Compiling Phrot

[rogue]

Originally Posted by Siemelink

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.

Fixed again. I only tested your first number, which passed, but didn't test your second number. The code had two multiplies that can cause an overflow. I fixed the one, presuming that it would resolve the error for both. I was wrong. I have now fixed the other as well. You can d/l from the aforementioned link.

[mdettweiler]

So...given these errors, is it safe to assume that all numbers that Phrot gives a "Klen>2" error for will be reported as composite regardless of whether they actually are prime? I too am running the Riesel base 3 project and get these errors sporadically throughout the files that I'm testing; however, I'm using Linux and the new, fixed builds are for Windows only. In fact, due to this I am still stuck using Geoff's Linux build of Phrot 0.51 from a little while back, i.e. the version without checkpointing or any of that stuff...

Rogue, has building a Linux binary for Phrot become easier since when you and I tried to build one on my machine a while back? If so, then I need to worry about waiting for someone else to produce an x86 Linux version of version 0.52.

[rogue]

Originally Posted by mdettweiler

So...given these errors, is it safe to assume that all numbers that Phrot gives a "Klen>2" error for will be reported as composite regardless of whether they actually are prime? I too am running the Riesel base 3 project and get these errors sporadically throughout the files that I'm testing; however, I'm using Linux and the new, fixed builds are for Windows only. In fact, due to this I am still stuck using Geoff's Linux build of Phrot 0.51 from a little while back, i.e. the version without checkpointing or any of that stuff...

Rogue, has building a Linux binary for Phrot become easier since when you and I tried to build one on my machine a while back? If so, then I need to worry about waiting for someone else to produce an x86 Linux version of version 0.52.

The Klen>2 message has no relationship to this issue.

As for building a linux binary, no. I think that the problems you were experiencing are related to something in your environment, such as the compiler.

[mdettweiler]

Originally Posted by rogue

The Klen>2 message has no relationship to this issue.

As for building a linux binary, no. I think that the problems you were experiencing are related to something in your environment, such as the compiler.

Ah, I see. Well, as long as the Klen>2 message is unrelated to the missed PRP's, then I'll keep on using Phrot 0.51 for my testing until a Linux build for Phrot 0.52 becomes available.

*me nudges Geoff and/or Nugget

[geoff]

Originally Posted by mdettweiler

*me nudges Geoff and/or Nugget

I have uploaded phrot 0.52 MinGW and Linux executables here.

How do you get it to go 28% faster nugget?

[mdettweiler]

Originally Posted by geoff

I have uploaded phrot 0.52 MinGW and Linux executables here.

How do you get it to go 28% faster nugget?

Thanks a bunch!

[rogue]

I am releasing phrot version 0.60. There are a number of changes to this version, including:

• Renamed source as phrot.c
• Removed support for MODULAR_REDUCE_2 and MODULAR_REDUCE_3 as they weren't used. This code can still be found in glprov.c, which should be included with the distribution of phrot.
• Removed old history from phrot.c
• Support many more ABC file formats
• Support some NewPGen file format
• Default to -b=3 to always output LLR compatible residues.
• Added "-d" option. Without it, phrot will not output lowest limbs after PRP test.
• Added "-e" option to force error checking for all tests. Without -e phrot will only do error checking if k or b is a power of 2. This effectively removes the need for the error checking build of phrot.
• Add "make clean" to makefile
• Added a few new messages to catch issues that prevent phrot from performing a PRP test.

As part of the enhancements to support more ABC formats and some NewPGen formats, phrot can now be used with output from gcwsieve and from the SR5 project manual reservation page.

This distribution includes a Windows build that required CygWin.

If anyone runs into any problems, please let me know.

--Mark

[mdettweiler]

Originally Posted by rogue

I am releasing phrot version 0.60. There are a number of changes to this version, including:

• Renamed source as phrot.c
• Removed support for MODULAR_REDUCE_2 and MODULAR_REDUCE_3 as they weren't used. This code can still be found in glprov.c, which should be included with the distribution of phrot.
• Removed old history from phrot.c
• Support many more ABC file formats
• Support some NewPGen file format
• Default to -b=3 to always output LLR compatible residues.
• Added "-d" option. Without it, phrot will not output lowest limbs after PRP test.
• Added "-e" option to force error checking for all tests. Without -e phrot will only do error checking if k or b is a power of 2. This effectively removes the need for the error checking build of phrot.
• Add "make clean" to makefile
• Added a few new messages to catch issues that prevent phrot from performing a PRP test.

As part of the enhancements to support more ABC formats and some NewPGen formats, phrot can now be used with output from gcwsieve and from the SR5 project manual reservation page.

This distribution includes a Windows build that required CygWin.

If anyone runs into any problems, please let me know.

--Mark

Cool! These features should come in quite handy, especially the NewPGen file support and default -b=3, at least for me personally.

Now all that's left before I can use the new version is to wait for someone to post a Linux build...

[nuggetprime]

On my linux machine I'm getting LLR64=000000.. for all numbers.
Also,the checkpointing slows things down somewhat... I suggest you get stuck with my 0.51 binaries.

--nuggetprime

[mdettweiler]

Originally Posted by nuggetprime

On my linux machine I'm getting LLR64=000000.. for all numbers.
Also,the checkpointing slows things down somewhat... I suggest you get stuck with my 0.51 binaries.

--nuggetprime

Hmm...maybe it would be better to, say, have Phrot write out its checkpoint data every 5 minutes or so, sort of like LLR does, instead of at every iteration?

[rogue]

Originally Posted by mdettweiler

Hmm...maybe it would be better to, say, have Phrot write out its checkpoint data every 5 minutes or so, sort of like LLR does, instead of at every iteration?

It doesn't write it out for every iteration, only when the counter is updated.