Compiling Phrot

[MyDogBuster]

Another option is to redirect the output to a file as before, but use ">>" instead of ">"--i.e., append to the output file instead of overwriting it.

I wasn't using the >> and was losing the original output file. Thanks

[Cruelty]

Originally Posted by rogue

I have been thinking about modifying phrot so that if k or b is a power of 2 that it will not perform the test unless using the error checking version. Does anyone have a strong opinion on that?

Good idea

[mdettweiler]

Originally Posted by rogue

I have finally added checkpointing to phrot and support for -s (stopOnPRP). Checkpointing means that phrot will write the current data to disk when during a test. When restarting, phrot will then restart from where it last checkpointed. This will be very helpful when running long tests and you need to shutdown or restart your computer and don't want to have to start a test from the beginning.

I have attached the sources for phrot 0.52. If anyone runs into any issues, please let me know.

Could somebody possibly make a Linux build of this?

[rogue]

Originally Posted by Cruelty

Good idea

I've been thinking of actually making the error checking a command line option so that there is only one build. That way someone can use the "-e" command line argument to force error checking for all tests. If I do that, then I can flip the switch on the fly for this case when "-e" is not used.

[Cruelty]

Originally Posted by rogue

I've been thinking of actually making the error checking a command line option so that there is only one build. That way someone can use the "-e" command line argument to force error checking for all tests. If I do that, then I can flip the switch on the fly for this case when "-e" is not used.

For a specific cases when Phrot is known to produce bad results it would make more sense to force error-checking feature just with some on screen information that it is being used and why. Of course a command-line switch could also be used in other cases if someone really wants to.

[mdettweiler]

Originally Posted by nuggetprime

Stop-on-prp files now available. Just add the -s switch and it will stop on the first prp.

--nuggetprime

When I try to run this version under 32-bit Linux (Ubuntu 8.10, to be exact), I get the following error:

./phrot: 1: Syntax error: ")" unexpected

However, a friend of mine who's running this exact same binary on 64-bit Ubuntu 8.04 didn't mention any problems. Interestingly enough, though, I have no problems running Geoff's build--though I'd kinda rather use yours given the ~28% speed increase you reported.

Also, when I try running the non-stop-on-prime version of your build, it gives me almost the same error message--the only difference being is that it gives the opposite parenthesis from the above error.

Do you have any idea what may be causing this?

[Siemelink]

Hi Rogue,

I've found two PRPs that give different results with different programs.

I have this one, found with LLR and I confirmed it with PFGW:
Primality testing 63063805242*3^5166-1 [N+1, Brillhart-Lehmer-Selfridge]
Running N+1 test using discriminant 3, base 3+sqrt(3)
Calling Brillhart-Lehmer-Selfridge with factored part 99.59%
63063805242*3^5166-1 is prime! (1.4525s+0.0004s)

On my old phrot this also resulted as PRP. But on the phrot.p3 and phrot.p3_e that I got from this thread this is reported composite.



And there is this one, where none of the phrots found the PRP.
Primality testing 63063960488*3^5184-1 [N+1, Brillhart-Lehmer-Selfridge]
Running N+1 test using discriminant 5, base 1+sqrt(5)
Running N+1 test using discriminant 5, base 2+sqrt(5)
Running N+1 test using discriminant 5, base 4+sqrt(5)
Running N+1 test using discriminant 5, base 5+sqrt(5)
Running N+1 test using discriminant 5, base 7+sqrt(5)
Running N+1 test using discriminant 5, base 10+sqrt(5)
Calling Brillhart-Lehmer-Selfridge with factored part 99.56%
63063960488*3^5184-1 is prime! (7.5649s+0.0005s)


To be sure that I am not running a corrupted program I checked some other PRPs with less outlandish k. These all reported the PRPs that I expected. So I am concluding that there is a limit on the size of k. What would it be? Can it be lifted?
Or I have a setup problem? I have windows XP and a Core Duo at 1.80 Ghz

Food for thought, Willem.

[rogue]

Originally Posted by Siemelink

Hi Rogue,

I've found two PRPs that give different results with different programs.

I have this one, found with LLR and I confirmed it with PFGW:
Primality testing 63063805242*3^5166-1 [N+1, Brillhart-Lehmer-Selfridge]
Running N+1 test using discriminant 3, base 3+sqrt(3)
Calling Brillhart-Lehmer-Selfridge with factored part 99.59%
63063805242*3^5166-1 is prime! (1.4525s+0.0004s)

On my old phrot this also resulted as PRP. But on the phrot.p3 and phrot.p3_e that I got from this thread this is reported composite.



And there is this one, where none of the phrots found the PRP.
Primality testing 63063960488*3^5184-1 [N+1, Brillhart-Lehmer-Selfridge]
Calling Brillhart-Lehmer-Selfridge with factored part 99.56%
63063960488*3^5184-1 is prime! (7.5649s+0.0005s)


To be sure that I am not running a corrupted program I checked some other PRPs with less outlandish k. These all reported the PRPs that I expected. So I am concluding that there is a limit on the size of k. What would it be? Can it be lifted?
Or I have a setup problem? I have windows XP and a Core Duo at 1.80 Ghz

Food for thought, Willem.

I'll take a look. Right now phrot on MacPPC and MacIntel (Core 2 Duo) report both of these as PRP. I do see that on Windows (Core 2 Duo) it returns composite for both. I don't know why. It will probably take me a little while to figure this one out.

[rogue]

Originally Posted by rogue

I'll take a look. Right now phrot on MacPPC and MacIntel (Core 2 Duo) report both of these as PRP. I do see that on Windows (Core 2 Duo) it returns composite for both. I don't know why. It will probably take me a little while to figure this one out.

I have found and fixed the issue. The problem was an integer overflow when phrot was choosing the optimal k/b for the test. I would recommend that if you are using phrot for the Sierpinski/Riesel base 3 attacks that you get the latest build from here. I've used phrot for other bases and it appears that it most likely when happen k is very large, i.e. > 2^32 and b is very small.

[Siemelink]

Originally Posted by rogue

I have found and fixed the issue. .

Great! I am glad it was easy to pin down.
I've got the new versions now. What is the difference again between phrot.p3 and phrot.p3_e? Oh never mind, I found it way back in this thread. _e for Error checking.

Thanks a bunch, Willem.

[Siemelink]

Originally Posted by rogue

I have found and fixed the issue.

Maybe. Is there more that I should do than to replace phrot.p3? My first example now gives PRP, but the second one still reports composite.

phrotp3 -b3 t17_b3.prp

Input 63063960488*3^5184-1 Actually testing 3723863802855912*1594323^398-1 (witness=3 401/896 limbs)
Klen>2 not supported for LLR residues
63063960488*3^5184-1 [-418304,450292,-649446,-387508] is composite LLR64=6c49bb5a5b6d4311. (t=0.94s)

Or phrotp3 -b3 t17_b3.prp

Input 63063960488*3^5184-1 Actually testing 3723863802855912*1594323^398-1 (witness=2 401/896 limbs)
63063960488*3^5184-1 [193024,-200289,-589860,-265982] is composite. (t=1.01s)

Thanks, Willem.