Re: [OMPI devel] opal_init/finalize counter --> boolean

2011-07-08 Thread Jeff Squyres
I see several main categories of leaks at the current head (I did not test before Ralph's changes): - a bunch of RTE leaks --> which is think is what Ralph is trying to pare down - OMPI pre-defined attribute leaks --> This will take a little thinking to fix; it's complicated - OMPI pre-defi

Re: [OMPI devel] opal_init/finalize counter --> boolean

2011-07-08 Thread George Bosilca
On Jul 8, 2011, at 16:15 , Ralph Castain wrote: >> So we have opal_init * 1 and opal_util * 2. Clearly the opal util is not a >> simple ON/OFF stuff. With Ralph patch the OPAL utilities will disappear as >> soon as the OMPI layer call orte_fini. Luckily, today there is nothing >> between the c

Re: [OMPI devel] opal_init/finalize counter --> boolean

2011-07-08 Thread Ralph Castain
On Jul 8, 2011, at 7:37 AM, George Bosilca wrote: > What memory leaks? My valgrind claims there are basically none at the MPI > level, so I'm wondering ... We leak nearly a megabyte, almost all of it from OPAL, when running a tool. If you look at the commit message, this wasn't about an MPI pr

Re: [OMPI devel] opal_init/finalize counter --> boolean

2011-07-08 Thread George Bosilca
What memory leaks? My valgrind claims there are basically none at the MPI level, so I'm wondering ... First, let's make sure everyone understand why there are two initialization functions in OPAL. One has to call opal_init_util before anything else otherwise no access to utilities dealing with

[OMPI devel] opal_init/finalize counter --> boolean

2011-07-08 Thread Jeff Squyres
Developers -- Ralph and I talked about this one the other day; he's trying to plug some memory leaks during finalize. Before this commit, there was a counter that counts how many times opal_init*() is invoked. opal_finalize() decremented the counter, but didn't actually do anything else unt

Re: [OMPI devel] [OMPI users] huge VmRSS on rank 0 after MPI_Init when using "btl_openib_receive_queues" option

2011-07-08 Thread Eloi Gaudry
what i cannot understand is the reason why this extra memory would be initialized on proc 0 only. as far as i know, this doesn't make sense. éloi On 22/04/2011 08:52, Eloi Gaudry wrote: it varies with the receive_queues specification *and* with the number of mpi processes: memory_consumed =

Re: [OMPI devel] Fix a hang in carto_base_select() if carto_module_init() fails

2011-07-08 Thread nadia . derbey
Yes, sure! Agreed. Regards, -- Nadia Derbey Phone: +33 (0)4 76 29 77 62 devel-boun...@open-mpi.org wrote on 07/08/2011 02:10:22 AM: > De : Jeff Squyres > A : Open MPI Developers > Date : 07/08/2011 02:10 AM > Objet : Re: [OMPI devel] Fix a hang in carto_base_select() if > carto_module_init(