if the only benefit is that I drop the temps from 95 to 80

Hi Charles.

 

If you haven't already done so I suggest you turn off multi-threading in your Bios. My temps reduced by  5 degrees and in FSX that won't cost you any performance hit. Other games maybe, but not FSX.

Some of the FSX overclockers are saying to just overclock two of the cores to the highest speed (say 4.6) and the rest at quite a bit lower speed (maybe 4.2 to 4.4) since we can use the Affinity_mask approach to run the main FSX loop on a particular core. This apparently reduces the overall temperature issues enough to maybe avoid delidding and H2O cooling. Anyone with practical experience on this idea?

Anthony Young:

 

Your chip is better than mine ... I did that same test (1.2V with 46x multiplier) and I was able to get to the point in Windows startup where the colored circles are twirling around but no further. 

 

My machine is fully prepped now and waiting on the arrival of the Arctic MX-4 tomorrow ... I have a trick up my sleeve to create a perfect application of TIM in a situation where the surfaces are as close to flat as possible (created by lapping the IHS). 

 

I just hope the thing boots up ... it's just my luck that after performing all of this work the thing is just going to sit there doing nothing!  Believe it or not this already happened to me.  I had just upgraded the rig by adding a 1TB HDD from my old rig.  I plugged everything in and absolutely nothing happened.  The motherboard appeared to be completely dead.  30 minutes of hand writhing later I'm staring at the mobo in stunned silence when I notice this little black button ... named ... DirectKey.  I pushed it after "I wonder what that key does" crossed my mind and boom ... the mobo came to life. 

 

The manual says:

 

This feature allows your system to go to the BIOS Setup program with the press of a button.  With DireckKey, you can enter the BIOS anytime without having to press the <Del> key during POST.  It also allows you to turn on or turn off your system and conveniently enter the BIOS during boot-up.

 

Whooda thunk it.  I was about ready to RMA the board.

 

Appliance:

 

Done ... that was part of NickN's setup guide for preparing to OC ... thanks for bringing that up.

 

WhitaV8:

 

That is the first I've heard of that ... the HERO motherboard allows for this setup (as I'm sure most other mobo's do too.)  I'm going to have to do some research about that Affinity Mask setting ... here is a good discussion about that, thanks for piquing my interest.

 

Thanks everyone ... until tomorrow ... listen to this ... in this context it's hilarious ... (I guess I'm getting giddy I'm so close now)

Do the good chips all come from a particular Lot number? Is there another way to tell if you are buying a top performing chip?

 

What I do to get a good chip is go to Chinese sites like amazon or ebay for china (they call it taobao), and they have tested chips to ensure you get certain OC capability with given vcore.  I'm sure I'll get critiques here and there but that's what I did to get my current IB.

 

And sure you pay for a bit more to ensure OCability, but at the end of the day I prefer doing that than come home with a i7 CPU of stock price but do not OC more than 4.2GHz without start cooking things ~~ Which was what happened to my first CPU.  The CPU was perfectly functional, except I do not feel comfortable pushing it given the performance.  I went to the Chinese sites to get a tested (still with warranty) i5, slightly cheaper than the price of i7, but with a mild/medium OC, I am running 4.8GHz stablish.

Another thought for the evening ... I always like to give things names ... and I didn't know what to call those "things" running down the left hand side of the thermal die ... some say they are "surface mounted capacitors" ... NickN says they are "SMD Resistors" ... but IBM calls them "voltage regulators".  I tend to trust Nick ... but the question remains ... are SMD Resistors also "voltage regulators"?

 

They apparently do not generate heat.  That being said some people are using Permatex Liquid Electrical Tape to cover them so that electrically conductive TIMs like Coollaboratory Liquid Pro won't short them out.  I really like this idea since, as I've said before, I like CLP under the hood but am always afraid that I'll let the IHS slip from right to left and cause the CLP to come in contact with those caps/resistors. 

 

Also, during my internet scouring this evening, I've run into several obscure posts of people who have "nicked" them, just like I think I did, and have them continue to function just fine (not recommended of course).  Evidently they are a pretty robust component.

 

This would be a perfect solution to my "manufacturing process" ... it would allow me to use CLP between the IHS and the thermal die with zero chance of making a mistake during mounting ... the idea is very intriguing as I like total peace of mind when working under that IHS.

 

C. T.

Have to watch this one!

Everything you EVER DIDN'T want to know about the i7 ... 4th Gen Core Family Desktop Vol. 1 Datasheet.

 

And, you just can't make this stuff up ... I was helped by a "tranny" ... bwhaahhh ... Intel really is an equal opportunity employer.

