Re: [bgl-discuss] MPI failure, simplified

[Darius Buntinas <buntinas@xxxxxxxxxxx>]

They're saying you need to guarantee that every send has a preposted 
receive before the send call is made.  Isn't that what ready-sends are 
for?

I'm assuming that someone over there has spent more time thinking about 
this issue than I have (~10 min or so), but it seems to me that they could 
be less aggressive with sends and isends by implementing a rendezvous 
protocol for larger messages, especially considering the BGL memory 
constraints.  Then they can be as aggressive as they want with 
ready-sends, where they can enforce preposted receives.

Just a thought.
-d

On Tue, 21 Feb 2006, Pete Beckman wrote:

> > Are they saying that 80MB of data are sent in eager mode?
>
> That's what I understood from their email.
>
> "Our implementation is perhaps too agressive in assuming there is enough 
> memory to accomodate the data"
>
> -Pete

- --------------------------------------------------------------------
To add or remove yourself from this mailing list, use the 'notifyme'
command on any BGL machine. To remove: notifyme -n, to add: notifyme -y.