Newb OCing E4300, DS3 Mobo

[Wirespeed]

Hi I'm a noob that has done lots of research and finally built my comp. Just want some advice, criticism, bashing, whatevers.

My goal is a stable computer that would last at least 5 years. I do not want to raise the vcore too high. I have tested stability of different vcores at different core speeds. Anything above 3050mhz, I would have to set my vcore to over 1.4v. I found a sweet spot of running 3041mhz at 1.3375v.

Right now I am running my comp with a 9x mult with FSB at 338. Ram Multiplier is set to 2 for a FSBRAM of 1:1. Ram is underclocked to 676mhz.

I was wondering if it would be a good idea to set multiplier to 8x and run FSB at 380 in order for the ram to run at 760MHZ. This way I would still have a 1:1 ratio and run my ram higher.

Also can somebody explain to me why a 1:1 ratio is good. I have done lots of research before buying and when actually overclocking. I read it is good to have a 1:1 ratio but never found out why.

In CoreTemp .94, my temps are 23-25C (idle), 53-54C (orthos small fft load)
Ambient Temp (70-75F)

Parts in my Sig

 

[bigblack]

My goal is a stable computer that would last at least 5 years. I do not want to raise the vcore too high. I have tested stability of different vcores at different core speeds. Anything above 3050mhz, I would have to set my vcore to over 1.4v. I found a sweet spot of running 3041mhz at 1.3375v.

Very wise choice not to raise the voltage more than 1.4. At 3.0Ghz you have a stable system with only 1.3375V which is very good and should last you 5 years. I seriously doubt that you will want to keep it for 5 years though

Right now I am running my comp with a 9x mult with FSB at 338. Ram Multiplier is set to 2 for a FSBRAM of 1:1. Ram is underclocked to 676mhz.

I was wondering if it would be a good idea to set multiplier to 8x and run FSB at 380 in order for the ram to run at 760MHZ. This way I would still have a 1:1 ratio and run my ram higher.

I think that if you run some benchmarks you will notice a small difference (better results) running at a lower multi and higher frequency. In real world applications this difference vanishes or is rather insignificant.

Also can somebody explain to me why a 1:1 ratio is good. I have done lots of research before buying and when actually overclocking. I read it is good to have a 1:1 ratio but never found out why.

You want a 1:1 ratio of FSB and ram in order not to introduce latency issues. A valid analogy goes like this: If you have a slow cargo ship and a faster cruiser ship that should escort the other one, you would have to make the cruiser ship go at the cargo ship's speed. Otherwise it would either not escort the cargo ship but race it, or it would have to go back and forth, spending fuel inefficiently. In this analogy the slow cargo ship is the FSB and the cruiser is your RAM. Latency is to go back and forth and it means introducing wait states in the communication between memory and FSB. I hope it is simple and clear.

In CoreTemp .94, my temps are 23-25C (idle), 53-54C (orthos small fft load)
Ambient Temp (70-75F)

This is a wide difference of 30C, double the normal difference! You would want to test your temps with TAT as well (TAT idle and TAT 100% load). What kind of fan do you use with the Thermalright and at what speed?

It seems you have selected a careful approach to overclocking which is very good and you try to learn more which is even better. Cheers!

 

[Wirespeed]

Thx for the analogy. I understand it now.

Update: now I'm running FSB at 336x9=3024mhz with vcore at 1.325 (stock) and ram set at 672mhz. Ran Orthos stable for 4+ hrs.

I'm using an Arctic Cooling AF9225 92x25mm fan
Rated Fan Speed: 2000 RPM
Air Flow: 34.0 CFM / 57.7 m3/h
However, I don't know how to check the speed its running at

TAT idle is 39C-43C,
Under 100% load on TAT (10min), it reports 68C-71C and 55C-59C in Coretemp.94

When I'm gaming (World of Warcraft), I get 37C-40C (most of the time on 37C) and that is the highest temp I ever see besides testing it with Orthos or TAT.

Are my temps too high?
I tried the stock hsf and it rose a few degrees.
For the Ultra-90, I reseated it a few times, tried different amounts of paste (lots, less, a thin line in the middle). I even lapped it and replace the stock bracket with nuts and bolts to have it more secure. And this is the coolest I can get it to.

 

[bigblack]

I would go with TAT measurements. They are more accurate than other utilities. Your load temperature is too high, especially compared with idle temperatures. Either your fan does not provide enough air flow, or something is wrong with your Ultra 90. It should give you significantly better temps than the stock cooler (10C less @load). You would want to keep you max load temps~60C in order to prolong the life of your CPU.

I would try a higher cfm fan first (like 60 cfm?). It is the cheapest of all possible tests. Something like this maybe will do the trick without breaking the bank.
http://www.newegg.com/Product/Product.aspx?Item=N82E16811999617
There is also the choice of thermal paste and its application. From various reviews i gather that the Shin Etsu thermal paste gives better results by 5C from what AS5 can do. But in your case i think air movement is the problem, since you have tried all the other tricks. Also try to disable all smart fan options from bios, so the fan moves always at full speed.

I don't know what else to suggest, except to sell the Ultra 90 and go for a Ultra 120 if everything else fails. Test and post results. Maybe you will get it right with just a fan...

 

[Wirespeed]

I would go with TAT measurements. They are more accurate than other utilities. Your load temperature is too high, especially compared with idle temperatures. Either your fan does not provide enough air flow, or something is wrong with your Ultra 90. It should give you significantly better temps than the stock cooler (10C less @load). You would want to keep you max load temps~60C in order to prolong the life of your CPU.

I would try a higher cfm fan first (like 60 cfm?). It is the cheapest of all possible tests. Something like this maybe will do the trick without breaking the bank.
http://www.newegg.com/Product/Product.aspx?Item=N82E16811999617
There is also the choice of thermal paste and its application. From various reviews i gather that the Shin Etsu thermal paste gives better results by 5C from what AS5 can do. But in your case i think air movement is the problem, since you have tried all the other tricks. Also try to disable all smart fan options from bios, so the fan moves always at full speed.

I don't know what else to suggest, except to sell the Ultra 90 and go for a Ultra 120 if everything else fails. Test and post results. Maybe you will get it right with just a fan...

buying materials, will report in a few days