[PATCH v2 00/11] sched: consolidation of cpu_power

[preeti@linux.vnet.ibm.com (Preeti U Murthy)]

On 05/26/2014 09:24 PM, Vincent Guittot wrote:
> Hi Preeti,
> 
> I have done ebizzy tests on my platforms but doesn't have similar
> results than you (my results below). It seems to be linked to SMT. I'm
> going to look at that part more deeply and try to find a more suitable
> HW for tests.

You are right Vincent. I tested this in smt-off mode and the regression
was not seen. But the regression was of the order 27% with higher number
of threads in smt-on mode. What is interesting is that the regression
increases in the range N=1 to N=24 and then it dips to 0 at N=48 on a 6
core, SMT 8 machine. Let me dig this further.

Let me dig further.

Regards
Preeti U Murthy
> 
> ebizzy -t N -S 20
> Quad cores
>  N  tip                 +patchset
>  1  100.00% (+/- 0.30%)  97.00% (+/- 0.42%)
>  2  100.00% (+/- 0.80%) 100.48% (+/- 0.88%)
>  4  100.00% (+/- 1.18%)  99.32% (+/- 1.05%)
>  6  100.00% (+/- 8.54%)  98.84% (+/- 1.39%)
>  8  100.00% (+/- 0.45%)  98.89% (+/- 0.91%)
> 10  100.00% (+/- 0.32%)  99.25% (+/- 0.31%)
> 12  100.00% (+/- 0.15%)  99.20% (+/- 0.86%)
> 14  100.00% (+/- 0.58%)  99.44% (+/- 0.55%)
> 
> Dual cores
>  N  tip                 +patchset
>  1  100.00% (+/- 1.70%)  99.35% (+/- 2.82%)
>  2  100.00% (+/- 2.75%) 100.48% (+/- 1.51%)
>  4  100.00% (+/- 2.37%) 102.63% (+/- 2.35%)
>  6  100.00% (+/- 3.11%)  97.65% (+/- 1.02%)
>  8  100.00% (+/- 0.26%) 103.68% (+/- 5.90%)
> 10  100.00% (+/- 0.30%) 106.71% (+/- 10.85%)
> 12  100.00% (+/- 1.18%)  98.95% (+/- 0.75%)
> 14  100.00% (+/- 1.82%) 102.89% (+/- 2.32%)
> 
> Regards,
> Vincent
> 
> On 26 May 2014 12:04, Vincent Guittot <vincent.guittot@linaro.org> wrote:
>> On 26 May 2014 11:44, Preeti U Murthy <preeti@linux.vnet.ibm.com> wrote:
>>> Hi Vincent,
>>>
>>> I conducted test runs of ebizzy on a Power8 box which had 48 cpus.
>>> 6 cores with SMT-8 to be precise. Its a single socket box. The results
>>> are as below.
>>>
>>> On 05/23/2014 09:22 PM, Vincent Guittot wrote:
>>>> Part of this patchset was previously part of the larger tasks packing patchset
>>>> [1]. I have splitted the latter in 3 different patchsets (at least) to make the
>>>> thing easier.
>>>> -configuration of sched_domain topology [2]
>>>> -update and consolidation of cpu_power (this patchset)
>>>> -tasks packing algorithm
>>>>
>>>> SMT system is no more the only system that can have a CPUs with an original
>>>> capacity that is different from the default value. We need to extend the use of
>>>> cpu_power_orig to all kind of platform so the scheduler will have both the
>>>> maximum capacity (cpu_power_orig/power_orig) and the current capacity
>>>> (cpu_power/power) of CPUs and sched_groups. A new function arch_scale_cpu_power
>>>> has been created and replace arch_scale_smt_power, which is SMT specifc in the
>>>> computation of the capapcity of a CPU.
>>>>
>>>> During load balance, the scheduler evaluates the number of tasks that a group
>>>> of CPUs can handle. The current method assumes that tasks have a fix load of
>>>> SCHED_LOAD_SCALE and CPUs have a default capacity of SCHED_POWER_SCALE.
>>>> This assumption generates wrong decision by creating ghost cores and by
>>>> removing real ones when the original capacity of CPUs is different from the
>>>> default SCHED_POWER_SCALE.
>>>>
>>>> Now that we have the original capacity of a CPUS and its activity/utilization,
>>>> we can evaluate more accuratly the capacity of a group of CPUs.
>>>>
>>>> This patchset mainly replaces the old capacity method by a new one and has kept
>>>> the policy almost unchanged whereas we can certainly take advantage of this new
>>>> statistic in several other places of the load balance.
>>>>
>>>> TODO:
>>>>  - align variable's and field's name with the renaming [3]
>>>>
>>>> Tests results:
>>>> I have put below results of 2 tests:
>>>> - hackbench -l 500 -s 4096
>>>> - scp of 100MB file on the platform
>>>>
>>>> on a dual cortex-A7
>>>>                   hackbench        scp
>>>> tip/master        25.75s(+/-0.25)  5.16MB/s(+/-1.49)
>>>> + patches 1,2     25.89s(+/-0.31)  5.18MB/s(+/-1.45)
>>>> + patches 3-10    25.68s(+/-0.22)  7.00MB/s(+/-1.88)
>>>> + irq accounting  25.80s(+/-0.25)  8.06MB/s(+/-0.05)
>>>>
>>>> on a quad cortex-A15
>>>>                   hackbench        scp
>>>> tip/master        15.69s(+/-0.16)  9.70MB/s(+/-0.04)
>>>> + patches 1,2     15.53s(+/-0.13)  9.72MB/s(+/-0.05)
>>>> + patches 3-10    15.56s(+/-0.22)  9.88MB/s(+/-0.05)
>>>> + irq accounting  15.99s(+/-0.08) 10.37MB/s(+/-0.03)
>>>>
>>>> The improvement of scp bandwidth happens when tasks and irq are using
>>>> different CPU which is a bit random without irq accounting config
>>>
>>> N -> Number of threads of ebizzy
>>>
>>> Each 'N' run was for 30 seconds with multiple iterations and averaging them.
>>>
>>> N          %change in number of records
>>>            read after patching
>>> ------------------------------------------
>>> 1          + 0.0038
>>> 4          -17.6429
>>> 8          -26.3989
>>> 12         -29.5070
>>> 16         -38.4842
>>> 20         -44.5747
>>> 24         -51.9792
>>> 28         -34.1863
>>> 32         -38.4029
>>> 38         -22.2490
>>> 42          -7.4843
>>> 47         -0.69676
>>>
>>> Let me profile it and check where the cause of this degradation is.
>>
>> Hi Preeti,
>>
>> Thanks for the test and the help to find the root cause of the
>> degration. I'm going to run the test on my platforms too and see if i
>> have similar results with my platforms
>>
>> Regards
>> Vincent
>>>
>>>
>>> Regards
>>> Preeti U Murthy
>>>
>