New online applet for factorization
 

Hi, as many of you recently used Factor, I deployed a server-based application to try and factor exponents above 2[sup]32[/sup].

The page is [URL="http://www.moregimps.it/mersenne-test"]here[/URL].
It is managed by a DB: just click on the numbers in the "#" column and make k rise :smile:
Please feel free to visit and test it :smile:

Luigi

The correct link seems to be [url]http://www.moregimps.it/mersenne-test[/url] :smile:

[QUOTE=debrouxl;253361]The correct link seems to be [url]http://www.moregimps.it/mersenne-test[/url] :smile:[/QUOTE]

Shat hippens... Oh, well. :smile:

Thanks debrouxl...

Luigi

When I click on "#" column it should keep running until it finds a factor, right? Here the applet runs for 2 seconds then stops.

quoting from the howto
[quote]
What happens after you clicked on the link? First of all, a bit of theory. A Mersenne number has form 2p-1, where p is a prime number. Each factor of a Mersenne number has form 2kp+1 where p is the exponent of the number and k is a natural number (1,2,3,4,5---). When you click the link, the program makes a bit of magics and tests about 500,000 successive values of k (well, not exactly... there are some heuristics that lower the number of tests done under 100,000) using a well known algorithm. The values are stored into a database that keeps track of every single click, and updates the state of the search.
[/quote]
the K grow by 500k chunk.
what worry me is that, once the 10-15 first click, the progress seem to stop. Maybe you can increase the number of tested k past 57 bits?

[QUOTE=em99010pepe;253380]When I click on "#" column it should keep running until it finds a factor, right? Here the applet runs for 2 seconds then stops.[/QUOTE]

No, the applet keeps running until the chunk of 500,000 k is computed, so you can find 1, 2 or even 3 factors on the first run.

Luigi

[QUOTE=firejuggler;253382]quoting from the howto

the K grow by 500k chunk.
what worry me is that, once the 10-15 first click, the progress seem to stop. Maybe you can increase the number of tested k past 57 bits?[/QUOTE]

There is no upper limit on k. Anyway, with higher k's the increment between 2 batches is less than 0.0001 bits, so you may not notice it.

Luigi

Why does it such small batches?

34.5850 bits , k=3
35.0000 bits, k= 4
35.3219 bits, k=5
35.8074 bits, k=7
.....
55.3532 bits, k= 5357831
57.0992 bits, k=2^3*7 * 320923
hmmm pm1 would have been a better method there

I have another issue ... when I click a random number, Firefox always says: 'No factor found' before returning to the startpage :question:

thats the normal thing . it does trial divide ... if it does not find a factor, why should it do?

[QUOTE=smh;253388]Why does it such small batches?[/QUOTE]

Because I don't want to overload the server... :smile:
It's still in beta (alpha?) testing.

Luigi

[QUOTE=CedricVonck;253390]I have another issue ... when I click a random number, Firefox always says: 'No factor found' before returning to the startpage :question:[/QUOTE]

You've not been lucky...
When you find a factor, the server answers "Factor found : xxxxxxx" before loading the startpage.

Luigi

P.S. I noticed that someone has already discovered a more-than-57-bit factor!
P.P.S. Maybe I'll have to insert some new exponent to the queue...

[QUOTE=firejuggler;253389]34.5850 bits , k=3
35.0000 bits, k= 4
35.3219 bits, k=5
35.8074 bits, k=7
.....
55.3532 bits, k= 5357831
57.0992 bits, k=2^3*7 * 320923
hmmm pm1 would have been a better method there[/QUOTE]

Written in php, with less than 50 lines of code, and running in batches of 3-4 seconds? I doubt it... :smile:

Luigi

[QUOTE=ET_;253397]Written in php, with less than 50 lines of code, and running in batches of 3-4 seconds? I doubt it... :smile:[/QUOTE]
Is this a machine that you have full time control over? Or have you figure out a way to get time on this machine by doing "fancy webpages"?

Suggestion:
Add a button to the right of the stats that says something like "Test a 'Random' number from the list".
Have it find the number with the lowest 'k' that has no known factor and test it.

What is the significance of the range?

mfaktc can only test TF up to 2^32-1. After, you have to use either factor5 (which is much slower) or use ET_ applet, which doesn't move fast either,but can help you relieve your boredoom, as each test take 4 or 5 seconds.
an other reason could be to see how evolve k in that range , its distribution... or will edington interest in 'large' mersenne factor. Or, even less probable, just a challenge in PHP, and revelant to the interest of ET?

[QUOTE=Uncwilly;253406]Is this a machine that you have full time control over? Or have you figure out a way to get time on this machine by doing "fancy webpages"?