 

Also, those "dots" at the top left edge of the process have no purpose whatsoever so they do not deserve a name other than "those dots at the top left corner of the processor" or TDATTLCOTP for short ... phonetically ... "te" + "dattel" + "cot"s ... the P is silent.

 

So, I'm wondering ... are those voltage regulators the physical manifestation of the on-die FIVR?

 

Processor related voltage rails have been moved onto the die. The motherboard supplies 1.8V via the processor socket to Intel’s on-die FIVR (fully integrated voltage regulators), which then convert the voltage down to required levels for various parts of the CPU.

 

Final for tonight ... look at the power consumption of an i7-4770K processor overclocked to 4.5GHz ... wow!

 

 

C. T.

I just figured out something totally annoying ...

 

1.  Everyone states a good FSX system really doesn't require anything more than 8GB

2.  NickN states that DDR3 2400 9-11-11-31 is a smokin' choice

3.  NickN also states that two sticks are better than four for overclocking

4.  I wanted 16GB for future growth.

 

So, I went shopping and found a killer deal on G.SKILL Trident X Series 8GB (2 x 4GB) DDR3 2400 for $107.99 ... the EXACT specification recommended by NickN.

 

So, I then went looking for an equivalent specification for 2 x 8GB chips from any manufacturer ... can't be found. 

 

I suppose that doubling the available space on a single stick slows things down a bit.  So, the question that comes up is this ... is it better to use two sticks (2 x 8GB) that have a slower CAS (10) instead of (9) or is it better to double the sticks (4 x 4GB) and keep the same timing over those four sticks (9-11-11-31)?

 

This is probably just splitting hairs as the differences are so small ... but NickN has said that at the higher clock speeds memory timing does affect the stability of the OC.

 

C. T.

G.Skill 2133MHz cas 8 worked worked with the same performance however cas7 was no go.

 

HLJAMES

!!! RESULTS !!!

 

After 1 hour of Prime95 and a Vcore of 1.344 (default provided by ASUS) at 4.4GHz:

 

Core i7 4770K Temperatures never exceeded 75C.

 

H100i Coolant temperature, previously, never went above 32.5/33.0C ... now it has shot up to 39C ... obviously all of that thermal energy is no longer being trapped under the IHS!

 

The MX-4 works great ... and a small amount goes a long way.  I put a 1/2 pea sized drop on the center of the IHS and spread it real thin.  However, I'm probably going to order more CLP from FrozenCPU and use that instead.  The surfaces between those two components (H100i cold plate and the lapped IHS) is so, so minute that the liquid metal characteristics of CLP just fits much better.

 

Here is the graphic ... going up to 45 just for fun.  As soon as NickN publishes Chapter 103 of the Haswell overclock guide I'll start tuning the Vcore downwards and the NB Frequency within 300MHz of the final OC multiplier.

HLJAMES:

 

Waltzed right through Intel Burn Test ... all levels ... temps for this particular test never got out of the mid 60C range.

 

Time to start flyin ... and stop this mucking around.

 

4.3 - 1.168

4.4 - 1.264

4.5 - 1.296

4.6 - 1.312

4.7 - 1.348

4.8 - 1.392

4.9 - 1.44

5.0 - 1.49

 

HLJAMES:

 

Looks like I might be running a bit high on a stable 4.4GHz OC ... Vcore manually set to 1.29 is rock solid at 4.4.

 

Would still love to be able to hit 4.7 stable ... time will tell.

 

Thanks for all your help ... Intel Burn Test didn't even make her sweat ...

Your ambient temp is 25* C, Your water temp is 39*C. This means your radiator is not able to dissipate the heat produced at 4.4GHz.

Thats what is limiting your overclock.

 

HLJAMES

HLJames:

 

I figured at this point that maybe I incorrectly second guessed the engineers at Corsair.  I removed the two Noctua fans (to be returned) and remounted the SP120s ... dropped the multiplier to 42x and the Vcore to 1.25 (starting over).  Tests with Prime95 after an hour saw the coolant temperature stay below 34 and the core temps further were reduced by 4 to 6 C on average.  I'm going to let Prime95 run all night as per another recommendation.  This takes a ton of time and effort!

 

I should have just let the H100i do the work it was engineered to do.  Unfortunately, the fan noise has gone up as the SP120s are l twice as loud at the Noctuas but hey, whatever works ... saved $50 with the return of the Noctuas.

 

Thanks for staying involved.

 

C. T.

Want 4.7GHz? assemble two H100 radiators in series.....no delidding necessary!

 

HLJAMES