Jeff Squyres wrote:
> On Apr 24, 2007, at 10:53 AM, Bert Wesarg wrote:
>
>>> Is it possible to have less than all of the (socket,core)
>>> information, the node ID information, and HTT information? Do we
>>> need to make the have_topo_information() function a bit more fine-
>>> grained? (i.e., indicate which, if any, of the topology information
>>> is available) And/or make *some* of the functions fail (gracefully)
>>> if the requested information is unavailable?
>> I think we should emulate missing information:
>>
>> * no node info: put all processor ids into node0
>> * no thread info: only one thread
>> :
>> :
>
> Hmm. Is that really a good idea? I'd think it was safer to say "I
> don't know" rather than "here's a fallback which may or may not be
> true."
Loud thinking:
What is the worst case that could happen, if we have no thread
information, aka the file thread_siblings is missing:
There are threads, so for two cpu ids cpu_i, cpu_j there should be two
tuple with (x, y, a) and (x, y, b) where a != b. So we end up with a
mapping from cpu_i and cpu_j to (x, y, 0).
So the best way is here to return topo unsupported. the best thing is to
catch this thing in the cache building.
Idea:
check in the compute map, that the mapping is indeed 1:1.
Bert
|
![[Flag]](../../../../../community/mirrors/flag-us.gif){kind=small}