Suggestion:
Add a button to the right of the stats that says something like "Test a 'Random' number from the list".
Have it find the number with the lowest 'k' that has no known factor and test it.[/QUOTE]

That's a good point, especially when the list grows up! :smile:
I hope I can add it next Friday.

Luigi

[QUOTE=Uncwilly;253424]What is the significance of the range?[/QUOTE]

I started from where mfaktc stops.

Luigi

[QUOTE=firejuggler;253447]Another reason could be to see how evolve k in that range , its distribution... or will edington interest in 'large' mersenne factor. Or, even less probable, just a challenge in PHP, and revelant to the interest of ET?[/QUOTE]

All your guesses are exact... :smile:

Luigi

are factors shown incorrectly ?
 

factors in the exponent table are not equal to 2kp+1. Am I wrongly interpreting your table ? For instance in the second row:
for P=4294967357 you list K=35000535 and Factor=34359738857,
but if I am not mistaken, this factor gives a k value of 4.
Can you tell me where I am wrong?

[QUOTE=ppo;253542]factors in the exponent table are not equal to 2kp+1. Am I wrongly interpreting your table ? For instance in the second row:
for P=4294967357 you list K=35000535 and Factor=34359738857,
but if I am not mistaken, this factor gives a k value of 4.
Can you tell me where I am wrong?[/QUOTE]

For the bit depth of factors you should check [URL="http://www.moregimps.it/mersenne-test/factors.php"]here[/URL] :smile:

The exponents table shows how deep each exponent has been trial-factored.

Luigi

[QUOTE=ppo;253542]factors in the exponent table are not equal to 2kp+1. Am I wrongly interpreting your table ? For instance in the second row:
for P=4294967357 you list K=35000535 and Factor=34359738857,
but if I am not mistaken, this factor gives a k value of 4.
Can you tell me where I am wrong?[/QUOTE]

Here's how to interpret it:
A factor is known at k=4, and all potential factors through k=35000535 have been checked. (i.e. one known factor, with some progress beyond that)

[QUOTE=ET_;253358]Hi, as many of you recently used Factor, I deployed a server-based application to try and factor exponents above 2[sup]32[/sup].

The page is [URL="http://www.moregimps.it/mersenne-test"]here[/URL].
It is managed by a DB: just click on the numbers in the "#" column and make k rise :smile:
Please feel free to visit and test it :smile:

Luigi[/QUOTE]
Very neat, Luigi! :tu: It's nice to see instant results.

Now, if I could write a script that would tell my P75 to click every 5 seconds or so on one of those numbers, then I'd have a way for that old machine to contribute...

Rodrigo

a P75? under windows?
well simple enough
simply create a text file

[code]
FOR /L %%G IN (1,1,1489) DO wget.exe http://www.moregimps.it/mersenne-test/compute.php?id=%%G
del compute.php*
[/code]then name it at your convenience and change the extension to .bat
you have to get wget in the same folder
the only problem with this script is that it will 'test' all the exponent, even the already factored one.
But this is the simpliest i can get.

[QUOTE=firejuggler;253570]a P75? under windows?
well simple enough
simply create a text file

[code]
FOR /L %%G IN (1,1,1489) DO wget.exe http://www.moregimps.it/mersenne-test/compute.php?id=%%G
del compute.php*
[/code]then name it at your convenience and change the extension to .bat
you have to get wget in the same folder
the only problem with this script is that it will 'test' all the exponent, even the already factored one.
But this is the simpliest i can get.[/QUOTE]

I don't think windows 95/98 has wget installed (my windows 7-64 bit doesn't have it).

You could just call the URL from a link using a simple random number generator to create the id.

There should also be a "task scheduler" application that lets you run anything every n seconds/minutes (please don't be too greedy) :smile:

Finally, I'm trying to implement links that allow automatization of simple tasks (like run some TF) on smallest/biggest factored/unfactored exponents.

Luigi

ok, now I understand. Thanks a lot.

[QUOTE=ET_;253581]I don't think windows 95/98 has wget installed (my windows 7-64 bit doesn't have it).[/QUOTE]It is a GNU package and can be downloaded. There is even a MS-DOS version.

I'm no mathematician, but I just happen to see this and found it interesting ... I'm sure it has no significance.

