View Full Version : Q6600 @ 3.6ghz .. Finally and Stable

08-11-2008, 08:37 AM
Well i decided to up the mhz now on the CPU and it took a lot of tinkering around in bios to what settings to use and a bit of trial and error of Priming Failures.

But i finally got it and after 11hrs 5min 4secs ... i stopped Orthos .. as i say she is good enough. I usually do 8-10 hours of orthos (when ever i OC) and i have never had any issues afterwards, but one never knows as you can prime for 24hrs and still have issues later on, but 8-10hrs is ideal for daily usage and or 24/7 usage.

Temps were decent, but i must inspect The Thermal Paste and heatsink mounting .. as it appears at times ... 2 cores are 8-10C higher then the other 2 ... cores 0-1 is 8-10C higher when OCing then cores 2-3, but at idle they are only 1-2C difference.

Idle right now is 33-32-28-30

Here is screen shot just before stopping Orthos which i just took this morning at 6:44am.

Also not sure why, but Orthos states it is at 4050mhz.. lol


08-11-2008, 10:15 AM
Ok i just ran Intels linpack Program...... wowzer it will increase Temps to 74C-84C max on this CPU... for now until i check TIM and mounting.

This program gets Cores HOTTER then Orthos Prime and is Intels CPU burn in program.

I need to figure out why 2 of my cores always read 10C higher then the other 2.

Linpack 0-1 Cores 84C and 2-3 Cores 74C Max.

Orthos = 0-1 Cores 74C and 2-3 Cores 64C Max. (Were talking like 2-3 instances of 3 secs of a Max of 74C)

Here is my test results just done about 12 min ago and all is OK Residual result numbers .. 5 passes all same. If you get not the same then system is not 100% stable, means you can orthos and/or prime 10hrs say or even 24hrs and run this program and FAIL, which has happened to many OCer's who think they are 100% stable.

Also ppl read (if you want to try it out) the read me text ... only use what free memory is available to use.

Program is here:



08-11-2008, 05:31 PM
congrats madmax, nicely done

What are you using for cooling?

also, why does CPU-Z identify your ram as DDR3 while you have it listed at DDR2?

08-11-2008, 05:46 PM

I am using a Thermalright Ultra 120 Extreme with 2 x 120mm fans on it .. one is a Noctua P12 and the other is just a 1200 rpm fan i had laying around. .. push pull configuration and the heatsink is installed in a North, South direction .. fan on bottom blowing up and fan on top blowing up/out towards the top case fan ... i have a Antec 900 case.

I lapped the Heatsink yesterday... /ubbthreads/images/%%GRAEMLIN_URL%%/smile.gif

Thing is i need to see if the IHS is concave which i do suspect as 90% are on intel CPU's and if that is my issue for the 10C higher temp on 2 cores .. other then that it is Awesome and i never see above (50-55 load) on many Daily Programs that push the CPU ... and as for gaming is good too.

But i like to fold again and i need to solve the 65C-70C full load on folding on those 2 cores .. 65C is like pushing what i like .. 70C well thats acceptable range still, but getting warmer .. so i not like it .. so for now i not folding on CPU till i figure what i can to solve issue and NO i am not underclocking to solve it... /ubbthreads/images/%%GRAEMLIN_URL%%/smile.gif

I may and if i really think i should .. may lap the IHS.

As for CPU-Z its a bug with it and it is the latest version too.

Here is the SPD section of CPU-Z: ... DDR2


08-11-2008, 06:05 PM
This is the ram here that i have:

http://newgskill.web-bi.net/bbs/view.php...desc=asc&no=106 (http://newgskill.web-bi.net/bbs/view.php?id=g_ddr2&page=2&sn1=&divpage=1&sn=off&ss=on&sc=on&select_arrange=headnum&desc=asc&no=106)

Also it is PC2-8000.... i update my signature again DAMN CPU-Z has it listed as PC2-6400... lol

Well it is right for PC2-6400 @ 400 speeds = DDR2 800, but at 500 speeds = 1000mhz it is PC2-8000

08-11-2008, 06:36 PM
I should also note i am not running bios 0704 anymore like in my testing results of my thread on the Q6600 and Ati 4850.

I am now running Bios 1103 which is a lot better for my setup, but there is a newer Bios looming on ASUS's FTP wesbite and its not a Beta from what i can see as it is not listed as such ... i downloaded it and i may flash Bios to it as some others in a forum i go to once in awhile have flashed to 1201 and they stated it is equal to the 1103 stability wise all that was added to it was more compatibility with certain memory modules... other then that it appears to be just as equal to the 1103 Bios .. some said it allowed them to OC higher then the 1103, but one never knows.

