PFGW 4.0.4 (with gwnum v30.10) Released

[rogue]

I have posted pfgw 4.0.4 over at sourceforge. Here are the updates:

   - Upgraded to Visual Studio 2022.
   - Added -A allowing the specification of single CPU for pfgw to use.  This
     is supported only for Windows at this time and should be primarily used on CPUs
     with a mix of efficiency and performance cores.
   - Added CC file format support.  This was added to support ccsieve (from the mtsieve
     framework) as the format is easier to understand than the newpgen format.  Read
     the ccfileformat.txt for more details.
   - Upgraded to gwnum v30.10b2.

I intend to post a Mac build later today. Can someone provide a linux build?

[hunson]

Hi,

I tested the new version of pfgw64 (4.0.4) and found a behavior that I can not explain when using the new cunningham chain input file from ccsieve.

1125262125*2^41003+1 is 3-PRP! (0.7271s+0.0003s)
1395485655*2^41003+1 is 3-PRP! (0.7270s+0.0002s)
2*(1395485655*2^41003+1)-1 is composite: RES64: [7B5D0590700BDD19] (0.7327s+0.0008s)
1462023480*2^41003+1 is 3-PRP! (0.7945s+0.0003s)
1488224595*2^41003+1 is 3-PRP! (0.7671s+0.0003s)
1541048355*2^41003+1 is 3-PRP! (0.7321s+0.0002s)
1895788596*2^41003+1 is 3-PRP! (0.7264s+0.0002s)
2*(1895788596*2^41003+1)-1 is composite: RES64: [2405F0A43800E1AB] (0.7328s+0.0007s)
1962067611*2^41003+1 is 3-PRP! (0.7419s+0.0002s)

Why does pfgw explicitly print out that 2*(1395485655*2^41003+1)-1 and 2*(1895788596*2^41003+1)-1 are composite? I assume the software always checks for this form if a prime was found like in all the other cases.

regards

[rogue]

Originally Posted by hunson

Hi,

I tested the new version of pfgw64 (4.0.4) and found a behavior that I can not explain when using the new cunningham chain input file from ccsieve.


Why does pfgw explicitly print out that 2*(1395485655*2^41003+1)-1 and 2*(1895788596*2^41003+1)-1 are composite? I assume the software always checks for this form if a prime was found like in all the other cases.

regards

The first term is PRP. The second term in the chain is composite. If you have length 3 and the first two are PRP, but the third is composite, it will output all three. It will not output the first term in the chain if it is composite.

Note that with base 2 that k can be even for the first time. That will be fixed in the next release. If you want to find records for longer chains then search primorials or factorials. fewer terms are removed in each range of k, but there are a lot more PRPs.

[hunson]

Thanks for the explanation.

[ryanp]

Originally Posted by rogue

I intend to post a Mac build later today. Can someone provide a linux build?

I was able to successfully build PFGW Version 4.0.4.64BIT.20221214.x86_Dev [GWNUM 30.10] on Linux, though I was initially getting linker errors working from SVN head. I had to tweak pform/pfio/makefile and pform/pfio/pfccfile.cpp as follows:

Index: pform/pfio/makefile
===================================================================
--- pform/pfio/makefile	(revision 679)
+++ pform/pfio/makefile	(working copy)
@@ -5,7 +5,7 @@
 vpath %.asm .
 
 OSPECIAL =
-OLIB = pfcpapfile.o pffile.o pfnewpgenfile.o pfoutput.o pfconsoleoutput.o pfini.o \
+OLIB = pfcpapfile.o pffile.o pfccfile.o pfnewpgenfile.o pfoutput.o pfconsoleoutput.o pfini.o \
 	pfabcfile.o pfabc2file.o pfprzfile.o pfscriptfile.o pfscriptfns.o pfabcdfile.o \
 	pfdecimalfile.o pfcheckfile.o prz.o pfabctaskcompleted.o pffilesymbol.o
 OUTDIR = .libs

Index: pform/pfio/pfccfile.cpp
===================================================================
--- pform/pfio/pfccfile.cpp	(revision 679)
+++ pform/pfio/pfccfile.cpp	(working copy)
@@ -27,6 +27,13 @@
 extern char g_ModularSieveString[256];
 extern bool g_bTerseOutput;
 
