On Wed, Nov 08, 2006 at 12:25:20PM +, Miguel Figueiredo Mascarenhas Sousa
Filipe wrote:
> > Unless sharing data repeatedly leads you to false sharing and a loss
> > in performance.
>
> what does that mean.. I did not understand that.
Google indexes a bunch of good webpages on "false sharing
On Wed, Nov 08, 2006 at 09:57:18AM -0500, Hugh Merz wrote:
> The conventional wisdom of pure MPI being as good as hybrid models
> is primarily driven by the fact that people haven't had much incentive
> to re-write their algorithms to support both models.
Actually, I was thinking of apps where pe
On Wed, 8 Nov 2006, Larry Stewart wrote:
Miguel Figueiredo Mascarenhas Sousa Filipe wrote:
H
the MPI model assumes you don't have a "shared memory" system..
therefore it is "message passing" oriented, and not designed to
perform optimally on shared memory systems (like SMPs, or numa-CCs).
For
Miguel Figueiredo Mascarenhas Sousa Filipe wrote:
> H
>>> the MPI model assumes you don't have a "shared memory" system..
>>> therefore it is "message passing" oriented, and not designed to
>>> perform optimally on shared memory systems (like SMPs, or numa-CCs).
>>>
>> For many programs with
On 11/8/06, Miguel Figueiredo Mascarenhas Sousa Filipe
wrote:
Hi,
On 11/8/06, Greg Lindahl wrote:
> On Tue, Nov 07, 2006 at 05:02:54PM +, Miguel Figueiredo Mascarenhas Sousa
Filipe wrote:
>
> > if your aplication is on one given node, sharing data is better than
> > copying data.
>
> Unle
Hi,
On 11/8/06, Greg Lindahl wrote:
On Tue, Nov 07, 2006 at 05:02:54PM +, Miguel Figueiredo Mascarenhas Sousa
Filipe wrote:
> if your aplication is on one given node, sharing data is better than
> copying data.
Unless sharing data repeatedly leads you to false sharing and a loss
in perfo
How does Open MPI use shared memory for communication between two processes
on the same host meaning who creates the shared memory?
Both the processes running on the host have their own private buffers. So
buf1 sent by Process 1 should reach buf2 of Process 2. How is this handled?
Is it that buf
The quick answers are yes and nothing. Open MPI seamlessly support
shared memory communications when we detect that 2 processes are on
the same node. In fact Open MPI can use different communication
methods (read networks) between the processes of the same application.
Please read our FAQ f
Hi,
Thanks for the reply,
A few Additional questions,
1. Does OpenMPI has the optimisations required to ensure that when send/recv
is called between 2 ranks on the same node, the shared memory kind of
methods should be used?
2. If a programmer wants to implement such a logic(optimisations for l
On Tue, Nov 07, 2006 at 05:02:54PM +, Miguel Figueiredo Mascarenhas Sousa
Filipe wrote:
> if your aplication is on one given node, sharing data is better than
> copying data.
Unless sharing data repeatedly leads you to false sharing and a loss
in performance.
> the MPI model assumes you don
Hi,
On 11/7/06, Chevchenkovic Chevchenkovic wrote:
Hi,
I had the following setup:
Rank 0 process on node 1 wants to send an array of particular size to Rank
1 process on same node.
1. What are the optimisations that can be done/invoked while running mpirun
to perform this memory to memory tr
Chev
Interesting question; I too would like to hear about it from the experts in
this forum. However, off the top of my head, I have the following advise for
you.
Yes, you could share the memory between processes using the shm_xxx system
calls of unix. However, it would be a lot easier if you us
Hi,
I had the following setup:
Rank 0 process on node 1 wants to send an array of particular size to Rank
1 process on same node.
1. What are the optimisations that can be done/invoked while running mpirun
to perform this memory to memory transfer efficiently?
2. Is there any performance gain if
13 matches
Mail list logo
