Best potential configuration

Associate
Joined
11 Jun 2006
Posts
1,569
E7200 @ 3.4ghz (Hits 3.6 stable max) 1.44/5volt current .138
Geil black dragon 2gb 800mhz 4-4-4-12 1.95
MSI p45 neo
Cpu is at 55c under load

400x8.5
1:1 ratio
800mhz ram

What is the best rato and combination i can get bearing in mind i am told my ram can reach around 1000mhz at 5-5-5-X timings
 
Hey Lighters!

If you have a good ideal of how your CPU and Chipset overclocks then it sounds like you are ready to get testing your RAM.

To get some free extra performance you would want to get those sticks overclocked to their very max. There is quite an easy way to test this by lowering your CPU and FSB overclock then have a play around with the Northbridge straps. I'm not familiar with your motherboard but hopefully there is an option to change the speed/ratio of the memory dividers. . .

Have a flick in the BIOS (or read the manual) and see if you can change the memory divider so that the RAM is running async and on an upward divider i.e the ram is running faster than the dialed in FSB rather than the same.

Then all you need to do is start upping the FSB and see how far those sticks can go, your tweaking tools will be a combination of vDimm, Memory timings and Northbridge/vMCH volts.

Download and burn a copy of the new Memtest86+ (v2.11) to CD-ROM and boot from that to really give the sticks a good diagnostic, any errors will make themselves known quickly.

If you reach a nice memory speed and it passes several hours of Memtest86 then you can take it to Windows testing using both IBT (Intel Burn Test) or Prime Large in-place FFTs, if either of these crashes or finds errors then your most likely looking at increasing the Northbridge volts to get things running nice :)

Post back if you need any more help! :cool:
 
will i likely get enough performance to warrant doing this? If you mean about the divider by 1:1 1:2 etc and 1:1.5 etc thats not a problem and changing the vdimm voltage is very easy but i dont know what the NB is specifically as its been a while
 
You got the info you need to get testing and draw your own conclusions, don't worry to much about the Northbridge volts yet, just get the ram overclocked and running fast and use Memtest86 to test.

You wanna know how fast the sticks can run with 4-4-4-12 timings and what vDimm, same for 5-5-5-18 timings. You will probably get the highest MHz using higher vDimm and slacker timings.
 
wot likely benefits will i see then? how can i draw conclusions? based on what exactly? Ie cpu and gpu overclocks = more fps
Edit 5min memtest on each
4-4-4-12
Default settings
2.1 voltage 851 stable 1:1.60
2.1 voltage 886 unstable 1:1.67 (Tried 2.2 with 1.148 mch)
2.1 voltage 880 stable 1:1.60
2.1 voltage 886 stable 1:1.60
2.1 voltage 889 unstable 1:1.60

Why did my result show unstable with a ratio of 1.167 and stable with 1.60 even though the mhz was the same?
Why did it not go stable after 4 +1volts and 4 clicks on the mch?
Its geil ram but will corsair dominator airflow fan mount help so i can increase voltage?

2.1 voltage 880 stable 1:1.60 Intel burn test stable
 
Last edited:
Hey Lighters!

Sorry If I was a little brief in my post before but I guess I didn't read your first post correctly . . .

right so your using an Intel® Core™2 Duo E7200 (nice!)

Multi: 9.5
FSB: 266MHz
System Bus: 1066MHz (FSB x4)
CPU Speed= Multi x FSB = 2527MHz

So right out the box that processor needs DDR2-533 (aka PC2-4200, 266MHz) to run at stock speeds above (9.5x266MHz) using synchronous memory ratio 1:1 (memory running identical to real speed of FSB, FSB=266, Memory real=266MHz but DDRDoubled to an effective 533MHz)

PC2-4200
Real MHz: 266
DDR2/Effective MHz: 533

So an Intel® Core™2 Duo E7200 in an Intel® P45 Express Chipset with DDR2 memory running at 1:1 has the following *main* frequencies (there are others such as PCI, PCI-E etc but I'm omitting them from this example), all frequencies are directly linked apart from one . . . . DDR2 Speed (in Yellow)

2527MHz
1066MHz
533MHz
266MHz

If you change any of the other frequencies it will effect them all . . .

