How does cinebench isolate the CPU?

[terminalinfinity]

Bit of a curiosity question, but I just recently adopted cinebench into my benchmark suite after staying with a set of tools Im more familiar with for years. 

 

Im really enjoying it, but Im currently OCing my memory and now Im curious: how exactly does cinebench isolate the CPU vs other benchmarks?  The reason I ask is because I had the same experience that Anthony had on the LTT video about ryzen memory clocks: cinebench just doesnt care about memory speed/latency in Ryzen, unlike every other CPU benchmark Ive tried.  You can punch in any ram speed, even 2133 at CL18 and your cinebench score will not change.  (I tried)

 

Whereas with every other benchmark Ive tried that claims to have a benchmark that isolates the CPU, you can clearly see gains in tighter timings and/or higher frequencies on Ryzen.

 

So what makes cinebench so special/different?  Just really curious.

[Enderman]

Well if you write a program that doesn't need to use ram and all the data can be stored in the CPU cache then ram speed won't make any difference...

[Fasauceome]

Ryzen is impacted by memory speed due to the fact that the core to core latency is reduced by more bandwidth and less latency. If cinebench isolates all the cores to complete the workload, the core to core latency isn't a factor.

[terminalinfinity]

Thanks.  That was extremely well put.  It also helped me understand the other end: benchmarks that just send as many calculations as possible and leave it to the processor to handle them as it wants would inversely be affected by core to core latency, correct?