FastECPP software and 50000 digit primality proof (reposted from NMBRTHRY)

[Batalov]

Interesting bits of advice from Andreas (and Bill):
1. Because libpari is built into the code, the -n '<value>' expression can also use Pari functions, and perhaps even complex statements, like first define an ad-hoc function, then ";", then call it. I am going to try this. This definitely works: -n 'fibonacci(37217)/4466041'
2. If your linux is proper (modern) then your command-line is essentially unlimited (like 2 megabytes) so you can call with -n `cat myfile`

[frmky]

You can also do something like

FILES="primo*"
for f in $FILES
do
  NUM=$(grep 'N$=' $f | sed '/^N$=/!d;s///;q')
  mpirun -np 32 ecpp-mpi -v -n "0x$NUM" -f cert-$f
done

On a related note, I've recently submitted about 1000 smaller certificates to FactorDB.

Edit: You'll also need to convert the results to Primo format with something like

FILES="cert*"
for f in $FILES
do
   cp $f /tmp/temp
   echo "C=read(\"/tmp/temp\");write(\"/tmp/temp.out\",primecertexport(C,1));" | gp -q
   sed 's/N=0x/N=$/g' /tmp/temp.out > $f.out
   rm /tmp/temp
   rm /tmp/temp.out
done

[rudy235]

Originally Posted by Cybertronic

Wow , 15k in 30h ..that is fast !


So I have 3 milestone 20k numbers for testing !


Smallest 20000 digit prime and prime twins.


10^19999+110949
10^19999+1514722609+d, d=0,2


Should be done, the prover extended this page...
https://pzktupel.de/smx000.html


regards

Perhaps they could include this PRP

10²⁹⁹⁹⁹+89821. *Smallest 30000 digit PRP*

[xilman]

Originally Posted by frmky

You can also do something like

FILES="primo*"
for f in $FILES
do
  NUM=$(grep 'N$=' $f | sed '/^N$=/!d;s///;q')
  mpirun -np 32 ecpp-mpi -v -n "0x$NUM" -f cert-$f
done

On a related note, I've recently submitted about 1000 smaller certificates to FactorDB.

Edit: You'll also need to convert the results to Primo format with something like

FILES="cert*"
for f in $FILES
do
   cp $f /tmp/temp
   echo "C=read(\"/tmp/temp\");write(\"/tmp/temp.out\",primecertexport(C,1));" | gp -q
   sed 's/N=0x/N=$/g' /tmp/temp.out > $f.out
   rm /tmp/temp
   rm /tmp/temp.out
done

A simple cut and paste gave me

  ***   at top-level: C=read("/tmp/temp");write("/tmp/temp.out",prim
  ***                   ^--------------------------------------------
  *** read: the PARI stack overflows !
  current stack size: 8000000 (7.629 Mbytes)
  [hint] set 'parisizemax' to a nonzero value in your GPRC

sed: can't read /tmp/temp.out: No such file or directory
rm: cannot remove '/tmp/temp.out': No such file or directory
  ***   syntax error, unexpected real number, expecting $end: 
  ***   ...05713.513159164.9156911606.091862374.224116268
  ***                               ^---------------------
sed: can't read /tmp/temp.out: No such file or directory
rm: cannot remove '/tmp/temp.out': No such file or directory
  ***   syntax error, unexpected real number, expecting $end: 
  ***   0000646.165922646.1821702.7308300.000000587.08
  ***                           ^----------------------
sed: can't read /tmp/temp.out: No such file or directory
rm: cannot remove '/tmp/temp.out': No such file or directory

The fix ought to be straightforward by following the hint given.

[Cybertronic]

@rudy235


All smallest full thousand digit proven primes k-tuplets are welcome !

[paulunderwood]

Originally Posted by Batalov

Interesting bits of advice from Andreas (and Bill):
1. Because libpari is built into the code, the -n '<value>' expression can also use Pari functions, and perhaps even complex statements, like first define an ad-hoc function, then ";", then call it. I am going to try this. This definitely works: -n 'fibonacci(37217)/4466041'
2. If your linux is proper (modern) then your command-line is essentially unlimited (like 2 megabytes) so you can call with -n `cat myfile`

One might have to type 'print(fibonacci(37217)/4466041)' to avoid trucation by Pari/GP. I certainly get truncation with echo fibonacci(37217)/4466041 | gp -q but maybe it works with the -n switch on ecpp

[chris2be8]

Is FastECPP significantly faster than primo for 300-2000 digit primes? I've automated primo by using xdotool and I'm working to prove 300+ digit PRPs in factordb. FastECPP would be easier to automate since it's a batch program, but would I save enough CPU time to pay for time spent scripting it?

[xilman]

Originally Posted by xilman

A simple cut and paste gave me

  ***   at top-level: C=read("/tmp/temp");write("/tmp/temp.out",prim
  ***                   ^--------------------------------------------
  *** read: the PARI stack overflows !
  current stack size: 8000000 (7.629 Mbytes)
  [hint] set 'parisizemax' to a nonzero value in your GPRC

sed: can't read /tmp/temp.out: No such file or directory
rm: cannot remove '/tmp/temp.out': No such file or directory
  ***   syntax error, unexpected real number, expecting $end: 
  ***   ...05713.513159164.9156911606.091862374.224116268
  ***                               ^---------------------
sed: can't read /tmp/temp.out: No such file or directory
rm: cannot remove '/tmp/temp.out': No such file or directory
  ***   syntax error, unexpected real number, expecting $end: 
  ***   0000646.165922646.1821702.7308300.000000587.08
  ***                           ^----------------------
sed: can't read /tmp/temp.out: No such file or directory
rm: cannot remove '/tmp/temp.out': No such file or directory

The fix ought to be straightforward by following the hint given.

Hint: gp -q -s 100000000; seemed to work nicely.

Certificate now uploaded. Let's see if it is acceptable.

[frmky]

Originally Posted by xilman

Hint: gp -q -s 100000000; seemed to work nicely.

Certificate now uploaded. Let's see if it is acceptable.

Placing this in $HOME/.gprc also fixes it.

\\ Limit PARI stack size to 1000 Mbytes = 10*10^8 bytes
parisizemax = 1000M

\\ Set PARI typical stack size to 40 Mbytes = 4*10^7 bytes (will grow as
\\ needed, up to parisizemax)
parisize = 40M

[xilman]

Originally Posted by frmky

Placing this in $HOME/.gprc also fixes it.

\\ Limit PARI stack size to 1000 Mbytes = 10*10^8 bytes
parisizemax = 1000M

\\ Set PARI typical stack size to 40 Mbytes = 4*10^7 bytes (will grow as
\\ needed, up to parisizemax)
parisize = 40M

I have little doubt that it will fix it, as claimed, but not that it is necessarily a good idea for general use in pari/gp, especially on systems which may be memory bound.

Most of the time I prefer a rapid OOM to a system which is paging itself to a standstill.

[xilman]

Anyone any idea how long it generally takes factordb to verify a certificate?

3178^1543+1543^3178 is still in the Processing state.