Hi, I struggled for days trying to get ET installed on Stampede2 (no problems on Stampede1, so some issue about the different environment, which is supposed to be KNL oriented). It worked 100% with very little intervention on my part for Bridges, compiling in about an hour. On Stampede2, after some suggestions from various people I managed to get all the object files and libraries built for all the thorns, but then it didn't like a couple of the arguments being passed to ld, which are hard to fix because it's not at all obvious how to modify the make-cycle. I could go into all the details of snags I hit for anyone who is interested. Anyways, as Roland suggests, trying a different resource might be a very good idea because the process seems rather sensitive the local environment. --Joel
-----Original Message----- From: users-bounces@einsteintoolkit.org [mailto:users-bounces@einsteintoolkit.org] On Behalf Of Roland Haas Sent: Friday, December 22, 2017 12:05 PM To: 林家暉 r06222015@ntu.edu.tw Cc: rhaas@ncsa.illinois.edu; users@einsteintoolkit.org Subject: Re: [Users] questions about compilation
Hello Chia Hui,
sorry for the long delay, I got caught up with Christmas preparations.
It also took several hours for me to compile the code which should be down in 20 min. Did the similar situation occur when Edison is tested last time ?
No I do not think that it took so long to compile on edison before. I am tempted to see if using the gnu compiler would be any faster. This is right now so painfully slow as to render the system almost unusable. it also does not help if one tries to compile in /dev/shm or from within an interactive job so the "usual" remedies for slow compilation do not apply.
Actually there are two other clusters I can use. But they are not supported by Einstein toolkit . I have tried to compile the code on them but failed since the required information for a new cluster is quite detailed and I cannot find all of them. Will you suggest me to shift to another cluster which is unsupported ?
If you want to use a different cluster then I would try and start from the files for a similar, working one. Eg for Cray machines the bluewaters-gnu files are a good starting point or maybe the cori ones (it being a newer cray).
We have a wiki page:
https://docs.einsteintoolkit.org/et-docs/Configuring_a_new_machine
that describes the procedure and would be happy to help out on the mailing list (or on the system if one of us has an account there).
Yours, Roland
-- My email is as private as my paper mail. I therefore support encrypting and signing email messages. Get my PGP key from http://keys.gnupg.net.