GMP-ECM help

[lukerichards]

Hi,

I'm having some difficulty installing GMP-ECM. Initially it could not located a GMP header, so I installed GMP from a binary downloaded from the GMP website.

Only now I think I have two versions of GMP on my computer. I get the following error message:

checking for gmp.h... yes
checking for recent GMP... yes
checking if GMP is MPIR... no
checking whether we can link against GMP... yes
checking if gmp.h version and libgmp version are the same... (6.1.2/6.1.0) no
configure: error: 'gmp.h' and 'libgmp' have different versions, you have to reinstall GMP properly, or use the --with-gmp parameter to tell configure the path to the GMP library and header you want to use


Can anyone offer any suggestions? I would use the --with-gmp parameter but I don't know how to find the location of the GMP library I want to use.

[EdH]

I don't know what OS you are using, but I have some detailed instructions that may help you sort it out in the following threads:

How I Install GMP onto my Ubuntu Machines

How I Install GMP-ECM onto my Ubuntu Machines

In the last step, when you install GMP, the screen output will tell you where it is installed. On a linux system, this is commonly /usr/local/ so you would use:

./configure --with-gmp=/usr/local/

This is all described in the referenced threads.

Let me know if the threads are helpful.

[Dubslow]

Originally Posted by lukerichards

checking if gmp.h version and libgmp version are the same... (6.1.2/6.1.0) no
configure: error: 'gmp.h' and 'libgmp' have different versions

Can you offer details of the binary you downloaded and installed? What was the binary's file name? It sounds like your computer already had a gmp installed, just not a development version, and if that is so, downloading a second version from the gmp website may have been a mistake.

Also what operating system? If this is some sort of Linux, please post the output of sudo updatedb; locate gmp.h; locate libgmp.so, in addition to the name of the file you downloaded.

[lukerichards]

luke@lukes-amda8:~$ sudo updatedb; locate gmp.h; locate libgmp.so
/usr/include/linux/igmp.h
/usr/include/netinet/igmp.h
/usr/local/include/gmp.h
/usr/src/linux-headers-4.10.0-28/include/linux/igmp.h
/usr/src/linux-headers-4.10.0-28/include/uapi/linux/igmp.h
/usr/src/linux-headers-4.13.0-26/include/linux/igmp.h
/usr/src/linux-headers-4.13.0-26/include/uapi/linux/igmp.h
/usr/src/linux-headers-4.13.0-32/include/linux/igmp.h
/usr/src/linux-headers-4.13.0-32/include/uapi/linux/igmp.h
/usr/src/linux-headers-4.13.0-36/include/linux/igmp.h
/usr/src/linux-headers-4.13.0-36/include/uapi/linux/igmp.h
/usr/src/linux-headers-4.13.0-37/include/linux/igmp.h
/usr/src/linux-headers-4.13.0-37/include/uapi/linux/igmp.h
/usr/src/linux-headers-4.13.0-38/include/linux/igmp.h
/usr/src/linux-headers-4.13.0-38/include/uapi/linux/igmp.h
/usr/src/linux-headers-4.13.0-39/include/linux/igmp.h
/usr/src/linux-headers-4.13.0-39/include/uapi/linux/igmp.h
/usr/lib/x86_64-linux-gnu/libgmp.so.10
/usr/lib/x86_64-linux-gnu/libgmp.so.10.3.0
/usr/lib/x86_64-linux-gnu/openssl-1.0.0/engines/libgmp.so
/usr/local/lib/libgmp.so
/usr/local/lib/libgmp.so.10
/usr/local/lib/libgmp.so.10.3.2

File downloaded was gmp-6.1.2.tar.lz

Running Lubuntu 16.04.