+int strpos(char *haystack, int needle) {
+   char *p = strchr(haystack, needle);
+   if (p)
+      return p - haystack;
+   return -1;   // Not found = -1.
+}
+
 int PFCCFile::LetterNumber(char Letter)
 {
    if (Letter >= 'A' && Letter <= 'Z')
@@ -192,7 +199,7 @@
    while (expPtr[0] == ' ')
       expPtr++;
 
-   if (strchr(expPtr, '&') > 0 || strchr(expPtr, '|') > 0) {
+   if (strpos(expPtr, '&') > 0 || strpos(expPtr, '|') > 0) {
       snprintf(errorMessage, sizeof(errorMessage), "\nCritical Error, first line in CC file does not support & or |\n");
       throw errorMessage;

@rogue, how should I get the binary to you?

[rogue]

Originally Posted by ryanp

I was able to successfully build PFGW Version 4.0.4.64BIT.20221214.x86_Dev [GWNUM 30.10] on Linux, though I was initially getting linker errors working from SVN head. I had to tweak pform/pfio/makefile and pform/pfio/pfccfile.cpp as follows:

Index: pform/pfio/makefile
===================================================================
--- pform/pfio/makefile	(revision 679)
+++ pform/pfio/makefile	(working copy)
@@ -5,7 +5,7 @@
 vpath %.asm .
 
 OSPECIAL =
-OLIB = pfcpapfile.o pffile.o pfnewpgenfile.o pfoutput.o pfconsoleoutput.o pfini.o \
+OLIB = pfcpapfile.o pffile.o pfccfile.o pfnewpgenfile.o pfoutput.o pfconsoleoutput.o pfini.o \
 	pfabcfile.o pfabc2file.o pfprzfile.o pfscriptfile.o pfscriptfns.o pfabcdfile.o \
 	pfdecimalfile.o pfcheckfile.o prz.o pfabctaskcompleted.o pffilesymbol.o
 OUTDIR = .libs

Index: pform/pfio/pfccfile.cpp
===================================================================
--- pform/pfio/pfccfile.cpp	(revision 679)
+++ pform/pfio/pfccfile.cpp	(working copy)
@@ -27,6 +27,13 @@
 extern char g_ModularSieveString[256];
 extern bool g_bTerseOutput;
 
+int strpos(char *haystack, int needle) {
+   char *p = strchr(haystack, needle);
+   if (p)
+      return p - haystack;
+   return -1;   // Not found = -1.
+}
+
 int PFCCFile::LetterNumber(char Letter)
 {
    if (Letter >= 'A' && Letter <= 'Z')
@@ -192,7 +199,7 @@
    while (expPtr[0] == ' ')
       expPtr++;
 
-   if (strchr(expPtr, '&') > 0 || strchr(expPtr, '|') > 0) {
+   if (strpos(expPtr, '&') > 0 || strpos(expPtr, '|') > 0) {
       snprintf(errorMessage, sizeof(errorMessage), "\nCritical Error, first line in CC file does not support & or |\n");
       throw errorMessage;

@rogue, how should I get the binary to you?

I'm curious what library is missing. I will update sourceforge.

You can send it to me via e-mail or dropbox. I just need you to zip (or 7-zip) the contents of the distribution folder.

Thank you!

[rogue]

I updated the makefile and fixed strchr to use != NULL. Let me know if that resolves your build issue.

[kruoli]

Do you already have received the Linux version? When building and taking the gwnum.a from the latest source, it still complains about a lot of missing references. Even to gwinit2.

[rogue]

Originally Posted by kruoli

Do you already have received the Linux version? When building and taking the gwnum.a from the latest source, it still complains about a lot of missing references. Even to gwinit2.

I cannot build a Linux version. Nobody has sent me one to post.

[kruoli]

Attached is a Linux build. I needed to change GMP from being statically to being dynamically linked. Please give it a try.

[rogue]

Originally Posted by kruoli

Attached is a Linux build. I needed to change GMP from being statically to being dynamically linked. Please give it a try.

I have no means to execute this. If someone else can verify that this runs in their environment, then I will put on sourceforge. My main concern is the dynamically linked libraries.