Prime95 beta version 28.3

[Prime95]

This was linked under Ubuntu 12.04.

Linux (or gcc) made it impossible to staticly link about a decade ago. IIRC, mprime was calling "dangerous" system routines like chdir.

[chalsall]

Originally Posted by Prime95

This was linked under Ubuntu 12.04.

Linux (or gcc) made it impossible to staticly link about a decade ago. IIRC, mprime was calling "dangerous" system routines like chdir.

Hmmmm...

Any chance you could upload the source, and I'll give compiling it statically a try?

Or, alternatively, I could give you (unprivileged) access to a CentOS 6.4 environment for you to work in.

[Mark Rose]

You should be able to call `gcc -static`. I use many pieces of software that are statically compiled.

[blip]

Works for me:
Ubuntu 13.10, Haswell i7-4702HQ
I can see a speed improvement of about 10% compared to v27.9

[Prime95]

Originally Posted by Mark Rose

You should be able to call `gcc -static`. I use many pieces of software that are statically compiled.

/usr/local/lib/libcurl.a(libcurl_la-netrc.o): In function `Curl_parsenetrc':
netrc.c:(.text+0x3a3): warning: Using 'getpwuid' in statically linked applications requires at runtime the shared libraries from the glibc version used for linking
/usr/local/lib/libcurl.a(libcurl_la-curl_addrinfo.o): In function `Curl_getaddrinfo_ex':
curl_addrinfo.c:(.text+0x6e): warning: Using 'getaddrinfo' in statically linked applications requires at runtime the shared libraries from the glibc version used for linking
prime.o: In function `pnHttpServer':
prime.c:(.text+0x17c82): warning: Using 'gethostbyname' in statically linked applications requires at runtime the shared libraries from the glibc version used for linking

[ewmayer]

Is it reasonable (I doubt it, but that's just my gut sense) to bundle specific glibc versions with each binary?

Or is it perhaps time to switch to a make-based approach and release sources/makefiles only?

{With my own code I have gone to pains to make builds as simple as possible, but I don't have all the primeNet/InterTubes-comms stuff to deal with.)

[chalsall]

Originally Posted by ewmayer

Or is it perhaps time to switch to a make-based approach and release sources/makefiles only?

You forget that a small amount of the Prime95/mprime code is "secret".

[ewmayer]

Originally Posted by chalsall

You forget that a small amount of the Prime95/mprime code is "secret".

Perhaps that aspect needs to be changed to a public-key model. But that would be a longer-term change.

[IBethune]

Hi George, is there a (planned) source release of gwnum 28.3? Didn't find it in the FTP directory at the moment. We are obviously keen to build a new version of LLR for testing at PrimeGrid.

Thanks for all your good work as always!

Cheers

- Iain

[Prime95]

Originally Posted by IBethune

Hi George, is there a (planned) source release of gwnum 28.3?

Let's wait for any bug reports. I'll make the source for 28.4 available then. I've already found an optimization to pass 1 length 768 which will help the performance of a few FFT lengths.

[Prime95]

Originally Posted by petrw1

Instead of one setting for # of iterations between screen updates...could it be either:
- # of minutes or hours between updates
Or
- allow a diff # of iterations for eack worker.

I'll see what I can come up with. It will be an undoc.txt feature.