but I can't seem to get a accurate memory reading. I've run multiple long duration blend an know my ram is stable and could probably go even higher. Mostly I see people using maxxmem on these forums... I've tried 2 versions and they give completely different results each time, secondly only one version was able to detect any of my hardware. Earlier I saw someone post opening chrome helped their score and I thought that was silly but I couldn't resist trying it and it gave me 5-10% higher scores with chrome open.
I took it a step further and tested it with my minecraft server up and populated, score went up more. So I tested it someone standing by a zombie spawner and it reduced my score drastically as did benching with litecoin mining 2xgpus. It would appear that programs reserving ram is raising my score while programs moving a lot obviously still lowers it. Or maybe I'm overanalyzing random recurring flukes.
Sisandra gives me the same 20.1gb/s @ 53ns each time and that seems too high where memaxx seems too low.
Spent a lot of time tweaking and testing my ram at this point - 8-9-9-24-7-5-7-7-14-300-33-1T with cpu/NB @ 2571
My old version of maxxmem = 10,000-13,000 copy, 9000-11,000 read, 4,000-6,500 write
maxxmem 1.99 = 13,500-16,000 copy, 12,000-14,000 copy, 8,000-8,400 write <- this version detects my timings except command rate.
Here's my best score with 12 chrome windows and 4 people on my minecraft server.
Leads me to think maxxmem is factoring my systems total ram vs how fast it can move the amount that's actually available to calculate score.
Or I could have a tinfoil hat on and hardware detection issue.
Update : Idea led to logical explanation for chrome = priority and core affinity. Score is lower and chrome trick doesn't work with maxxmem at high priority. This may have happend after I applied the win7 fx hotfix.
Guess that means sisandra is probably right.