08-13-2008, 02:28 PM
Update on 10C core difference:

Well i managed to solve my 10C core differences.

I took heatsink off again and lapped it some more .. like 3 hours more ... smooth as a babes butt.

Then instead if using AS5 i figured i would try out the thermalright chillfactor paste that came with the thermalright ultra 120 extreme heatsink.

I spread the paste accross the CPU CAP only which is what is recommened for it, but not on heatsink even though it says to do so, but i don't agree as the sink is not an exposed PIPE sink and to much paste will hinder in heat transfering.

My CPU cap is flat .. not concave ... which is suprising as most are on intel CPU's, but at times you get a good one that ain't.

I put it all back together and loaded up to windows and WOW it dropped idle temps even more about 4-6C from what was previously running at which were decent in the first place.

Now at idle before the temps were 4-6C difference between core 0-1 and cores 2-3, now they are only 1-3C at idle.

As for full load Orthos the temps were only 2-4C difference now . not like before 10C and at times 12C difference between 0-1 and 2-3 cores.

Now how is Chillfactor ..... very similar temps as AS5, but a little lower by about 2C now.... that is pretty good and no break in time is required with it.

I also used a penny in between the mounting bracket and block of the heatsink to help comepensate for the lapping = 2-4mm when done so heatisnk is snug now.

I am folding again and the temps are Averaging: 63-60-55-57 MAX .... i have seen (Peak Temp for 1-2secs) is 64-66C on core 0.

Thats pretty awesome at 3.6ghz Air cooled for a Q6600 running 1.43v Max Load.

In Orthos for about 11-12 min MAX temp i seen was 63C (Thats about 4-6C drop), again on Core 0 which is typical as core 0 usually is about 2-3C higher then the rest of them on any Quad-Core and many Dual-Cores.

I may go with water cooling in the future, but $300.00-$400.00 for decent/great setup for only 6-10C less is expensive for the amount of degrees in reduction plus as many know the Q9450 and newer CPU's generate less heat supposily, but i read about many ppl who have such CPU's and at 3.6ghz their temps are equal to mine and they even at a lower Voltage then i am at.

So water cooling is not really needed that much anymore unless your looking to get 4.0ghz or higher and thats what i may do in the future on a future CPU.

08-15-2008, 02:52 PM
<span style="color: #FF0000">Updated Linpack version now it is v1.4.</span>

VERSION 4 has some minor updates the maker of the GUI interface made some updates is all .... <span style="color: #FF0000">it has Version 3 Binaries </span>as <span style="color: #CC0000">there is no update yet to intels linpack files</span> .. so version 1.5 will have those .. which right now there is BETA ones, but the maker of this GUI interface is not going to be a guinea pig for them.

CHANGE LOG IS BELOW in this post:

The binaries in v3 are harder testing system stability .... i even had to adjust my NB, Vdimm and PLL volts just had to up the PLL and NB 1 notch and lower Vdimm a couple notches ... versions 1 and 2 passed np's.

This program is better then Orthos and Prime put together and is only good for Intel CPU machines, but when you think your stable.. you may think again after running this.

Remember it will push the CPU to 70-88 degrees depends on you OC and cooling of course ... those temps i listed are based upon what i have been reading on many ppl who are running this test on OC's of 3.4-4.0ghz and on good air cooling or water cooling and dual-core or quad-core.

I would state use Orthos for about 10 hours and then run this puppy for 5-10 loops = about 15-20 min if using 3000mb of memory = 3GB being used for testing for errors and CPU stability.


vice versa this program 15-30min and then Orthos 10hrs minimum, but many will only run this program as it is FAST and harsher then many prime programs and is the BEST for overall PC stability.

Like any program its use at your own risk as if you damage anything its your risk, but so far i have only read a few that had some complaints about this program killing a onboard NIC card or a CPU ... but both were due to cooling issues and or Older Mobo that could not hack the stress.

If anyone has any questions about this program and how to use it PM.

