Hi all,

Thanks for the info! I think it's valuable to keep the documentation up to date on this front, so I've gone ahead and added Steele to the documentation as well as to the warning message in the install_script.sh. It's really important to me that yt be easy to install, even though it sits on top of a couple different packages, so all this feedback is very helpful. If anyone else has tips about machines that might have tricky installation procedures, please let me know -- I'm happy to put notices in the install_script and the documentation about any modules, commands, etc.

And, John, you're a step up on me -- I always ended up giving up around the time numpy came around in the stack when trying to compile with Intel!

-Matt

On Thu, Feb 11, 2010 at 1:24 PM, John Wise jwise@astro.princeton.edu wrote:

In general, I've found that loading the same compiler module as the one used to compile Python reduces headaches.  For instance, I've compiled Python+numpy+yt+etc with the Intel compiler, but I wouldn't recommend it after going through it...

John

On 02/11/2010 04:15 PM, gso@physics.ucsd.edu wrote:

...

Yes a similar thing should be done on Verne or Kraken at NICS or perhaps any cluster, do a module swap from the pgi or intel (or other) compiler to gnu compiler before trying to compile YT.

From G.S.

...

Hi Matt,

I managed to install and run yt1.6 successfully on Steele. I had to do the following...

module purge (to get rid of the mpi amd intel modules) module load gcc  (this loaded gcc4.2.1)

Just in case anybody wanted to know how to get yt1.6 running on Steele.

Thanks a lot.

Shankar KU Cosmology

---

yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org

---

yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org

---

yt-users mailing list yt-users@lists.spacepope.org http://lists.spacepope.org/listinfo.cgi/yt-users-spacepope.org