Only the memory divider (aka DDR2-Speeds, Northbridge Straps, Memory Dividers, Memory Ratios speeds etc etc) allows you to directly change the speed of the memory without any increase to the other frequencies . . . and that's your key to becoming a memory overclocking gEEk! :cool:

You have already done a lot of testing which is good, you know roughlyhow many MHz your CPU can produce, you also know roughly how many MHz your Chipset (aka Northbridge, MCH, FSB) can run . . . so what you need to do next is kinda start your testing from scratch but the second time around you are changing your memory from 1:1 *sync* to a faster speed using the dividers, as you start increasing the FSB the memory MHz will go flying off and you got Memtest86 at hand to do your testing.

Now does all that information answer your question? probably not but a bit of background doesn't hurt for anyone else lurking/reading lol! :D

"Best potential configuration"

It is my belief you will get better performance from your Intel® Core™2 Duo E7200 if you can get that memory running at an effective 1066MHz (aka DDR2-8500, 533MHz) :eek:

If you did that you would be an overclocking god! but I don't think you will be able to safely/stably :(

You just need to give those sticks a good test and make some notes, once you realise its speeds you can make the final adjustments, a balance of CPU MHz, FSB MHz and Memory MHz/Timings and volts

You want them all (CPU/FSB/MEM) to be as high as possible for the best overall performance boost!

The real world difference would be worth it, your computer would be faster, games would be faster, encoding would be faster, everything would be faster!
 
Cool, thanks for the informative reply!
Im currently testing 5-5-5-15 in the conclusion 880 is the max i can achieve with 4-4-4-12.
I am running 5 minute tests in memtest error free and then orthos crashes within 2 minutes. I made a stain in my underwear as i jumped from 933-953 and my computer/speakers made this loud cracking sound :( I restarted and tested previous orthos and all is stable.

I cant see myself reaching 1066mhz on 5-5-5-15 latencies
 
hehe cool! :D

Your off, just relax and enjoy, don't worry, just makes sure your sticks don't start getting burning hot for extended periods (touch them carefully with your fingers, careful!).

Using memtest I normally do a couple of full passes to test, I also do 1-50 custom loops of Test #5

C then 1 then 3 then 5 then enter then 0 then Space Bar

If the stick pass Memtest86 but fail in Prime Small in-place FFTs then the problem is the Northbridge pOOping out!

In that case you need to carefully start giving the Northbridge some extra juice, again just make sure the Northbridge heatsink doesn't get burning hot for extended periods (touch it carefully with your fingers, careful!).

If you find that the Memory or Northbridge heatsink gets too hot for you to keep touching them for 5 seconds without a sizzling skin sound and feeling then you need to get some active cooling sorted out! :p

I've got a baby overclocking board with built in graphics and have just lobbed a 120mm fan blowing directly down onto the Northbridge, Memory Sticks and Southbridge, they never get hot ever!

Active NB/RAM/SB Ghetto Cooling, 120mm Sharkoon SE1000
eco2017eum6.jpg

ECO-II Prototype (Intel® Core™2 Duo) - 52w Idle - 78w Load - 87w Max Load
 
thats awesome, thanks for info. I will definate update more tomorrow as g/f is begging me to come to bed lol :) Thanks a bunch will the corsair dominator airflow help me?
 
I've not used a corsair dominator airflow thingie but I'm sure it will work fine keeping the memory sticks cool (maybe a bit noisy?), not really helping keep the NB cool! (that's if it needs it).
 
http://www.ixbt.com/mainboard/memdb/corsair_ddr2_dominator/dominator_airflow.jpg
880 4-4-4-12
949 5-5-5-15
3519 @ 1.44 voltage intel stable
3411 @ 1.38 voltage orthos stable (havent tested intel yet)
These are on 2.1 and standard NB volts with ratio of 1:1.60 and 1:1.67 stable in orthos and intel burn test
1 problem i did have was when i played music when testing and my pc restarted itself showing signs of instability but then it did a full intel burn test fine just after

Assuming 3519 is my best stable clock which settings will give me optimum results
3511 477 RAM intel burn test i ran 30 tests of max memory and got 1 error http://e.imagehost.org/view/0339/intelburntest
 
Last edited:
come on someone, whats better 40404012 at 880 or 5-5-5-15 at 949? and does the 1:1 1:1.20 ratio make a difference if the end result is 880?
 
Back
Top Bottom