I don't recall ever having an ET call where this was discussed. Should it be scheduled?
Cheers, Steve
On 10/23/2013 11:23 AM, Erik Schnetter wrote:
I believe these benchmark parameter files are wildly out of date and should be removed.
These days, we auto-generate benchmark files based on the target system to ensure that the benchmark parameters are scaled correctly. I have some script in Simfactory that do this. There is a low-level interface to the scripts where you (unfortunately) need to set many parameters (nodes, cores, processes, threads, SMT threads, memory to be used, CPU time to be used, etc.), and a high-level interface where you (unfortunately) basically cannot set anything since these are all chosen automatically... I should give an overview in person; would next Monday during the ET call make sense?
-erik
On 2013-10-23, at 11:40 , Roland Haas roland.haas@physics.gatech.edu wrote:
Signed PGP part Hello Steve,
here the ones I know of:
BenchIO BSSN_MoL
this is the CCATIE code. AEI private and not to be used anymore as far as I know (superseded by MacLachlan).
FlexIO IOFlexIO
Old IO libraries. Used only with PUGH.
whisky_init_data
this is now GRHydro_InitData
whisky
this is now GRHydro
whisky_rnsid
this uses a private version of RNSID (original RNSID stand alone code by Nick Stergioulas then extended). Rumour (and me talking to Nick when I met him accidentally in Germany) has it that a new public version of RNSID that contains (most of?) these features is being prepared. Frank would know I think.
Are these out of date? Are there better files to use for benchmarking?
Mostly they are out of data I'd say. For benchmarking I'd use the tov and vacuum files from the xirel project. Alternatively I think Erik send around some of his benchmarking files a while ago.
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.
Users mailing list Users@einsteintoolkit.org http://lists.einsteintoolkit.org/mailman/listinfo/users
Users mailing list Users@einsteintoolkit.org http://lists.einsteintoolkit.org/mailman/listinfo/users