Factor5 revamping

[ET_]

Hi all.

I decided to put my Mersenne factorization program (factor5) back online here.

I would like to add the nice feature of RES64 on the final result: would anybody help me teach how to achieve it?

Luigi

[Stargate38]

Any chance you could upload a Windows version? Not all OBD participants use Linux/Ubuntu.

[rogue]

Originally Posted by Stargate38

Any chance you could upload a Windows version? Not all OBD participants use Linux/Ubuntu.

It could theoretically be built using cygwin or minnow, but not Visual Studio, as it relies on pthreads.

[wombatman]

Originally Posted by rogue

It could theoretically be built using cygwin or minnow, but not Visual Studio, as it relies on pthreads.

What are the linux compilation instructions? I'll see if the Ubuntu shell can build it, which would be viable for all Windows 10 users.

[ET_]

Originally Posted by wombatman

What are the linux compilation instructions? I'll see if the Ubuntu shell can build it, which would be viable for all Windows 10 users.

I use

gcc factor5.c -lm -lgmp -lpthread -O3 -ofactor5

Cosider that OBD is stalling as the ranges have been "completed" by GPUs (by far faster than any CPU program) and the lone Mersenne hunters did the rest
Anyway I will put the program back online (probably on DoubleMersennes website) once you provide the Windows10 version, thank you!

[wombatman]

Unfortunately, it looks like there is an issue related to the size_t definition? Error log is attached.

GCC version is 4.8.4.

[rogue]

Change the line to this:

if (strcmp(argv[1], "--resume") != 0) {

Apparently your compiler doesn't like using "!" for an integer value.

[wombatman]

Still throwing size_t related errors (new error log is attached). Most likely culprit would seem to be this line:

In file included from /usr/include/stdio.h:33:0,
                 from factor5.c:21:
/usr/lib/gcc/x86_64-linux-gnu/4.8/include/stddef.h:212:1: error: expected identifier or ‘(’ before ‘typedef’
 typedef __SIZE_TYPE__ size_t;

Any thoughts?

[ET_]

Originally Posted by wombatman

Still throwing size_t related errors (new error log is attached). Most likely culprit would seem to be this line:

In file included from /usr/include/stdio.h:33:0,
                 from factor5.c:21:
/usr/lib/gcc/x86_64-linux-gnu/4.8/include/stddef.h:212:1: error: expected identifier or ‘(’ before ‘typedef’
 typedef __SIZE_TYPE__ size_t;

Any thoughts?

Yep.
Try adding the flag -std=c99 on the compile line

[monst]

I see the array... unsigned long int mexp[]
could have a few members added.

[ET_]

The array has been corrected in today's version 5.02

The problem is that when I tried to recompile the code, I had the same list of errors Ben had. The compilation for code v5.01 was successful less than 1 month ago, on both Intel and ARM platforms.

Something weird is happening under the hood...

I did not restart the PC after a series of updates, maybe some library needs a reboot?

Here is version 5.02 with the version update, the strcmp update and the array update.