Open MPI logo

Open MPI Development Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Development mailing list

Subject: Re: [OMPI devel] [EXTERNAL] Re: RFC: Change ompi_proc_t endpoint data lookup
From: Barrett, Brian W (bwbarre_at_[hidden])
Date: 2013-07-22 12:22:51


On 7/22/13 9:19 AM, "David Goodell (dgoodell)" <dgoodell_at_[hidden]> wrote:

>On Jul 20, 2013, at 4:42 PM, "Barrett, Brian W" <bwbarre_at_[hidden]>
>wrote:
>
>> On 7/20/13 3:33 PM, "George Bosilca" <bosilca_at_[hidden]> wrote:
>>
>>> - In terms of memory this solution provide an approach where there
>>>will never be an extra overhead. The ompi_proc_t is not changed, and
>>>the extra array of endpoints is only created if the components that
>>>share it, are all loaded and enabled.
>>
>> I agree. Jeff and I talked about a similar concept, but the dependent
>>load was an idea crusher to me.
>
>I'm not really familiar with the code being discussed here, but could you
>insert a small fixed-size cache in front of this in order to mitigate
>this second load in the most common cases

That's essentially what I was proposing, except that the size of the cache
would be based on how many components claim they need a cache entry during
configure time. George seems to think that number will blow up. I think
that since we're adding a dynamic interface, we can use the RFC process
that's already in place to make sure it stays small.

Brian

--
  Brian W. Barrett
  Scalable System Software Group
  Sandia National Laboratories


  • application/pkcs7-signature attachment: smime.p7s