[LEFT][FONT=Arial][SIZE=2]Exponet (p) --- [/SIZE][/FONT][FONT=Arial][SIZE=2]Factor 10(p)+1[/SIZE][/FONT][/LEFT]
[SIZE=2][COLOR=#0000aa]4294967563 42949675631[/COLOR][/SIZE]
[SIZE=2][COLOR=#0000aa]4294967983 [/COLOR][/SIZE][SIZE=2][COLOR=#0000aa]42949679831[/COLOR][/SIZE]
[SIZE=2][COLOR=#0000aa]4294972063 [/COLOR][/SIZE][SIZE=2][COLOR=#0000aa]42949720631[/COLOR][/SIZE]
[SIZE=2][COLOR=#0000aa]4294972267 [/COLOR][/SIZE][SIZE=2][COLOR=#0000aa]42949722671[/COLOR][/SIZE]
[SIZE=2][COLOR=#0000aa]4294973791 [/COLOR][/SIZE][SIZE=2][COLOR=#0000aa]42949737911[/COLOR][/SIZE]
[SIZE=2][COLOR=#0000aa]4294980763 [/COLOR][/SIZE][SIZE=2][COLOR=#0000aa]42949807631[/COLOR][/SIZE]
[SIZE=2][COLOR=#0000aa]4294982191 [/COLOR][/SIZE][SIZE=2][COLOR=#0000aa]42949821911[/COLOR][/SIZE]
[SIZE=2][COLOR=#0000aa]4294983811 [/COLOR][/SIZE][SIZE=2][COLOR=#0000aa]42949838111[/COLOR][/SIZE]
[SIZE=2][COLOR=#0000aa]4294996051 [/COLOR][/SIZE][SIZE=2][COLOR=#0000aa]42949960511[/COLOR][/SIZE]
[SIZE=2][COLOR=#0000aa]4294999363 [/COLOR][/SIZE][SIZE=2][COLOR=#0000aa]42949993631[/COLOR][/SIZE]

[QUOTE=drh;253594]I'm no mathematician, but I just happen to see this and found it interesting ... I'm sure it has no significance.

[LEFT][FONT=Arial][SIZE=2]Exponet (p) --- [/SIZE][/FONT][FONT=Arial][SIZE=2]Factor 10(p)+1[/SIZE][/FONT][/LEFT]
[SIZE=2][COLOR=#0000aa]4294967563 42949675631[/COLOR][/SIZE]
[SIZE=2][COLOR=#0000aa]4294967983 [/COLOR][/SIZE][SIZE=2][COLOR=#0000aa]42949679831[/COLOR][/SIZE]
[SIZE=2][COLOR=#0000aa]4294972063 [/COLOR][/SIZE][SIZE=2][COLOR=#0000aa]42949720631[/COLOR][/SIZE]
[SIZE=2][COLOR=#0000aa]4294972267 [/COLOR][/SIZE][SIZE=2][COLOR=#0000aa]42949722671[/COLOR][/SIZE]
[SIZE=2][COLOR=#0000aa]4294973791 [/COLOR][/SIZE][SIZE=2][COLOR=#0000aa]42949737911[/COLOR][/SIZE]
[SIZE=2][COLOR=#0000aa]4294980763 [/COLOR][/SIZE][SIZE=2][COLOR=#0000aa]42949807631[/COLOR][/SIZE]
[SIZE=2][COLOR=#0000aa]4294982191 [/COLOR][/SIZE][SIZE=2][COLOR=#0000aa]42949821911[/COLOR][/SIZE]
[SIZE=2][COLOR=#0000aa]4294983811 [/COLOR][/SIZE][SIZE=2][COLOR=#0000aa]42949838111[/COLOR][/SIZE]
[SIZE=2][COLOR=#0000aa]4294996051 [/COLOR][/SIZE][SIZE=2][COLOR=#0000aa]42949960511[/COLOR][/SIZE]
[SIZE=2][COLOR=#0000aa]4294999363 [/COLOR][/SIZE][SIZE=2][COLOR=#0000aa]42949993631[/COLOR][/SIZE][/QUOTE]

They all have k=5... :smile:

Luigi

Here's a small WIN script for running automatically:

script called "runGIMPS.bat":
[code]
@echo off
:begin
echo.>>all.txt
wget -q -O in.txt "http://www.moregimps.it/mersenne-test/compute.php?id=%1"
copy all.txt+in.txt all.txt >nul
del in.txt
goto begin
[/code]

Calling "runGIMPS.bat <index>" (for example "runGIMPS.bat 1") will run exponent 494967311 (index=1) automated (CTRL-C will stop this).

The results will collected in the file "all.txt" so you can look (while run over night) if there were a factor found.

The zip contains wget.exe and this DOS-batch.

Other idea:
- reading first the page which expos without any factor
- run any of those

@kar_bon
@firejuggler

Wow, thank you so much! With the combined info from your posts (what to get + how to use it) I should be able to get this to work.

And yes, I'll start with exponents that don't have any factors yet.

