Re: CCL:gaussian 03 on macintosh

 dear all,
MacOSX is just another regular UNIX and one can compile the standard g03 source code on that machine (and optimize for the G5!). I did so and it runs very fast. On a typical DFT job, it is 2-3 times (yes, that much!) faster than on a Xeon (comparing a single 2 GHz Xeon CPU to a single 2 GHz G5 CPU). For visualization, I use mostly viewmol, an excellent program for looking at molecules, vibrations, orbitals.... Added advantage: its free. 
 There are a couple of points to consider if you want to compile g03 on MacOSX:
I used the IBM xlf compiler (8.1 beta). The final version of the compiler is available now, I am about to order that one. It order to compile, you need a UNIX filesystem partition on the disk where g03 is compiling, as hfs and hfs+ are not really case sensitive and in the preprocessor stage of the compilation (cpp), it needs to distinguish between xxxx.F and xxxx.f. The makefile is a hybrid of the rs6k.make and the linux makefile. I also added a few lines of fortran code and slightly modified some c source files (only on some files in the bsd subdir). Of course one has to keep in mind that the byteorder on a PowerPC is different from Intel and that, although the CPU is 64 bit, MacOSX is not (no support for 8 byte pointers). I also used the Apple BLAS (veclib).
The resulting code runs very well, 16 GB scratchfiles are no problem (no need to segment into 2 GB chunks). Bigger scratchfiles do not work, I guess the reason are the 4 byte pointers.
I tried to compile it for 2 CPUs (using openMP). That did not work and since I don't care, I gave up quickly. My feeling is that after spending some time, that would work as well. 
 Good luck,
 Gert
 At 16:53 +0200 7/4/04, Andreas Serr wrote:
 
 Dear all,
 we are considering using Gaussian 03 on a Mac OS X
 system. Has anybody experience in that field? In
 particular:
 - G03M is not optimised for Mac G5 architecture, so
 does it make sense to run it on G5 rather than on G4?
 - is GaussView M working fine with G5 computers?
 - does anyone have experience/benchmark tests for
 Gaussian running on Mac (G5/G4/...) in comparison
 with Gaussian on Intel Xeon-dual process.  #*at*#  2 GHz
 (Linux system)
 Regards,
 Andy
 andreas.serr~at~uni-bielefeld.de
 -= This is automatically added to each message by the mailing script =-
 To send e-mail to subscribers of CCL put the string CCL: on your Subject: line
 and send your message to:  CHEMISTRY~at~ccl.net
 Send your subscription/unsubscription requests to: CHEMISTRY-REQUEST~at~ccl.net
 HOME Page: http://www.ccl.net
 | Jobs Page: http://www.ccl.net/jobs
 If your mail is bouncing from CCL.NET domain send it to the maintainer:
 Jan Labanowski,  jlabanow~at~nd.edu (read about it on CCL Home Page)
 -+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
 
 --
 -------------------------------------
 Gert von Helden
 Fritz-Haber-Institut der Max-Planck-Gesellschaft
 Faradayweg 4-6
 14195 Berlin
 tel.: +49-30-8413 5615
 fax: +49-30-8413 5603