[OMPI devel] [ofa-general] OpenMPI and RDMA-CM

[Jeff Squyres]

In the "FYI" category...

There was discussion about the udpal BTL over OFED today on the  
weekly developer teleconference (per my earlier post, a user is  
reporting that it doesn't work).  Andrew Friedley is going to work  
with the Sun developers -- he thinks he might know where the problem  
is coming from but is in process of physically relocating, and  
therefore couldn't look at it until late next week at the earliest.

Sun may be able to pick up the issue -- but if so, I don't know what  
their timeframe will be (and it may depend on the severity of the  
problem).


On May 8, 2007, at 9:47 AM, Steve Wise wrote:

> On Mon, 2007-05-07 at 20:39 -0400, Jeff Squyres wrote:
>> On May 7, 2007, at 6:52 PM, Steve Wise wrote:
>>
>>> Also, there appears to be a DAPL BTL in OMPI.  Is this BTL complete
>>> and
>>> enabled for the ofed-1.2 udapl library?
>>
>> Yes, it is complete and is well-tested in Solaris.
>>
>> It is not well tested in Linux/OFED (we've been concentrating on the
>> verbs interface on the OFED side of things -- the "openib" BTL [we
>> never renamed it when OpenIB changed names to OpenFabrics]).  In
>> fact, we've had scattered reports of it not working properly in  
>> Linux/
>> OFED, but those could well have been pilot error (i.e., me not trying
>> to run properly -- I know just about zilch about udapl).
>>
>
> The reason I'm asking is twofold:
>
> 1) this can get OMPI running on iwarp devices today if it works.
>
> 2) the udapl code can be a model for the rdma-cm piece, since the two
> are similary (client / server connection set, ipaddr/port based,  
> etc)...
>
> I'll try it out on T3.
>
>
> Steve.


-- 
Jeff Squyres
Cisco Systems