Potential Limit Of A Memory Kit ?

[Avenger411]

Hello!

I am in the process of finding the limit of my kit AND internal memory controller. In previous posts, i found out that running tighter timings and less mhz was more beneficial. Hence i continued my tests and now i am somewhere weird. I have this kit :

F4-3000C15Q-16GRR (from g.skills)

The kit is stated 3000mhz@cas15-15-15-35-2T. Right now, i am at 2750mhz@cas13-13-13-33-1T@1.35v. I am baffled at how the timings can be trimmed down. In the following days i'll try 12-12-12-32-1T@1.35v and i'll keep working my way down but i guess i'll hit a wall soon. My question, is it normal for such a kit to end up being much better than what it is actually stated for ? From my findings, from CAS15 to CAS13 there's a HUGE difference. I'm just wondering.

Max

[kaese]

I know the difference in CAS is often more important than the just the clock speed. But actually I dont see what makes you say the difference between stated and actual performance would be huge.
As delays are given in clock cycles and you go down from 3000 to 2750 one would expect: 15*2750/3000=13.75, and you achieve 13 - that is 5%. Anyways I guess you can be happy to be able run 1T command rate what is more of a difference.

[Avenger411]

Hey,

The difference is in synthetic benchmarks. Not in day to day usage.

Max

[Menthol]

Avenger,
Very nice, there are always some kits that have very good sample of chips on the modules, my first kit was a Gskil 3000mhz 15-15-15-, it ran very well at 3200mhz 14-15-15 at 100 blck, I believe it took at little more voltage but very good performance. If you search you may find a thread of Arne's, he spent some time working on timings with this kit
Like you say it's not in day to day performance, you learn a lot about your system playing with settings like this, knowledge gained is always a plus

[Avenger411]

Hey Menthol,

Indeed.

Max