OK. My i7-3770K is now running and being benched with stock GTX 690 at 4.8GHz with 8GB system RAM running at 1866MHz (versus running identical settings with 4GB, just completed).

3DMark11 scored 400 points higher with 8GB


. . . then i realized that i ran it with 4GB with the power target at zero and it is always +135%. i had to restart to add +.01V and forgot to reset it. With this kind of testing, it is pretty hard to not catch user errors in testing

Weird. i got system instability after upgrading from 4GB to 8GB of RAM. i had to add +.02V to both the DIMM and the CPU voltage to stabilize it. It took me a few hours to troubleshoot it but now it is solid again.

And i am finding some surprising results

Not that weird really, you are pushing the memory controller hard and putting extra load on it.

i guess that is why most overclocking is done with minimal RAM. However, i was surprised that simply adding two DIMMs affected system stability so much. i wonder if this will vary from MB to MB. i will be interested to see if the EVGA MB needs as much voltage as the ECS MB (and i wonder if i can OC the CPU even higher).

i also need to repeat results with 4GB of RAM to be absolutely sure of my results. Very interesting

The EVGA should be a slight improvement over the ECS board, but probably not by a lot I'd guess without knowing details of how both companies engineered the memory slots, if they provide extra prower separately or not. Even then, it is the CPU that will have the most influence, it contains the memory controller. It's a shame you don't also have an i7 SB to compare the i7 IB against.

i really think SB is near identical clock per clock as IB. I see people post stuff like IB is ~5% faster ipc but I dont think its true at all. Most of the original reviews of the 3700k showed it barely edging out the 2600k but its clocked slightly higher and that by itself accounts for the tiny bit of performance gain. The real improvements are igp and the energy savings which goes away when overclocking.

IB is great for intel. Its really is more efficient at stock clocks over SB and it brings a large leap in graphics. Its a nice step in the right direction and it has accomplished Intel's goals. But for the enthusiast who buys the k-series for long term overclocking, IB is just not any better than SB.

results?


Virtual Memory...................................

nevermind, lets wait to get into this. Like after the article. No spoilers!!!!!!!!

About 4GB vs 8GB of RAM. But then i am re-running some benches to be sure before i say anything
(like that guy at OBR )

nope. Now i almost feel like the owner of OBR ... "BIG ANNOUNCEMENT!!!!!!!!!!!"

(then later)

"nothing to see, oops"


Nope. nothing surprising. Interesting, yes.


i was originally referring to my 4GB vs 8GB system RAM comparison; the issue was that i accidentally set my Power Target on the GTX 690 back to zero (i test AMD cards with PowerTune maxed and test Nvidia cards with the power target set to Max - everything else on the cards is kept stock)

i wonder if there would be a difference if i used only 2 DIMMs - 2x4GB RAM. i need to ask Kingston about this.

ANYWAY ... you mentioned SB vs IB - Don Karnage did some great tests
http://forums.anandtech.com/showpost.ph ... stcount=44