Jeff Squyres, le Thu 22 Sep 2011 21:36:31 +0200, a écrit :
> (SIDENOTE: reading this letter after I wrote it, I see that it sounds
> overwhelmingly negative. Please do not take it that way!
Sure! If these are the only negative points of our API, it's actually a
very positive mail ;)
> 1. The de
Hello,
Sorry for coming late in the thread, I've been busy at conference,
traveling, teaching, moving between flats :)
Brice Goglin, le Thu 22 Sep 2011 22:25:33 +0200, a écrit :
> > (2d) more specifically: since all caches are of type HWLOC_OBJ_CACHE, we
> > find ourselves putting in special cas
On Sep 24, 2011, at 8:43 AM, Jeff Squyres wrote:
> On Sep 24, 2011, at 10:36 AM, Ralph Castain wrote:
>
>> Looks good - thx!
>>
>> Pointing to specific sections that correlate to specific code is important
>> if you intend to rely on the pdf doc for people to understand the code. Not
>> many
On Sep 24, 2011, at 10:36 AM, Ralph Castain wrote:
> Looks good - thx!
>
> Pointing to specific sections that correlate to specific code is important if
> you intend to rely on the pdf doc for people to understand the code. Not many
> of us have the time to read a 100+ page doc just to figure o
Looks good - thx!
Pointing to specific sections that correlate to specific code is important if
you intend to rely on the pdf doc for people to understand the code. Not many
of us have the time to read a 100+ page doc just to figure out how to use an
accessor function.
On Sep 24, 2011, at 8:2
How's this?
https://svn.open-mpi.org/trac/hwloc/changeset/3853
On Sep 24, 2011, at 10:01 AM, Ralph Castain wrote:
>
> On Sep 24, 2011, at 5:52 AM, Jeff Squyres wrote:
>
>> On Sep 24, 2011, at 7:46 AM, Jeff Squyres wrote:
>>
The funky thing here is that the parent/child links between
On Sep 24, 2011, at 5:52 AM, Jeff Squyres wrote:
> On Sep 24, 2011, at 7:46 AM, Jeff Squyres wrote:
>
>>> The funky thing here is that the parent/child links between the first
>>> socket and its core go across level 2 because nothing matches there. In
>>> the first socket, you have Socket(depth1
On Sep 24, 2011, at 7:46 AM, Jeff Squyres wrote:
>> The funky thing here is that the parent/child links between the first
>> socket and its core go across level 2 because nothing matches there. In
>> the first socket, you have Socket(depth1)->Core(depth3) while in the
>> second socket you have Soc
On Sep 22, 2011, at 5:05 PM, Brice Goglin wrote:
> No, the level always contain all elements of the same type (+depth for
> caches), even if they are not at the same "distance" to the root (not
> "depth").
>
> Let's say you have two single-core sockets. One with no cache. One with
> a L1.
> What
On Sep 22, 2011, at 3:05 PM, Brice Goglin wrote:
> Le 22/09/2011 22:42, Ralph Castain a écrit :
>> I guess I didn't get that from your documentation. Since caches sit
>> between socket and core, they appear to affect the depth of the core
>> in a given socket. Thus, if there are different numbers
Le 22/09/2011 22:42, Ralph Castain a écrit :
> I guess I didn't get that from your documentation. Since caches sit
> between socket and core, they appear to affect the depth of the core
> in a given socket. Thus, if there are different numbers of caches in
> the different sockets on a node, then th
On Sep 22, 2011, at 2:25 PM, Brice Goglin wrote:
> Le 22/09/2011 21:36, Jeff Squyres a écrit :
>> 1. The depth-specific accessors are Bad. Given the warning language in the
>> docs paired with the practical realities that some people actually do mix
>> and match CPUs in a single server (especi
Le 22/09/2011 21:36, Jeff Squyres a écrit :
> 1. The depth-specific accessors are Bad. Given the warning language in the
> docs paired with the practical realities that some people actually do mix and
> match CPUs in a single server (especially when testing new chips), the
> depth-based accesso
We're integrating hwloc deeper into Open MPI, and are also using it in some
other projects. As a result, I have some practical feedback.
(SIDENOTE: reading this letter after I wrote it, I see that it sounds
overwhelmingly negative. Please do not take it that way! We love the use of
hwloc in
14 matches
Mail list logo
