Re: [PATCH] perf: make perf.data more self-descriptive (v3)

[Stephane Eranian]

On Wed, Sep 7, 2011 at 9:28 PM, Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
> On Wed, 2011-09-07 at 14:55 +0200, Stephane Eranian wrote:
>> On Wed, Sep 7, 2011 at 10:12 AM, Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
>> > On Tue, 2011-09-06 at 22:46 +0200, Stephane Eranian wrote:
>> >> +static const char *topo_fmt[] = {
>> >> + Â Â Â "/sys/devices/system/cpu/cpu%d/topology/core_siblings_list",
>> >> + Â Â Â "/sys/devices/system/cpu/cpu%d/topology/thread_siblings_list",
>> >> + Â Â Â NULL
>> >> +};
>> >
>> > No nodes?
>> >
>> To record what information? core/thread siblings or memory stuff?
>
> Dunno really,.. but nodes are part of the topology and with people
> interested in all this uncore business it might make sense to have that.
> Maybe even record the node-distance table so you can reconstruct the
> interconnect topology.
>
>
I posted v4 which captures the numa topology (meminfo, cpulist). When you
combine with what I already had, you get the complete picture.
>
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/