Question about frenquencies displayed in benchmarks

[frwho]

Hi !

I've exhausted all Aida64 benchmarks (memory + cpu).

My question is :  when testing CPU Queen perfs for example, the cpu frenquency displayed is 3900 Mhz with overclock due to turbo-boost (i7-3770 not K, base is 3.4 Ghz). All other cpus display results and base frequencies associated, let's say 3500 Mhz for i7-3770K (base 3.5 Ghz, T.B. 2.0 is 3.9 Ghz).

Does this means that all other cpu compared are executed with turbo-boost disabled ? or only real frequency is displayed for the cpu which is tested ? Or, in another words, do I have to disable turbo-boost to benchmark my cpu ?

Thank's for clarify those infos...

[Fiery]

In the reference benchmarks list we indicate the base clock for every processors, without taking CPB or Turbo Boost into consideration.  Every reference processors were running with stock (default) configuration, so if they support CPB/Turbo, we leave it enabled to maximize their performance.

Regards,

Fiery

[frwho]

Thank you for your answer

[MAA]

Fiery

Did I understand correctly that in the reference results you publish the base frequency, but the benchmark itself is performed at the turbo frequency?

In my opinion this is an incorrect approach, which is misleading.

Some processors have a very high turbo frequency (up to +200% to Base), but it works very short time:

 

In my opinion, you should publish the Turbo frequency, or increase the benchmarks time to several minutes.

[Fiery]

On 12/26/2017 at 12:47 PM, MAA said:

Fiery

Did I understand correctly that in the reference results you publish the base frequency, but the benchmark itself is performed at the turbo frequency?

In my opinion this is an incorrect approach, which is misleading.

Some processors have a very high turbo frequency (up to +200% to Base), but it works very short time:

In my opinion, you should publish the Turbo frequency, or increase the benchmarks time to several minutes.

Since only the base frequency is guaranteed, indicating the turbo frequency would be less relevant IMHO.  Not to mention the fact that modern processors have different turbo frequencies defined for different workloads (e.g. AVX-512 turbo is different to SSE turbo, etc).  Only the base frequency is somewhat constant as a single MHz value.

Our benchmarks do not enable or disable Turbo Boost, so the benchmarks are performed at whatever frequency the CPU chooses to operate on.  Most of our benchmarks complete in 10 seconds, but we're planning to implement some sort of long-haul benchmarks in the future in order to better measure the long-term performance you can expect from such low-TDP parts that turbo like crazy in the short run   Exactly as the graph from the TechSpot article shows.

[MAA]

On 1/3/2018 at 2:41 PM, Fiery said:

but we're planning to implement some sort of long-haul benchmarks in the future

maybe just add long running versions of existing benchmarks?