It looks like SDSC Rocks is shipping with Open MPI (through Sun Cluster tools). Who do we know at SDSC who works on Rocks? How do we get MPICH2 in there? -- Pavan -- Pavan Balaji http://www.mcs.anl.gov/~balaji
Is Phil Papadapoulos still there? Rusty On Tuesday,Nov 24, 2009, at 3:49 PM, Pavan Balaji wrote:
It looks like SDSC Rocks is shipping with Open MPI (through Sun Cluster tools). Who do we know at SDSC who works on Rocks? How do we get MPICH2 in there?
-- Pavan
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
Looks like he is: http://users.sdsc.edu/~phil/homepage.html -- Pavan On 11/24/2009 03:51 PM, Rusty Lusk wrote:
Is Phil Papadapoulos still there?
Rusty
On Tuesday,Nov 24, 2009, at 3:49 PM, Pavan Balaji wrote:
It looks like SDSC Rocks is shipping with Open MPI (through Sun Cluster tools). Who do we know at SDSC who works on Rocks? How do we get MPICH2 in there?
-- Pavan
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
-- Pavan Balaji http://www.mcs.anl.gov/~balaji
On Tue, Nov 24, 2009 at 03:49:33PM -0600, Pavan Balaji wrote:
It looks like SDSC Rocks is shipping with Open MPI (through Sun Cluster tools). Who do we know at SDSC who works on Rocks? How do we get MPICH2 in there?
If PVFS is any example, you don't want your software anywhere near ROCKS. we get a ton of bug reports from rocks users due to awfully mis-configured "rolls" of PVFS. (wow, i'm old... i only know "npaci rocks", which hasn't existed since 2004) ==rob -- Rob Latham Mathematics and Computer Science Division Argonne National Lab, IL USA
On 11/24/2009 04:27 PM, Rob Latham wrote:
On Tue, Nov 24, 2009 at 03:49:33PM -0600, Pavan Balaji wrote:
It looks like SDSC Rocks is shipping with Open MPI (through Sun Cluster tools). Who do we know at SDSC who works on Rocks? How do we get MPICH2 in there?
If PVFS is any example, you don't want your software anywhere near ROCKS. we get a ton of bug reports from rocks users due to awfully mis-configured "rolls" of PVFS.
(wow, i'm old... i only know "npaci rocks", which hasn't existed since 2004)
Based on our experience with binary packagers, it seems a lot easier to work with packagers, than users who don't understand what a configure failure that reads "C++ compiler not found" means. -- Pavan -- Pavan Balaji http://www.mcs.anl.gov/~balaji
participants (3)
-
Pavan Balaji -
Rob Latham -
Rusty Lusk