know how you get on with a fresh machine.

EdH · 2022-11-30, 15:53

Quote:

Originally Posted by paulunderwood

Yes. Later versions produce a primo file, although you have rename it with a ".out" extension and chage it to "n=$...." inside the primo file, to get factoDB to accept it. There are scripts using sed to do this in another thread. Or of course you can alter the source and recompile and re-install.

Actually, with the ones I tested, I didn't need to change anything to get factordb to accept and process them correctly - no editing and no name change. I just browsed to the .primo file and sent it. This was via the html page at factordb. I haven't tried a curl upload yet.

EdH · 2022-11-30, 16:20

Quote:

Originally Posted by paulunderwood

You might have to sudo make uninstall firstly. Let us know how you get on with a fresh machine.

No Joy! This was a totally different machine so no uninstall was used. I'm faced with the libmpfrcx.so.1 trouble:

Code:

FAIL: tcm
=========

./tcm: error while loading shared libraries: libmpfrcx.so.1: cannot open shared object file: No such file or directory
FAIL tcm (exit status: 127)

FAIL: ttwist
============

./ttwist: error while loading shared libraries: libmpfrcx.so.1: cannot open shared object file: No such file or directory
FAIL ttwist (exit status: 127)

FAIL: tecpp
===========

./tecpp: error while loading shared libraries: libmpfrcx.so.1: cannot open shared object file: No such file or directory
FAIL tecpp (exit status: 127)

For mpfrcx, I used:

Code:

./configure - no errors
make - no errors
make check - no errors
sudo make install - no errors

But for cm:

Code:

./configure - no errors
make - no errors
make check - three failures

as shown above. I tried:

Code:

$ whereis libmpfrcx.so.1
libmpfrcx.so: /usr/local/lib/libmpfrcx.so /usr/local/lib/libmpfrcx.so.1

and, even added /usr/local/lib to my PATH. No progress was made.

2022-11-30, 13:47	#12
EdH "Ed Hall" Dec 2009 Adirondack Mtns 5260₁₀ Posts	I originally used the repository libmpc-dev: Code: $ sudo apt list libmpc-dev Listing... Done libmpc-dev/focal,now 1.1.0-1 amd64 [installed] libmpc-dev/focal 1.1.0-1 i386 and mpc: Code: $ sudo apt list mpc Listing... Done mpc/focal,now 0.33-1 amd64 [installed] and that machine is the only one that is "almost" running fine. It did crash for a particular candidate. This is the only machine that creates Primo certificates. Oddly, it never complained about mpc being too old. I've been trying all available (mpc mpfrcx and cm) packages from multiprecision.org and mpfr from mpfr.org. The compiled ecpp does run, but if it isn't going to write Primo certificates and its check program complains (and it failed on one candidate that Primo worked fine with), then I don't intend to use it. The mpfr I retrieved is 4.1.1, while mpc wants >=4.1.0. I've considered trying mpfr-4.1.0 just to see if mpc likes that version. To Chris' OS point: I currently have over three dozen Ubuntu machines. Although Ubunu is constantly studying me to see how much more annoyance I will tolerate with its new anti-features and added steps for everything, I don't see me moving to another OS. But, I do have a few with Debian already on them. I suppose I should give one of them a try for test completeness.

2022-11-30, 15:11	#15
paulunderwood Sep 2002 Database er0rr 5×29×31 Posts	The next question is: are you you using the same version of cm? Here is what I am using: Code: ecpp -v CM: include 0.4.1dev, lib 0.4.1dev GMP: include 6.2.1, lib 6.2.1 MPFR: include 4.1.0, lib 4.1.0 MPC: include 1.2.1, lib 1.2.1 MPFRCX: include 0.6.3, lib 0.6.3 PARI: include 2.13.4, lib 2.13.4 ***** Error: cm_ecpp called with composite number. PS. In notice "CM" is not listed on the machine that does not produce a primo file for you, Is this because you did not run sudo make install? Last fiddled with by paulunderwood on 2022-11-30 at 15:32

2022-11-30, 15:41	#18
EdH "Ed Hall" Dec 2009 Adirondack Mtns 2²·5·263 Posts	No. I ran install and I just ran it again with no change. But how did I get the dev version, 0.4.1dev? I don't see that on the site. I'm going to try that on a totally separate machine and see what happens with it. Last fiddled with by paulunderwood on 2022-11-30 at 15:50

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Gpuowl Windows builds	kriesel	GpuOwl	34	2022-12-10 19:18
Would Personal Builds run win-11?	a1call	Hardware	29	2021-11-27 14:42
Apparent Ubuntu Version Troubles with ecmpi	EdH	GMP-ECM	1	2020-05-04 13:16
ARM builds and SIMD-assembler prospects	ewmayer	Mlucas	183	2019-02-25 08:17
Running 32-bit builds on a Win7 system	ewmayer	Programming	34	2010-10-18 22:36

2022-11-30, 15:47	#19
EdH "Ed Hall" Dec 2009 Adirondack Mtns 2²×5×263 Posts	I found it! The original download was from the news section of the CM pages. All my other downloads were from the CM download page "Latest version" which is only 0.4.0. I will step back for a moment and then renew my efforts. I still need to see if the other issues cause trouble. Thank you very much for helping me sort this part out.