Home > IBM > IBM developerWorks Open Source Compiler Farm IBM developerWorks Open Source Compiler Farm Submitted by Jeff 2002-02-28 IBM 5 Comments Members of developerWorks open source projects can now compile their code on the first compiler farm in the open source community that runs on an IBM eServer zSeries server. The zSeries is the world’s first dedicated Linux mainframe server. About The Author Eugenia Loli Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker. Follow me on Twitter @EugeniaLoli 5 Comments 2002-02-28 7:13 am hmm.. i think i just had a neat idea.. you know how distributed computing things work.. like seti at home. mabey they could do that and compile stuff for people 2002-02-28 9:39 am hmm.. I think you gave me a neat idea.. you realise how much crunch that server farm probably has? Guessing loads. Anyway, write your hard problem as # precompiler commands and get the farm to do the crunching… http://www.ioccc.org/years.html#1995_vanschnitz comes to mind 😀 2002-02-28 2:11 pm Wouldnt different processor types affect the compile? If one node is a pentium, and another is an athlon…while they are ‘compatible’ they can be optimized differently, and im sure 3DNow! code wouldnt go over well on the Pentium based proc. Or even if you go above that… if the farm has Alphas or something, you would get screwed again no? Well, of course if every node has a cross compiler, but then you start getting dirty. It would be neat if it would require very little more software than what most people already have. 2002-02-28 5:57 pm this is a mainframe with many linux images running on it. 2002-02-28 6:05 pm Wouldnt different processor types affect the compile? If one node is a pentium, and another is an athlon…while they are ‘compatible’ they can be optimized differently, and im sure 3DNow! code wouldnt go over well on the Pentium based proc That’s why those optimizations are options in most compilers, and which optimizations are compiled into the program have nothing to do with what processor is being used to compile the program in the first place (you can compile a program with P4 optimizations to run under Windows using a compiler on a Mac under OS X if you really wanted to, you just couldn’t run the resulting executable without emulation). The most important part of writing a distributed compiler, though, would definitely be keeping track of the options and making it fairly easy for the person submitting the program for compilation to set those options according to what they need.