It's been a hectic day today, but I can't wait to try this tomorrow. Will report!

Rodrigo

[QUOTE=firejuggler;253570]a P75? under windows?
[/QUOTE]
firejuggler,

Believe it or not -- Windows for Workgroups 3.11 and MS-DOS 6.20. :blush:

This was my main work computer from 1995 to 2002. (I do have other, newer machines crunching for OBD and GIMPS already.)

Thank you VERY much for providing this info!

Rodrigo

[QUOTE=kar_bon;253609]Other idea:
- reading first the page which expos without any factor
- run any of those[/QUOTE]

Here's that sort of script:

- Download [url=www.rieselprime.de/dl/GIMPS_IT.zip]this[/url] zip-file containing all scripts and programs (and the script from above, too)

- unzip the file in an empty folder
- run "do.bat"

Explanations:
- the script first downloads the page with all exponents and factors found so far
- it creates a file called "expo.txt" with all exponents without a factor
- from this list another temporary batch called "gimpsget.bat" is created to test one of those exponents (by random or one after each other)
- the result (factor found or not) is written in "results.txt"
- after 100 attempts, the exponents-page is downloaded again, to be sure, only expos w/o factors are searched
- found factors are then collected in "factors.txt" and "results.txt" is deleted

[code]
@echo off

:begin
set /a n=1
if not exist results.txt goto begin1
findstr /c:"Factor found :" results.txt >>factors.txt
echo.>>factors.txt
del results.txt

:begin1
wget -O gimps.txt "http://www.moregimps.it/mersenne-test/"
gawk -f get_expo.awk gimps.txt >expo.txt

:next
if exist expo_neu.txt del expo_neu.txt
gawk -f make.awk -v random=1 expo.txt
del expo.txt
ren expo_neu.txt expo.txt

echo.>>results.txt
call gimpsget.bat
copy results.txt+in.txt results.txt >nul
del in.txt
set /a n=%n%+1
if %n% GTR 100 goto begin

goto next
[/code]

I've just found the factor 110648174743660201 of exponent #683 = 4294981591 with this script.

Set "random=0" to go through the whole list from top to down.
Doing this, the counter for 'n' should be set higher than 100 (see line "if %n% GTR 100 goto begin" at the end),
otherwise only first 100 expos will tested (currently should be 877 or more).

PS: I think, the DOS-command "findstr" is not available in V6.20 so please check first.

The currently tested expo-no is shown only in the DOS-BOX title because all outputs are disabled during testing (echo off).

[QUOTE=ET_;253581]Finally, I'm trying to implement links that allow automatization of simple tasks (like run some TF) on smallest/biggest factored/unfactored exponents.[/QUOTE]

Can you also allow a sort of k from smallest to lowest? That way it would be easier to request the lowest k's that have have no factor.

[QUOTE=Uncwilly;253660]Can you also allow a sort of k from smallest to lowest? That way it would be easier to request the lowest k's that have have no factor.[/QUOTE]

Of course, that makes sense! :smile:

You should have those links by the end of this week.

Luigi

[QUOTE=ET_;253672]Of course, that makes sense! :smile:

You should have those links by the end of this week.

Luigi[/QUOTE]

It seems that my page is having more success than expected... :smile:

I'm getting too many [COLOR="Red"]500 Internal Server Error[/COLOR], so I will lower the k request from 500,000 to 400,000 tonight, and see if the problem goes away.

Meanwhile, I'd appreciate if you put a sleep call in your scripts to avoid web server clobbering... :smile:

Sorry for the inconvenience.

Luigi

[QUOTE=ET_;253692]It seems that my page is having more success than expected... :smile:

I'm getting too many [COLOR=red]500 Internal Server Error[/COLOR], so I will lower the k request from 500,000 to 400,000 tonight, and see if the problem goes away.

Meanwhile, I'd appreciate if you put a sleep call in your scripts to avoid web server clobbering... :smile:

Sorry for the inconvenience.

Luigi[/QUOTE]
Ain't me (yet)! But I will keep your request in mind as I work on the script. :smile:

Rodrigo

Update
 

OK, I’ve done some experimenting with these batch files, and here’s what I have so far.

In Windows Vista, they work to perfection from Start – Run. DO.BAT also works if I double-click on it from Windows Explorer. It opens a DOS window and does its thing. RUNGIMPS.BAT (by itself) will also run, but it doesn’t seem to accomplish anything since from that route there is no way to tell it which exponent to test.

At first, neither of the batch files would work if I opened a DOS window first and then tried to launch them from there. For example, in these circumstances DO.BAT was returning an error saying that, “’do’ is not recognized as an internal or external command, operable program or batch file” (!). But today both batch files are working properly if I open a DOS window first.

