So I am searching about this latency issue. When I mean latency, I mean the total time from when the app needs the data, cpu requests it, goes over the memory bus to the dimm, and returns with the data. The CAS latency you brought up before is only the time it takes for a single module on dimm to return data.
However, the timings that you find on memory is only for the single modules themselves, not for the entire dimm. These modules are controlled by the memory controller. If the memory controller needs to handle more of these modules (many 2GB and 4GB memory dimms are double-sided), then the locate to read/write ops should take longer.
However, even though memory latencies may increase slightly (I retract my statement of double), it seems that somehow windows xp is still able to increase performance with more memory. http://www.pcauthority.com.au/Featur...ifference.aspx
I think what we need someone to do is use windows xp and use everest and record 1GB and 2GB memory latency times.