Alternative Sieving for 10M digit prime search

[Citrix]

The sieve program has become twice as fast, thanks to Geoff. Please download the latest verison to sieve.

4005-4010 reserved.

[m_f_h]

Originally Posted by Citrix

The sieve program has become twice as fast, thanks to Geoff. Please download the latest verison to sieve.

4005-4010 reserved.

I may want to join you.
I tried to compile with ARCH=k8 (for my AMD Athlon 64 dual core)
but this gives bsgs.o : undefined reference to powmod64_k8 :-( !

(will try some more generic x86-amd but I fear I'll loose speed )

update: x68-amd gives same error...

[m_f_h]

euh sorry, I typed the command line

sr1sieve -i 3_16data.txt -o 3_16data.txt -f 3_16factors.txt -p4040000000000 -P4050000000000 --verbose

to see if it works.
This seems quite fast although I use the x86 executable on an amd-64
(it's doing 8330278 p/sec - is this OK?)
Anyway, it says will finish in about 20mins.
What do you want me to do with the results ?
(sorry for my ignorance, I did not yet have time to think about what's going on...)


UPDATE:
sr1sieve stopped: at p=4050000000000 because range is complete.
Wrote 657301 terms for 43046721*2^n+1 to NewPGen file `3_16data.txt'.
Found factors for 39 terms (expected about 44.93).

PS: I attach the new "factors", I could not attach the "terms" since the 657301 lines make up 1.7MB, even gzip'ed.
Still no plan what's happenin'...

[Citrix]

Thanks for the factors. I do not need the terms. I think you are using the client correct. Only thing is that 4010 to 4040 is unreserved, try working on lowest range unreserved.

As for compiling the client, there is an AMD version available inn version 15 on the website, so you don't really need to compile.

[ValerieVonck]

Range overview:

0 - 200G - unknown
200G - 4000G - CedricVonck - Done
4000G - 4005G - Citrix
4005G - 4010G - Citrix
4010G - 4040G - Unknown
4040G - 4100G - CedricVonck

[m_f_h]

Originally Posted by Citrix

Thanks for the factors. I do not need the terms. I think you are using the client correct. Only thing is that 4010 to 4040 is unreserved, try working on lowest range unreserved.

Well I took something a bit ahead in order not to duplicate work.
Now I see it's not ahead but reserved by CedricVonck
Anyway, I believe I have done 4040-4050, in about 30 min. Is this possible ?

I noticed that the _data file is identical to the previous version, except for the first line.
Could you please provide a little explanation (link) ? Thanks in advance.

PS: I'd be OK for doing 4010-4040 ; if previous assumptions are correct, this should take only 1h30... ???

[ValerieVonck]

4100G => 4500G CedricVonck - Reserved

[Citrix]

Originally Posted by m_f_h

Well I took something a bit ahead in order not to duplicate work.
Now I see it's not ahead but reserved by CedricVonck
Anyway, I believe I have done 4040-4050, in about 30 min. Is this possible ?

I noticed that the _data file is identical to the previous version, except for the first line.
Could you please provide a little explanation (link) ? Thanks in advance.

PS: I'd be OK for doing 4010-4040 ; if previous assumptions are correct, this should take only 1h30... ???

yes it should only take 30 min. 4010-4040 is being done by CerdicVonck, so please reserve a new range.
First line of the 3_16data.txt acts as the minimum sieve value.

[Citrix]

Originally Posted by Citrix

yes it should only take 30 min. 4010-4040 is being done by CerdicVonck, so please reserve a new range.
First line of the 3_16data.txt acts as the minimum sieve value.

Sorry, 4010-4040 is open for you. So you can work on it. I will try to get this thread more organized, later this week.

[m_f_h]

sorry, could you please explain or at least confirm:
1) we have a list of 658089 numbers of the form 3^16 * 2^n +1, with 1e6 <= n < 5e7
2) such a number is a 1e7 digit prime for n>3.5e7
3) we try to eliminate them by finding a factor q in the given range e.g.4010e9 < q <4040e9
4) is there a limit for the number of remaining candidates when we should stop trial factoring and do ECM or so on these ?

Q: How comes that after I processed the range 4040..4050, the 3-16data.txt file did not change at all, even though I found factors ?
e.g. in ...factors.txt :
4043905711873 | 43046721*2^49823744+1
in ...data.txt :
43046721 49823744

What should I do so that these composite exponents disappear ?
How many candidates are there really left ?

[ValerieVonck]

M_f_h,

This is my command line:

srsieve -i 3_16data.txt -o 3_16data.txt -f 3_16factors.txt --pmax <YMMV> --verbose

Citrix,

I calculated my score so far: 20.830.046,1475 for 104.061 factors

with following algorithm:

Dim txtBaseFactor As Variant
Dim txtFactor As Variant
Dim txtK As Variant
Dim txtTempN As Variant
Dim txtN As Variant
Dim txtScore As Variant
Dim txtTotScore As Variant
Dim testVariant As Variant
Dim i As Variant

i = 0

Open CurrentProject.Path & "\" & "3_16factors1.txt" For Input As #1
 Do Until EOF(1)
    Line Input #1, testVariant
     
    txtBaseFactor = testVariant
    txtFactor = Mid(txtBaseFactor, 1, InStr(1, txtBaseFactor, "|") - 2)
    txtK = Mid(txtBaseFactor, InStr(txtBaseFactor, "*") - 1, 1)

    txtTempN = Mid(txtBaseFactor, InStr(txtBaseFactor, "^") + 1)
    txtN = Mid(txtTempN, 1, Len(txtTempN) - 2)
    
    txtScore = txtFactor / 100000000000# * txtN / 1000000
    txtTotScore = Round(CDbl(txtTotScore) + CDbl(txtScore), 4)
    
    i = i + 1
Loop
 
Close #1

Me.txtScore = txtTotScore & " for " & i & " factors"

Ps. should we have a site for our own?
I can host a couple of "static" pages.... and / or a central dat file (or input.txt)