The real test, though, is on the Windows for Workgroups 3.11/DOS 6.20 computer, which is the one that I’m trying to add to the project. I haven’t been able to get either of the batch files to work there. They both return a whole mess of error messages, no matter how I try to launch them. The most serious error appears to be that, “This program cannot be run in DOS mode.” This happens whether I click on the batch file through File Manager, open a DOS prompt, or use Run.

The next step will be to try these files on one of my Windows 98 systems. Windows 98 is of course at an intermediate stage between Windows 3.11 and Vista, so I'm curious to see how well it handles these batch files.

When I get the chance to (hopefully, later today) I will try a couple of ideas I’ve come up with to see if they solve (or work around) this issue. Suggestions are, of course, welcome!

Rodrigo

[QUOTE=Rodrigo;254002]RUNGIMPS.BAT (by itself) will also run, but it doesn’t seem to accomplish anything since from that route there is no way to tell it which exponent to test.

RUNGIMPS.BAT needs a parameter "%1" so calling "RUNGIMPS.BAT 1" will test cand. No.1.

Creating a shortcut on your desktop, you can add this param. at the target.
Doubleclick should work now.
[/quote]

[QUOTE=Rodrigo;254002]
At first, neither of the batch files would work if I opened a DOS window first and then tried to launch them from there. For example, in these circumstances DO.BAT was returning an error saying that, “’do’ is not recognized as an internal or external command, operable program or batch file” (!). But today both batch files are working properly if I open a DOS window first.
[/quote]

Perhaps the extension 'bat' was not assiciated properly to the cmd-command.

[QUOTE=Rodrigo;254002]
The real test, though, is on the Windows for Workgroups 3.11/DOS 6.20 computer, which is the one that I’m trying to add to the project. I haven’t been able to get either of the batch files to work there. They both return a whole mess of error messages, no matter how I try to launch them. The most serious error appears to be that, “This program cannot be run in DOS mode.” This happens whether I click on the batch file through File Manager, open a DOS prompt, or use Run.
[/quote]

The two used programs (wget and gawk) seems not working for those old WfW.
I have no chance to make this working here! So sorry for this task.

[QUOTE=Rodrigo;254002]
The next step will be to try these files on one of my Windows 98 systems. Windows 98 is of course at an intermediate stage between Windows 3.11 and Vista, so I'm curious to see how well it handles these batch files.
[/QUOTE]

Under Win 98 all should work fine but not the command "findstr".
Instead you can use
[code]
find "Factor found :" results.txt >>factors.txt
[/code]

Another point: the line with "echo.>>factors.txt" can/should be deleted.

[QUOTE=Rodrigo;254002]The real test, though, is on the Windows for Workgroups 3.11/DOS 6.20 computer, which is the one that I’m trying to add to the project. I haven’t been able to get either of the batch files to work there. They both return a whole mess of error messages, no matter how I try to launch them. The most serious error appears to be that, “This program cannot be run in DOS mode.” This happens whether I click on the batch file through File Manager, open a DOS prompt, or use Run.[/QUOTE]The version on wget attached above is not for the older DOS, there is a version that should run, check here:
[url]http://wget.addictivecode.org/FrequentlyAskedQuestions?action=show&redirect=Faq#download[/url]

And for DOS gawk, check here:
[url]http://www.gnu.org/software/gawk/manual/gawk.html#PC-Installation[/url]

I have taken to manually fetching the status page and then sorting it in Excel. Then I paste the expos that I want into the expo.txt file. I modified the do.bat to just loop the selected expos in order.

[QUOTE=kar_bon;254013]
The two used programs (wget and gawk) seems not working for those old WfW.
I have no chance to make this working here! So sorry for this task.
[/QUOTE]
Not a problem, kar_bon. Thank you for trying. And, in addition to the factoring work that my Vista did with these programs during testing, it looks like other folks have adopted the batch file for their own use with the project. So your efforts were not in vain.

[QUOTE=kar_bon;254013]Under Win 98 all should work fine but not the command "findstr".
Instead you can use
find "Factor found :" results.txt >>factors.txt
[/QUOTE]
I will try that. When launching DO.BAT in Win98, the results were stranger than in WFWG. The DOS window opened and the program appeared to start working, but unlike in Vista, there was no visible output after the first line or two, and all it showed was a blinking cursor. Then entering CTRL-C did not stop the program -- got an error message saying that I could not close the program from Windows. The Pause key didn't work, either. Used Task Manager to close the application, but I could still hear the PC churning afterward. :confused: Ended up having to shut down completely.

