Compiling Phrot - Page 13

rogue · 2008-12-23, 02:46

Quote:

Originally Posted by mdettweiler

Well, this range had actually been run with 0.61 from the beginning--so, any and all checkpoint/phrot.ini files had been saved by 0.61.

If I remember correctly, yes, the LastLine= value was a little off in 0.61, but in 0.52 it's right on.

Are you certain that you were using 0.61, not 0.60? I spent a lot of time addressing this in 0.60, thus the 0.61 release. It is possible I missed a case. Did LastLine point to the last completed test of your input file or not?

mdettweiler · 2008-12-23, 02:52

Quote:

Originally Posted by rogue

Are you certain that you were using 0.61, not 0.60? I spent a lot of time addressing this in 0.60, thus the 0.61 release. It is possible I missed a case. Did LastLine point to the last completed test of your input file or not?

I'm sure that I was using 0.61. However, I don't quite remember exactly what LastLine pointed to. I'll run some tests on it later tonight and report back here with the results.

geoff · 2008-12-23, 03:24

Quote:

Originally Posted by mdettweiler

I'm sure that I was using 0.61. However, I don't quite remember exactly what LastLine pointed to. I'll run some tests on it later tonight and report back here with the results.

If you are using the binary from my site then it could be that I didn't compile it from the latest source. I'll check and upload another binary tomorrow if necessary.

mdettweiler · 2008-12-23, 03:55

Quote:

Originally Posted by geoff

If you are using the binary from my site then it could be that I didn't compile it from the latest source. I'll check and upload another binary tomorrow if necessary.

Yes, I am using the binary from your site--thanks.

geoff · 2008-12-24, 00:35

Quote:

Originally Posted by mdettweiler

Yes, I am using the binary from your site--thanks.

I have just uploaded a new binary, the old one was built from the wrong source and didn't update phrot.ini properly.

mdettweiler · 2008-12-24, 00:58

Quote:

Originally Posted by geoff

I have just uploaded a new binary, the old one was built from the wrong source and didn't update phrot.ini properly.

Thanks! I'll try it out sometime tonight.

mdettweiler · 2008-12-24, 03:06

Quote:

Originally Posted by mdettweiler

Thanks! I'll try it out sometime tonight.

Okay, I've checked it out, and all seems well. Thanks!

thommy · 2009-01-05, 17:39

I'm awaiting the PRPnet server for srb5. Nice work, Mark.

geoff · 2009-01-09, 02:41

For anyone who has been using the Windows Phrot executable I posted at http://www.geocities.com/g_w_reynolds/phrot/ I am sorry to report that versions 0.52 to 0.62 had a fault that caused the PRP residue to be printed incorrectly.

If you are using the Windows executable from one of these versions then please download version 0.63 to replace it.

The fault didn't affect the calculation of the residue itself, only the printing, so there is no need to fear that a PRP would have been missed. But it means that the first 8 digits of the PRP residue will not match a double-check.

The Linux executables were not affected, and the Cygwin Windows executable in the original source doesn't suffer from this fault, just the one posted on my site above.

mdettweiler · 2009-01-09, 03:07

I was just testing Geoff's Linux build of Phrot 0.63 on some Phrot numbers a moment ago, and they don't match up with the respective LLR residuals, even when I specify -b=3 on the command line:

Code:

Phrot:
3*2^12348+1 is composite LLR64=05f1d8bfc1ff6e88. (t=0.69s)
LLR:
3*2^12348+1 is not prime.  Proth RES64: 5DBFF15E5EF169ED  Time : 263.788 ms.

Yet, it matches the residual produced by PRP:

Code:

3*2^12348+1 is not prime.  RES64: 05F1D8BFC1FF6E88.  OLD64: 11D58A3F45FE4B95  Time: 248.767 ms.

Possibly there's some bug in Phrot that's keeping it from "knowing" when to do a Proth test on k*2^n+1 numbers, and thus it's doing a PRP test instead, like it would do for any other number?

rogue · 2009-01-09, 03:42

Quote:

Originally Posted by mdettweiler

I was just testing Geoff's Linux build of Phrot 0.63 on some Phrot numbers a moment ago, and they don't match up with the respective LLR residuals, even when I specify -b=3 on the command line:

Code:

Phrot:
3*2^12348+1 is composite LLR64=05f1d8bfc1ff6e88. (t=0.69s)
LLR:
3*2^12348+1 is not prime.  Proth RES64: 5DBFF15E5EF169ED  Time : 263.788 ms.

Yet, it matches the residual produced by PRP:

Code:

3*2^12348+1 is not prime.  RES64: 05F1D8BFC1FF6E88.  OLD64: 11D58A3F45FE4B95  Time: 248.767 ms.

Possibly there's some bug in Phrot that's keeping it from "knowing" when to do a Proth test on k*2^n+1 numbers, and thus it's doing a PRP test instead, like it would do for any other number?

See here

2009-01-05, 17:39	#140
thommy Dec 2006 3·11 Posts	I'm awaiting the PRPnet server for srb5. Nice work, Mark. Last fiddled with by thommy on 2009-01-05 at 17:39

2009-01-09, 03:07	#142
mdettweiler A Sunny Moo Aug 2007 USA (GMT-5) 3×2,083 Posts	I was just testing Geoff's Linux build of Phrot 0.63 on some Phrot numbers a moment ago, and they don't match up with the respective LLR residuals, even when I specify -b=3 on the command line: Code: Phrot: 32^12348+1 is composite LLR64=05f1d8bfc1ff6e88. (t=0.69s) LLR:* 32^12348+1 is not prime. Proth RES64: 5DBFF15E5EF169ED Time : 263.788 ms. Yet, it matches the residual produced by PRP: Code: 32^12348+1 is not prime. RES64: 05F1D8BFC1FF6E88. OLD64: 11D58A3F45FE4B95 Time: 248.767 ms. Possibly there's some bug in Phrot that's keeping it from "knowing" when to do a Proth test on k*2^n+1 numbers, and thus it's doing a PRP test instead, like it would do for any other number?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
PRPNET & Phrot discussion	masser	Sierpinski/Riesel Base 5	27	2010-09-08 03:10
Phrot announcements	rogue	Conjectures 'R Us	33	2010-01-22 19:39
LLR/PRP/phrot/pfgw:- Pick A Range	masser	Sierpinski/Riesel Base 5	8	2009-08-18 19:44
Using Phrot For LLRNET Reservations	thommy	Sierpinski/Riesel Base 5	1	2008-11-03 14:53
Programmer needed to write networked Phrot app!	mdettweiler	Programming	0	2008-04-07 21:25

2009-01-09, 02:41	#141
geoff Mar 2003 New Zealand 13·89 Posts	For anyone who has been using the Windows Phrot executable I posted at http://www.geocities.com/g_w_reynolds/phrot/ I am sorry to report that versions 0.52 to 0.62 had a fault that caused the PRP residue to be printed incorrectly. If you are using the Windows executable from one of these versions then please download version 0.63 to replace it. The fault didn't affect the calculation of the residue itself, only the printing, so there is no need to fear that a PRP would have been missed. But it means that the first 8 digits of the PRP residue will not match a double-check. The Linux executables were not affected, and the Cygwin Windows executable in the original source doesn't suffer from this fault, just the one posted on my site above.