post #1 of 1
Thread Starter 
Hi All!

My first post on this forum but I am not new to the OC game, definitely well read on the subject and have seen millions of threads. I’ll walk you through my process and all my benchmark data will be at the bottom of the thread. I have some questions around the voltages and timings and I want to get some opinions on how to perfect this.

I want to give you my “quick” (sorry) back story so you know this was thorough. A couple weeks ago my PC crashed with a memory problem which was the BSoD I had in years. When I restarted and I noticed I was using over 90% of my 6GB with various games, skype, chrome with a bunch of tabs, and a couple other things being open. So I was thinking about buying some more RAM, then I considered maybe it’s time for an entirely new machine because there is so much cool new stuff out there and I just built one for friend. I settled on buying a new SSD and buying more RAM (another 2x3GB kit of the same part number) since my machine really isn’t that old and I think I can just get a little more out of it. During this project I figured I would benchmark everything before and after each change, then I would optimize everything and re-overclock etc.

I started by installing the SSD, reinstalling windows on it, and benchmarking. The SSD had a huge increase on my disk mark in Passmark as I expected, but I also saw increase in my Graphics and Memory which was cool. (See table below - I ran most tests twice for each of the 3 configurations)



Then I increased my RAM from 3x2GB to 6x2GB. First I checked my BIOs and saw all 12gigs, then I loaded up windows and checked CPU-Z again saw 12gigs. But then I went to the task manager and saw the max memory at 6 gigs so I went to the resource monitor and saw 6 gigs allocated as “Hardware Reserved”. I immediately ran to the forums thinking Windows was being annoying and a setting holding back some of my RAM. So I checked a few Windows settings, ran Memtest86, swapped my RAM around in different slots, etc, etc, etc – all of which did nothing and I still had 12 gigs of ram, 6gigs usable according to Windows. So then I started playing in the BIOS again.

My RAM is rated at 1.65v & 1600 Mhz and that is how I configured it in my BIOs. But after a million different variations of voltages and changes to other settings I couldn’t get it to work. So I gave up on the 1600 Mhz and went the STD voltage( 1.5v) and STD speeds(1333 Mhz). Then when I loaded up Windows I had all 12gigs usable. BUT when I ran my benchmarks, my scores dropped a bit from the previous tests. Yes there was a small change in the base clock that affected the CPU but I would think the 40 Mhz or whatever would be negligible. This is what leads to my question!

How do I use 12GB of RAM at 1600 Mhz? (and have Windows declare 12 gigs usable smile.gif) I feel like it has something to do with the DIMM voltage and the QPI voltage but I can’t figure it out. Primary goal is to maximize my benchmarks, hopefully the “how does it feel” will increase as well. But I feel like the benchmarks are a good somewhat scientific way to prove results.

I know my cooler can handle a 3.8 -4Ghz OC since My CPU in Prime95 overnight never breaks 65 C so I have a lot of headroom on the CPU cooling which will allow me increase my base clock a lot more. But I am still worried about hitting 1600 and needing to increase my Dimm voltage which sets my RAM back to 6 GB. If the only stable build with 12GB will be my current current settings then I can accept it but I want to get your ideas. I think the key is in the multipliers.

I found a guide in a different thread that I am trying to follow so I made some notes on my settings. Please advise!
(Great guide here http://www.evga.com/forumsarchive/tm.asp?m=716108)
1. Vcore vtt +/-.5 from Dimm Voltage (Check 1.22 vtt vs 1.5 Dimm)
2. Difference between QPI PLL from Dimm Voltage should be between .1 and .5 (Check 1.1 qpi vs 1.5 dimm)
3. I reduced my memory frequency to 2:8 to get my memory near 1333 with a 166 clock. If I leave it 2:10 it goes over 1600 and I feel like I would need to raise my DIMM voltage to 1.65 which puts me back at 6 gigs usable.
4. I read in a guide that 2:8 memory should have a reduced core clock ratio to 16x, down from 20x. But if I do that I need my base clock to be 200 to get my CPU @ 3.2 Ghz and 1600 Ram. That baseclock number seems a bit high.

Any advice, tips, ideas are welcome. Thanks in advance! I hope I can solve this problem and this thread can be of use for others.


My build.
Mother Board (EVGA X58 SLI LE)
CPU (i7 920)
CPU Cooler (Xigmatec Dark Knight)
Memory (Corsair XMS3 12GB (6x2gb) TR3X6G1600C9)
PSU (Corsair TX650)
GPU ( EVGA GTX 570 )
Operating System ( Windows 7 Pro 64-bit )

MyBios.
Frequency Control
CPU Clock Ratio (20X)
CPU Host Frequency (Mhz) ( 166) [Previously at 160 before changing my memory settings forcing me to up it]
MCH Strap (Auto)
CPU Uncore Frequency (Mhz) (AUTO)
CPU Clock Skew (0 ps)
Spread Spectrum (Disabled)
PCIE Frequency (Mhz) (100)

Memory Feature
Memory Speed ( Standard )
Memory Control Setting ( Enabled )
Memory Frequency ( XXX / 2:8 ) Target =1330 [recent change, I want it back at 1600 as stated above]
Channel Interleave Setting ( 6 Way )
Rank Interleave Setting ( 4 Way )
Memory Low Gap ( Auto )
tCL Setting ( 9 )
tRCD Setting (9 )
tRP Setting (9 )
tRAS Setting (24 )
tRFC Setting ( 88 )
Command Rate ( 1t ) Should I bump to 2T?

Voltage Control
EVGA VDroop Control ( With VDroop )
CPU VCore ( AUTO ~1.25)
CPU VTT Voltage ( +100mv )
CPU PLL VCore ( AUTO ~1.8 )
DIMM Voltage (1.5)[Also a recent change and would expect to go back new 1.65]
DIMM DQ Vref (AUTO ~+0mV )
QPI PLL VCore (1.1 )
IOH VCore ( Auto~1.1 )
IOH/ICH I/O Voltage ( Auto~1.5 )
ICH VCore ( Auto~1.05 )

CPU Feature
Intel SpeedStep (Disabled )
Turbo Mode Function ( Disabled )
CxE Function ( Disabled )
Execute Disable Bit ( Enable )
Virtualization Technology ( Disabled )
Intel HT Technology ( Enabled )
Active Processor Cores ( All )
QPI Control Settings ( Enabled )
QPI Link Fast Mode ( Enabled )
QPI Frequency Selection ( Auto )