By coincidence, the MOREGIMPS.IT server may also have been down just at the time when I tried these batch files, and that may have influenced the results. (Going to the page via IE8 in Vista was yielding an "Internal Server Error" message.)

But Win98 was just an experiment for our purposes. I'll look into MS-DOS versions of GAWK.EXE and WGET.EXE.

Thanks again!

Rodrigo

[QUOTE=Rodrigo;254016]The DOS window opened and the program appeared to start working, but unlike in Vista, there was no visible output after the first line or two, and all it showed was a blinking cursor.[/QUOTE]

Comment out the first line with "REM @echo off". This will now show any output of the batch, good for testing.

[QUOTE=Uncwilly;254014]The version on wget attached above is not for the older DOS, there is a version that should run, check here:
[URL]http://wget.addictivecode.org/FrequentlyAskedQuestions?action=show&redirect=Faq#download[/URL]

And for DOS gawk, check here:
[URL]http://www.gnu.org/software/gawk/manual/gawk.html#PC-Installation[/URL]

I have taken to manually fetching the status page and then sorting it in Excel. Then I paste the expos that I want into the expo.txt file. I modified the do.bat to just loop the selected expos in order.[/QUOTE]
Thanks, Uncwilly.

The link to WGET for DOS didn't work for me ("Explorer cannot display the page"). But I did find this: [URL]http://www.rahul.net/dkaufman/[/URL]. Do a search on that page for "wget 1.8.2", it's about halfway down. What do you think?

Will also look into GAWK for DOS. This is taking me back 20 years! :smile: It could end up being a lot of fun, or highly frustrating. Hopefully it'll be like what they say for riding a bicycle, that you never forget and it comes back to you...

Rodrigo

wget 1.8.2 will not work here!
It does not support the option -O to download a page into a file.

I've used this before, too, until I was aware of this.

[QUOTE=kar_bon;254020]wget 1.8.2 will not work here!
It does not support the option -O to download a page into a file.

I've used this before, too, until I was aware of this.[/QUOTE]
Hmm...I use wget all the time to download files on the command line and have never needed to use -O. I just use "wget [I]url[/I]" which takes the file at the specified URL and saves it to the current directory. (Or does that not work with older wget versions?)

Can older wget versions at least print the downloaded file to stdout? If so, then that could probably work with some slight modifications to the batch script.

[QUOTE=kar_bon;254020]wget 1.8.2 will not work here!
It does not support the option -O to download a page into a file.

I've used this before, too, until I was aware of this.[/QUOTE]
Rats! Thanks for the warning -- good thing I hadn't gotten around to trying it yet. I was testing DO.BAT with echo on, on Win98. :wink:

Rodrigo

P.S. Got some partial output on Win98. DO.BAT does run, but the server is still down so no actual work is getting downloaded. While I wait, I will change the command from FINDSTR -- you were right, it wasn't working.

At what point should someone consider to discontinue the search on an exponent ... 66, 67, 68 bits?

[QUOTE=ET_;253692]It seems that my page is having more success than expected... :smile:

I'm getting too many [COLOR=Red]500 Internal Server Error[/COLOR], so I will lower the k request from 500,000 to 400,000 tonight, and see if the problem goes away.
[/QUOTE]Luigi,

When I use your page manually now, 500 Internal Server Error occurs about 3/4 of the time.

kar_bon,

I got both DO.BAT and runGIMPS.BAT to work properly on the Win98 machine -- thanks, at least in part, to your changes. With runGIMPS.BAT, where I can set the specific exponent to test, I can refresh and see the numbers on the webpage changing each time there's a result. :smile:

Great work!

Rodrigo

fatal error
 

I'm getting this since a few minutes:
Fatal error: Out of memory (allocated 1310720) (tried to allocate 16 bytes)

[QUOTE=cheesehead;254043]Luigi,

When I use your page manually now, 500 Internal Server Error occurs about 3/4 of the time.[/QUOTE]

I know. Unfortunately, I have not been home last weekend. Hope I can modify the script soon, sorry for the inconvenience...

Luigi

[COLOR="Red"]Edit[/COLOR]: I lowered the chunk of k to work with. Now the page appears more responsive, and I guess you can start your scriptds to test it. :smile:
Let me know if it keeps working.

Question for the users of the online factoring applet:

I need to know how many f you are interested in testing already factored exponents. If no one is interested, I may visualize only non factored exponents and maybe raise the delta_k in the inner routine.

Luigi

I'm still only getting fatal out of memory errors ... for the last few hours.

[QUOTE=drh;254080]I'm still only getting fatal out of memory errors ... for the last few hours.[/QUOTE]

It works for me, now... :smile:

Luigi

[QUOTE=ET_;254088]It works for me, now... :smile:

Luigi[/QUOTE]

It still doesn't work for me, but had been since the beginning. If I try it manually, it works about half the time.