[Dubslow]

Originally Posted by lukerichards

luke@lukes-amda8:~$ sudo updatedb; locate gmp.h; locate libgmp.so
...
/usr/local/include/gmp.h
...
/usr/lib/x86_64-linux-gnu/libgmp.so.10
/usr/lib/x86_64-linux-gnu/libgmp.so.10.3.0
...
/usr/local/lib/libgmp.so
/usr/local/lib/libgmp.so.10
/usr/local/lib/libgmp.so.10.3.2

File downloaded was gmp-6.1.2.tar.lz

Running Lubuntu 16.04.

I believe this confirms my hypothesis. You have the 6.1.0 gmp libraries in your /usr/lib/x86_64-linux-gnu/, installed by the system package, as well as the 6.1.2 libraries and development files installed by the package you downloaded.

If you compiled the downloaded tar yourself, it might be better optimized for your system. At any rate, since you have already done the legwork there, you may as well use that for GMP-ECM; as the above lines indicate, --with-gmp=/usr/local in your configure command ought to do the trick, as suggested by EdH. Try it and let us know.

[pepi37]

How to output factors found with GMP-ECM to text file?
I was following help, and got that GPM-ECM record something on file, but it record all except factors ( and that is most important to me)
syntax I use is

ecm -one -savea factors.txt -c 100 4000 6000 < test.txt

In as factor is found I got this on screen

********** Factor found in step 2: 7977946786400951
Found prime factor of 16 digits: 7977946786400951
Composite cofactor (285601*2^216+1)/7977946786400951 has 55 digits

But it is never recorded to output text.

[VBCurtis]

Redirect the output to a file: add ">output.txt" or similar to the end of your invocation.

[pepi37]

Originally Posted by VBCurtis

Redirect the output to a file: add ">output.txt" or similar to the end of your invocation.

Thanks!
Works perfect!

[kosta]

Or add a pipe like this
ecm ...... | tee -a foo.log
You will have both the file and on screen.

[kriesel]

Originally Posted by VBCurtis

Redirect the output to a file: add ">output.txt" or similar to the end of your invocation.

It's safer to use >> which appends (creating the output file if it doesn't yet exist), rather than blowing away what might be there from a previous run with >.

[kriesel]

Hi, very new to gmp-ecm here. Is this right? Prime95 is orders of magnitude faster than gmp-ecm for P-1 on 2^~88E6 -1 on the same i3 hardware? Or what am I doing wrong, other than running the Core 2 code that I could find, on an i3?
The difference in gmp-ecm between -v -v and -v is about 12% more time for more verbosity per a spot check I made at p=99991.

-v put the following in the output file, versus -v -v including some rather large integer decimal expansions.

GMP-ECM 7.0.5-dev [configured with GMP 6.1.2, --enable-asm-redc] [P-1]
Input number is 2^99991-1 (30101 digits)
Using special division for factor of 2^99991-1
Using lmax = 813 without NTT which takes about 181MB of memory
Using B1=10000, B2=773118, polynomial x^1, x0=271186255
P = 1155, l = 813, s_1 = 480, k = s_2 = 1, m_1 = 3
Probability of finding a factor of n digits:
20    25    30    35    40    45    50    55    60    65
0.0052    0.0002    5.3e-006    9.9e-008    1.4e-009    2.1e-011    0    0    0    0
Writing checkpoint to m99991.ckp at p = 10000
Step 1 took 9828ms
Computing F from factored S_1 took 7956ms
Computing h took 686ms
Multi-point evaluation 1 of 1:
Computing g_i took 1654ms
TMulGen of g and h took 4087ms
Computing product of F(g_i) took 359ms
Step 2 took 14805ms
********** Factor found in step 2: 1799839
Found prime factor of 7 digits: 1799839
Composite cofactor (2^99991-1)/1799839 has 30095 digits
Peak memory usage: 187MB
gmpecm returned errorlevel 6

Total observed runtime was about 275. seconds, much larger than the ~39 second total of the program's timing values in the output above.

Running several widely spaced smallish exponents in gmp-ecm, and extrapolating significantly, I get an estimate of several years for p~88M on gmp-ecm, compared to about a week on prime95, same cpu. (476 weeks, 9+ years)

See the attachment for more info.