I'm already aware of these questions:
- how to know if RAM is my bottleneck
- How to determine if CPU or RAM or DISK is the bottleneck on Windows 7
- How can I know if the speed of RAM is the bottleneck rather than CPU speed?
To summarize the answers: They focus on total RAM usage (amount of bytes) and paging, but not on the actual CAS latency and/or speed of the RAM.
I'm not asking about paging, nor about speed, nor about if 99% RAM usage is a bottleneck.
My question is specifically targeted at the latency of the RAM. Most RAM sold today is between 8 to 20ns true/total latency. Performance kits can go as low as 6ns. Worse kits go up to 55ns.
So my question:
In either Windows or Linux based systems, how would I, objectively, measure how much the performance of an application is impacted by the RAM latency?
- Which (and how?) measurements should I perform?
- How would it show up in measurements? i.e. how do I interpret the results?