<Later> More Checking ...

I've found that if I move to an exponent with a much smaller k, it works fine. I've been concentrating on #'s 1,3,4,8 & 11. When I move to #12, it works fine, at least so far.

I get always fatal error clicking on #1 or #8 , not on #3

[QUOTE=ppo;254094]I get always fatal error clicking on #1, not on #3[/QUOTE]

I've stopped #'s 1, 3, and 4, now running without errors on #'s 12, 13, and 14.

[QUOTE=drh;254095]I've stopped #'s 1, 3, and 4, now running without errors on #'s 12, 13, and 14.[/QUOTE]

My opinion: the problem is not related with the length of k, as the routine stops in different places, and never while allocating memory for the calculation.

It seems that the issue turns on during the call of the index page on highly requested exponents, that's why I asked your advice to generate a smaller index page and test the routine with higher delta_k.

Luigi

Well, I've made a LOT of progress getting this to work on MS-DOS.

I copied the previously mentioned WGET 1.8.2 (for DOS) to its own directory with RUNGIMPS.BAT, then used the Run command in WFW3.11 to specify an exponent. (I figured that if I can get this simpler batch file to work, then I can start working on the more complex DO.BAT file that uses GAWK.)

The result was an error message, "NO PACKET DRIVER FOUND". So I researched what the heck it was talking about, found the appropriate packet driver for my NIC, and copied it to the same directory. No dice.

Researched the topic some more, and found that I needed to load the packet driver via AUTOEXEC.BAT. Whoa, I'm not going to mess with my Windows autoexec.bat, so how about if I put everything on a boot floppy and a one-line AUTOEXEC.BAT file to load the NIC packet driver. (Windows isn't involved anyway.)

Did that, and rebooted. Next thing I get is a new error message:
"sin_mask" is 0.0.0.0 !
Cannot set local ip-address

Here I am drawing a blank. I'm not sure what IP address it wants me to enter... or how/where to enter it.

If I can figure this out, I can at least get RUNGIMPS.BAT to work on single exponents. Then I can start working on a more automated version using GAWK and DO.BAT. I already found two versions of GAWK ([URL]http://www.uni-leipzig.de/~gksprach/homepages/dirkj/gawk/index.html[/URL]) that are supposed to work in DOS and in WFW3.11.

Rodrigo

[QUOTE=Rodrigo;254129]Did that, and rebooted. Next thing I get is a new error message:
"sin_mask" is 0.0.0.0 !
Cannot set local ip-address

Here I am drawing a blank. I'm not sure what IP address it wants me to enter... or how/where to enter it.[/QUOTE]
Have you ever had that PC hooked up to the net via your current network connection? That is where you have to look.

[QUOTE=Uncwilly;254135]Have you ever had that PC hooked up to the net via your current network connection? That is where you have to look.[/QUOTE]
Yes. It's been connected via Ethernet card (through the DSL router/modem) ever since I installed the card last year. (Well, not connected ALL of the time, but you know what I mean...)

Although, of course, the PC's not electronically connected (only physically so) when I boot with a floppy.

Rodrigo

As an FYI/sanity check, factor5 takes about 50 seconds per exponent to go from a bit depth of 1 to 62 on my machine.

[CODE]Trial-factoring M4294989379 in [2^1, 2^62-1]
M4294989379 has 0 factors in [2^1, 2^62-1].
Trial-factoring M4294972037 in [2^1, 2^62-1]
M4294972037 has 0 factors in [2^1, 2^62-1].
Trial-factoring M4294988641 in [2^1, 2^62-1]
M4294988641 has 0 factors in [2^1, 2^62-1].
Trial-factoring M4294978223 in [2^1, 2^62-1]
M4294978223 has 0 factors in [2^1, 2^62-1].
Trial-factoring M4294991927 in [2^1, 2^62-1]
M4294991927 has 0 factors in [2^1, 2^62-1].
Trial-factoring M4294972663 in [2^1, 2^62-1]
[/CODE]

[QUOTE=Uncwilly;254255]As an FYI/sanity check, factor5 takes about 50 seconds per exponent to go from a bit depth of 1 to 62 on my machine.

