So far it looks like it is working...

[code]GMP-ECM 5.0.3 [powered by GMP 4.1.4] [ECM]
Resuming ECM residue saved with Prime95
Input number is 24707306311927565716857342128774085333197833223161879682238935306082805123046306993647507776054336486228891340858985829027076261887914242781617846672453431386903982455635542158748401823985988322905245077938567513252198179128990807936780194781391547404884040101606295111368825026273254703636026307207764436438929167613951 (320 digits)
Using special division for factor of 2^1061-1
Using B1=1, B2=110000000-680270182898, polynomial Dickson(30), sigma=4109972587527535
a=19738248110159599042754684441285702005161916686176762569452308094787997313970404763967728170341577895744665813660759436332673677122770652958642797523995222082337950187661171249809926616350511222735416190547058868430638213560987458443712576897417183503264114430743794947251483670115479871851531170886077885139221233503063
starting point: x=7359335644544843960877197431683542289090375881866979286423727711105731547174554832161414538758696270423001534679096983708734360783511582845694614426344065116204050994066053141926838464878818108595725465273170062949007088939778149415695241823389161884993853755070732948863367875158553061595089558211543201695119892661612
[b]Step 1 took 0ms[/b]
x=7359335644544843960877197431683542289090375881866979286423727711105731547174554832161414538758696270423001534679096983708734360783511582845694614426344065116204050994066053141926838464878818108595725465273170062949007088939778149415695241823389161884993853755070732948863367875158553061595089558211543201695119892661612[/code]
After it does one curve properly I will test it with many curves... For the initial test my input file has only one entry...

:smile:

It looks like it worked fine with multiple lines...

:smile:

[QUOTE=Xyzzy]It looks like it worked fine with multiple lines...

:smile:[/QUOTE]

Very good!

What are the timings per curve?

I tested on a 2.25 Ghz P4, and it used ~12000 secounds using gmp-ecm for stage 2.

Thomas
:coffee:

~2300s per curve...

[QUOTE=Xyzzy]~2300s per curve...[/QUOTE]

:shock:

Cool!
I have started doing 100 curves of M1061 B1=110M stage 1 only.
They should finish in 4 days time.

Thomas
:coffee:

[QUOTE=thomasn]I tested on a 2.25 Ghz P4, and it used ~12000 secounds using gmp-ecm for stage 2.[/QUOTE]
There must be something wrong here, on my 2.9GHz P4 a default (B2=680e9) stage 2 takes less than 4000 seconds.

[QUOTE=geoff]There must be something wrong here, on my 2.9GHz P4 a default (B2=680e9) stage 2 takes less than 4000 seconds.[/QUOTE]

Hmm. Strange. Maybe I compiled the binary with sub-optimal compiler options. Or maybe something else ran on the machine when I timed it. I will re-run one stage 2 curve in a few days.

Thomas
:coffee:

10 curves using B1=110000000 & B2=680270182898.

Mprime was used by me for stage 1, and Xyzzy used gmp-ecm for stage 2.

Thomas
:coffee:

[QUOTE=thomasn]Hmm. Strange. Maybe I compiled the binary with sub-optimal compiler options. Or maybe something else ran on the machine when I timed it.[/QUOTE]

Did you use a newer GMP release (currently 4.1.4) or the one your OS came equipped with?

[QUOTE=Mystwalker]Did you use a newer GMP release (currently 4.1.4) or the one your OS came equipped with?[/QUOTE]

Ah!
That is most likely the problem, as I used the one the OS came with. I will install the newest lib and recompile later.

Thomas
:coffee:

[Sun Jan 02 02:35:43 2005]
M1061 completed 100 ECM curves, B1=110000000, B2=4290000000

Thomas
:coffee: