On Thursday 2009-12-03 23:02, David Miller wrote:
>> On Wednesday 2009-12-02 23:57, David Miller wrote:
>>>The T1 cores are repesented as SMT units.
>>
>> It seems so.
>>
>> /sys/devices/system/cpu/
>> cpu0/topology/physical_package_id:0
>> cpu1/topology/physical_package_id:0
>> cpu10/topology/physical_package_id:2
>> cpu11/topology/physical_package_id:2
>> cpu12/topology/physical_package_id:3
>> cpu13/topology/physical_package_id:3
>> cpu14/topology/physical_package_id:3
>> cpu15/topology/physical_package_id:3
>> cpu16/topology/physical_package_id:4
>> ...
>>
>> Why is this done, when they are, in fact, not multiple physical
>> packages?
>
>Because I need two levels of grouping to represent chips like T2.
>
>I use SMT to represent the cores.
>
>And I use MC to represent the 2 integer units within a core on T2.
Since 2.6.32 was recently released, I thought I give that a try (the
2.6.31 state carrying my Netfilter changes was throwing a recursive
fault on sparc), and the scheduling toplogy has changed slightly:
CPU0 attaching sched-domain:
domain 0: span 0-3 level SIBLING
groups: 0 (cpu_power = 294) 1 (cpu_power = 294) 2 (cpu_power = 294) 3 (cpu_pow
domain 1: span 0-3 level MC
groups: 0-3 (cpu_power = 1176)
domain 2: span 0-23 level CPU
groups: 0-3 (cpu_power = 1176) 4-7 (cpu_power = 1176) 8-11 (cpu_power = 1176
This is what I would have really expected before, though Fedora "only"
shipped 2.6.31 so I may have looked at something outdated.
Where did the NODE domain actually go, or are sched-domains limited to a
depth of 3?
thanks
Jan
--
To unsubscribe from this list: send the line "unsubscribe sparclinux" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