[CODE]Trial-factoring M4294989379 in [2^1, 2^62-1]
M4294989379 has 0 factors in [2^1, 2^62-1].
Trial-factoring M4294972037 in [2^1, 2^62-1]
M4294972037 has 0 factors in [2^1, 2^62-1].
Trial-factoring M4294988641 in [2^1, 2^62-1]
M4294988641 has 0 factors in [2^1, 2^62-1].
Trial-factoring M4294978223 in [2^1, 2^62-1]
M4294978223 has 0 factors in [2^1, 2^62-1].
Trial-factoring M4294991927 in [2^1, 2^62-1]
M4294991927 has 0 factors in [2^1, 2^62-1].
Trial-factoring M4294972663 in [2^1, 2^62-1]
[/CODE][/QUOTE]
Uncwilly,

I hear ya. It's not surprising that a CPU doing all the work locally would be much faster than having to retrieve and send data over the
Net. But it's still a very cool app. :cool: Just knowing that something like this can be done is amazing!

Personally, I was hoping it'd serve as the solution for recruiting my last PC into one of the factoring projects, but that's not looking very promising at the moment. The programs that the batch files call want to run in Windows rather than DOS mode, while the batch files themselves must run in DOS mode.

Still hoping that I can get a version of a (any) factoring program that's been compiled for either WFW3.11 or for one of the tiny Linux versions that runs off a floppy... or learn how to do these things myself.

Rodrigo

[QUOTE=Rodrigo;254273]Uncwilly,

I hear ya. It's not surprising that a CPU doing all the work locally would be much faster than having to retrieve and send data over the
Net. But it's still a very cool app. :cool: Just knowing that something like this can be done is amazing!

Personally, I was hoping it'd serve as the solution for recruiting my last PC into one of the factoring projects, but that's not looking very promising at the moment. The programs that the batch files call want to run in Windows rather than DOS mode, while the batch files themselves must run in DOS mode.

Still hoping that I can get a version of a (any) factoring program that's been compiled for either WFW3.11 or for one of the tiny Linux versions that runs off a floppy... or learn how to do these things myself.

Rodrigo[/QUOTE]

Rodrigo, if the program wants to run under Windows, try to associate a .PIF (program Information File).
To activate the advanced functions of MS-DOS emulator you should recreate the PIF file. It's an easy task: open the directory where your MS-DOS program is and look for the file having the correct name ; right-click on its icon and choose "properties". You'll have some additional tabs on the information window. Choose "Screen", select your preferred options, then click "OK": windows will automagically save a new PIF file.

Luigi

[QUOTE=ET_;254275]Rodrigo, if the program wants to run under Windows, try to associate a .PIF (program Information File).
Luigi[/QUOTE]
Luigi,

I tried that tonight. It was actually a fun experience that took me back almost twenty years. Actually I probably know more now about WFW3.11 than I did back then. :smile: Dusted off the old manuals and really paid attention to the details...

But, unfortunately, creating the PIF file hasn't worked in this case. As soon as I click on either runGIMPS.bat or runGIMPS.pif, the DOS window opens with an immediate (and continually scrolling) warning that, "This program cannot be run in DOS mode." I guess that one could say it's "real" DOS and not a DOS emulation!

I did find a DOS version of WGET as I reported a few posts back, and after it complained about "no packet driver" I even set the NIC driver in that directory, but the program now comes to a halt with a "sin_mask" error instead, on which Google didn't find any useful info (or, at least, no info that I could make sense of). Therefore I'm stymied on both the Windows and the DOS side. But it HAS been an enjoyable experience -- I feel like I've gotten my hands dirty diving into the inner workings of the system. Sort of like tinkering with an old car.

Anyway, it looks like I'm back to setting up a tiny Linux for OBD. A couple of weeks ago I found some versions that run off a floppy diskette, so that the existing Windows installation can stay unaffected. What are the steps required in order to get any version of FactorX to run on (for example) BasicLinux?

Thanks for the PIF suggestion. I'm not out of options yet, I just need to figure out how to implement them.

Rodrigo

[QUOTE=Rodrigo;254432]What are the steps required in order to get any version of FactorX to run on (for example) BasicLinux?

Rodrigo[/QUOTE]

The actual FactorX on Linux usually only needs the GMP dynamic library, but I'm sure it is possible to create a statically linked executable.

Luigi

[QUOTE=ET_;254458]The actual FactorX on Linux usually only needs the GMP dynamic library, but I'm sure it is possible to create a statically linked executable.

Luigi[/QUOTE]
Thanks Luigi, I will look into that. If nothing else, pursuing this objective has helped me to learn quite a bit about the nuts and bolts of computing.

Rodrigo

The website [url]http://www.moregimps.it/mersenne-test/index.php[/url] appears to be active but not much activity lately.

[QUOTE=RichD;374785]The website [url]http://www.moregimps.it/mersenne-test/index.php[/url] appears to be active but not much activity lately.[/QUOTE]

Born as a server side experiment, it's there to be used from time to time. :smile: