Re: [Patch v6 0/7] Introduce Thermal Pressure

[Thara Gopinath]

On 12/16/2019 09:56 AM, Peter Zijlstra wrote:
> On Wed, Dec 11, 2019 at 11:11:41PM -0500, Thara Gopinath wrote:
>> Test Results
>>
>> Hackbench: 1 group , 30000 loops, 10 runs
>>                                                Result         SD
>>                                                (Secs)     (% of mean)
>>  No Thermal Pressure                            14.03       2.69%
>>  Thermal Pressure PELT Algo. Decay : 32 ms      13.29       0.56%
>>  Thermal Pressure PELT Algo. Decay : 64 ms      12.57       1.56%
>>  Thermal Pressure PELT Algo. Decay : 128 ms     12.71       1.04%
>>  Thermal Pressure PELT Algo. Decay : 256 ms     12.29       1.42%
>>  Thermal Pressure PELT Algo. Decay : 512 ms     12.42       1.15%
>>
>> Dhrystone Run Time  : 20 threads, 3000 MLOOPS
>>                                                  Result      SD
>>                                                  (Secs)    (% of mean)
>>  No Thermal Pressure                              9.452      4.49%
>>  Thermal Pressure PELT Algo. Decay : 32 ms        8.793      5.30%
>>  Thermal Pressure PELT Algo. Decay : 64 ms        8.981      5.29%
>>  Thermal Pressure PELT Algo. Decay : 128 ms       8.647      6.62%
>>  Thermal Pressure PELT Algo. Decay : 256 ms       8.774      6.45%
>>  Thermal Pressure PELT Algo. Decay : 512 ms       8.603      5.41%
> 
> What is the conclusion, if any from these results? Clearly thermal
> pressuse seems to help, but what window? ISTR we default to 32ms, which
> is a wash for drystone, but sub-optimal for hackbench.
Hi Peter,

Thanks for the reviews. IMHO, the conclusion is that thermal pressure is
beneficial but the decay period to be used depends on the architecture
and/or use-cases. Sticking to 32ms should give some improvement but it
can be tuned depending on the system.

> 
> 
> Anyway, the patches look more or less acceptible, just a bunch of nits,
> the biggest being the fact that even if an architecture does not support
> this there is still the code and runtime overhead.

I am fixing this and sending out a v7.

> 


-- 
Warm Regards
Thara