on the motherboard . On a substandard platform , testing is made easy , however it does get complicated with any other board that is competent at memory overclocking like the APEX . As it is capable of achieving what is otherwise not possible on many other Z270 offerings .
Looking at the results , one should be mindful of how nonlinear performance scaling is due to the automatic configuration of memory settings outside of your primary timings and perhaps the command rate . The motherboard will set the memory and IMC to particular values which may result in you setting lower primary timings and a higher frequency , but ending up with worse results . In light of this , the testing was done backwards which is to say from the highest frequency down to the lowest . All secondary and tertiary timings were set manually and identical across all frequencies with only the primary timings and
“ There ’ s not much that is left to say about this memory . It ’ s beautiful , is in limited quantities and overclocks exceptionally well .” voltages changing as shown in the graphs . In this way , it is a real representation of scaling with reduced variation outside of what it is that is being tested .
No testing was done on X99 platforms but , other than the low memory frequency ceiling , the timings should be possible on both Haswell-E and Broadwell-E systems .
What is evident in the results is that the two best settings for the memory are 4000MHz at 1.45V or perhaps 3,866MHz . This is important given just how tough 4000MHz can be on many 4-DIMM boards , but
20 The OverClocker Issue 40 | 2017