Compiling Phrot

[mdettweiler]

Originally Posted by Anonymous

Okay, thanks, that worked! I was able to compile the "libs" folder's contents OK. Then when I move over to the phrot folder, and run the "make" command, I get a load of errors, so many that it overflows the scroll buffer.

Thanks to a quick Google search, I figured out how to redirect all output (i.e. both stdout and stderr) to a file, so I was able to capture all the "make" command output to a file. When I looked at the file's contents, though, I found the source of all the errors: I had forgotten to change one of the references to point to my copy of the glucas source! So, it was still pointing to wherever it would have been if it was on Rogue's machine.

So, now I'm going to go fix the reference in whatever makefile it's in, and try again...

Well, I fixed the reference to point to the correct glucas source directory, but I still got those gazillions of error messages. Again, I redirected the output to a file, and looked at the file to find that, though it was apparently looking in the right location, it still couldn't find a bunch of files!

A look at the glucas src directory told me that the files it was looking for really weren't there. So, I re-downloaded the glucas source, and extracted it afresh. Same error. Even when I tried some slightly older versions of glucas, the files weren't there.

I'm attaching the file that I redirected all the "make" command output to--Rogue, do you know what's going wrong this time?

[em99010pepe]

With the latest version the -o flag doesn't work.

phrot -b3 -o totest.txt

[tnerual]

Originally Posted by em99010pepe

With the latest version the -o flag doesn't work.

phrot -b3 -o totest.txt

use this:

phrot -b3 totest.txt >residues.txt

by the way, if you got a prp, it appear in "phrot.prp" file

[em99010pepe]

Thanks. I use this batch file to start up the client:

Start /min /low C:\dc\....\phrot.exe -b3 totest.txt > residues.txt

Made my first test with phrot:

Actually testing 21298250*390625^28790-1 (28792/65536 limbs)
170386*5^230323-1 [-163569,-180246,102328,-81013] is composite LLR64=C9ED90BA45018002. (e=0.01562 (0.0194567~6.03482e-16@0.986) t=2823.12s)

And LLRnet:

170386*5^230323-1 is not prime. RES64: C9ED90BA45018002 Time: 6133.648 sec.

Edit: Hey masser, please send me something like 10 000 Riesel k/n pairs (n=240k to n=260k)

[tnerual]

Originally Posted by em99010pepe

Thanks. I use this batch file to start up the client:

Start /min /low C:\dc\....\phrot.exe -b3 totest.txt > residues.txt

Made my first test with phrot:

Actually testing 21298250*390625^28790-1 (28792/65536 limbs)
170386*5^230323-1 [-163569,-180246,102328,-81013] is composite LLR64=C9ED90BA45018002. (e=0.01562 (0.0194567~6.03482e-16@0.986) t=2823.12s)

And LLRnet:

170386*5^230323-1 is not prime. RES64: C9ED90BA45018002 Time: 6133.648 sec.

Edit: Hey masser, please send me something like 10 000 Riesel k/n pairs (n=240k to n=260k)

pay attention to the fact that with this version of phrot, you have to remove the allready processed pairs from the "totest.txt" file before restarting it.
when starting again phrot, it will delete the old "residues.txt" file so you will have to make a copy of it each time you stop the client or restart the computer ==> no batch file in your startup folder

[rogue]

Originally Posted by tnerual

pay attention to the fact that with this version of phrot, you have to remove the allready processed pairs from the "totest.txt" file before restarting it.
when starting again phrot, it will delete the old "residues.txt" file so you will have to make a copy of it each time you stop the client or restart the computer ==> no batch file in your startup folder

I've tested about 13000 pairs for n = 25x,xxx, so you would probably be doing a double-check if you take anything in that range.

Originally Posted by em99010pepe

With the latest version the -o flag doesn't work.

That's odd. For each completed test, you should the results appended to a filed named "results.txt". What do you see if you type "phrot" on the command line without any options? BTW, which OS are you using? Finally, did you try to build with -DUNROLLED_MR to see if the timings are any better?

[rogue]

Originally Posted by Anonymous

Well, I fixed the reference to point to the correct glucas source directory, but I still got those gazillions of error messages. Again, I redirected the output to a file, and looked at the file to find that, though it was apparently looking in the right location, it still couldn't find a bunch of files!

A look at the glucas src directory told me that the files it was looking for really weren't there. So, I re-downloaded the glucas source, and extracted it afresh. Same error. Even when I tried some slightly older versions of glucas, the files weren't there.

I'm attaching the file that I redirected all the "make" command output to--Rogue, do you know what's going wrong this time?

Do you want to solve this via e-mail? We will probably get faster responses to one another and get this resolved more quickly.

It appears to me that your PHILMATH variable, .//libs, is incorrect. I assume that should be ../libs. From the glprov directory, what happens if you do this:

ls ../glucas/src/yeafft.h

If you get "No such file or directory, then your YEALIB variable is incorrect.

[em99010pepe]

Originally Posted by rogue

That's odd. For each completed test, you should the results appended to a filed named "results.txt". What do you see if you type "phrot" on the command line without any options? BTW, which OS are you using? Finally, did you try to build with -DUNROLLED_MR to see if the timings are any better?

I get this:

C:\test>phrot
Phil Carmody's Phrot (0.50e)
phrot -q expr   to PRP expr
phrot filename  to batch
-b=3         to perform 3-PRP test, and give LLR residue.

I'm using XP and did not compile the client, used the way tnerual said.

[mdettweiler]

Originally Posted by rogue

Do you want to solve this via e-mail? We will probably get faster responses to one another and get this resolved more quickly.

It appears to me that your PHILMATH variable, .//libs, is incorrect. I assume that should be ../libs. From the glprov directory, what happens if you do this:

ls ../glucas/src/yeafft.h

If you get "No such file or directory, then your YEALIB variable is incorrect.

Yeah, email would probably be better since this seems to be dragging on farther than I originally expected. I believe I mentioned my email earlier in this thread, or you could use the "email this user" link from my profile page (I have "receive email from other members" turned on).

About those variables, though, okay, I fixed it to say "./libs", because the Makefile.inc file is in the directory containing the glprov, glucas, and libs directories. As for the YEALIB variable, though--you're right, if I run that ls command, I get blank response (essentially a "no such file or directory" on my system). But, isn't yeafft.h supposed to be in the glucas src directory? After all, that is the correct path to my glucas directory. Is my glucas source incomplete, or something weird like that?

[mdettweiler]

Was the -o switch added in rogue's version? I don't think it's in the original 0.5 version, which Carlos is apparently using.

BTW Carlos and tnerual, why not simply say ">> results.txt" in the command line instead of "> results.txt"? That will tell the OS to append to results.txt rather than overwrite it.

[rogue]

Originally Posted by Anonymous

Was the -o switch added in rogue's version? I don't think it's in the original 0.5 version, which Carlos is apparently using.

BTW Carlos and tnerual, why not simply say ">> results.txt" in the command line instead of "> results.txt"? That will tell the OS to append to results.txt rather than overwrite it.

Go to this thread for version 0.51. It is Phil's version with some changes I've introduced.