<span style="color: #FF0000">This versions change long.</span>

- Fixed a bug with percentage display that resulted in displaying 0% instead of the actual
pass rate
- Removed redundant prompt at the beginning asking if user wants to run the program
- Added customize option for selecting the stress level

- Added automatic error-detection feature (choose at the beginning)
- Added elapsed time at the end
- Added updated 32-bit and 64-bit Intel(R) Linpack binaries from Intel(R)

- First non-beta version
- Added automatic detection of total available physical memory
- Added ability to select testing stress level
- Added automatic detection of OS type (X86/X64) [Itanium is not, and will not be supported]
- Further program optimizations applied
- Added more notes in the program for user-friendliness
- Added changelog
- Added credits section in readme (was missing before for some reason)

v1.1 beta
- Optimized memory equation (originally by Cronos) to use maximum memory available and leave
headroom to prevent memory paging


08-16-2008, 10:31 PM
It looks like it increases the temp, but it does not really. Press CTRL C on the program window while having your temp reading program open. It will drop 20 Celsius as soon as the program stops. No cooler can cause such a big drop so fast...
I think the reason why it shows such high temps is because it does not share any cpu cycles with anything. This causes your temp reading program to report incorrect temperatures.

08-17-2008, 12:25 AM
Yes my cooler drops it that fast .. 80C to 35C within secs.

Its realtemp that i use and trust me its a true temperature that it reads when it says 75-80C on the cores.

Its not incorrect tempertures ... its fluctuating temps on the cores when heat/stress is applied.

Also when you touch the heatsink it gets warm when using that program and more hot air is blowing out of case.

Its just like prime and Orthos when full load = 63-60-57-58 on cores . but stop program = down to 35C in secs for me anyway.

This intel linpack program is Intels program that is used to test CPU's . its their burn in test they use before sending them out for sale.

And it does/will increase temps by at least 15C-20C then what Prime95 or Orthos will do.

Again no program be made 100% to check for instability, but this one will find it a lot faster then Orthos or Prime95 will.

Where say Orthos may error in 3 hours .. this program will find it within in 15-20 min.

I will use this from now on for fast checking for MAX OC settings/stability and then Orthos 10-12 hrs at least.

For best results run at least 10-15 test loops.

08-18-2008, 10:48 AM
I use the stock cooler with stock thermal paste. I highly doubt it can drop 20C within the blink of an eye...

08-18-2008, 12:21 PM
Well it drops on mine when the CPU is stressed the fluctuation of the cores temps go up and down and once the program stops whetehr it is Prime95/Orthos or Linpack the cores temps drop quickly due to no more heat stress on them and with good/great cooling the temps will drop within secs.

The cores are not going to stay at 80C when the testing stops as good cooling will allow the temps to drop very very fast.

Like i don't even know what your argument is as over the past 5-6 years all benching programs stress the CPU and or memory and video card ... depends on the test used.

Even running folding on my Video the temps go to 64C full load and if i shut it folding off the temps drop within secs (1-2) to 48C.

What your saying is that effective cooling can't drop temps fast?

Now if you have non effective cooling then yes the temps will remain higher once tests have stopped and slowly go down.

A good Water Cooling Set would superior to what i have and what i have is Great Air Cooling for now, but in the near future i may go with a water cooling setup, if its worth while and i maybe able to drop about 15C more on full load .. which would be awesome.

Also when my cores were at 80C and the test was done it dropped to 38C with secs . then after about a minute it dropped to 35C on core 0 and the rest are always a little cooler then core 0 so they would have been around 28C-30C .... YES within secs - 65 secs i would say.

08-18-2008, 01:48 PM
What I am saying is that this program is the only program where my temperatures are being displayed as 80C. On all others it displays as 50-55C.
Something is affecting the temperature reading, and I highly doubt it's the program putting more stress onto the cores.

And I am not talking about a couple of seconds either, I am talking about dropping 20C as soon as I stop the program.

08-18-2008, 02:17 PM
Here is a lot of info .. read it all:

Yes its the only one that stresses the CPU like heck and memory as well the more memory USED the more stressful it will be on the CPU and memory.

The CPU temp programs DON'T lie .. those are the temps under high-high-high stressful CPU and Memory Calculations which there is no other program that i am aware of that does that, Not Orthos, Not Prime95, not CPU-BURN, Not OCCT.

And with effective cooling the temps will drop just like Orthos and prime95 and so forth ... good cooling is all.

"LINPACK is a collection of Fortran subroutines that analyze and solve linear equations and linear least-squares problems. The package solves linear systems whose matrices are general, banded, symmetric indefinite, symmetric positive definite, triangular, and tridiagonal square. In addition, the package computes the QR and singular value decompositions of rectangular matrices and applies them to least-squares problems. LINPACK uses column-oriented algorithms to increase efficiency by preserving locality of reference."

"LINPACK was designed for supercomputers in use in the 1970s and early 1980s"

"The Linpack benchmark is a method of measuring the floating point rate of execution of a computer by running a program that solves a system of linear equations.

Linpack was originally developed by Jack Dongarra to give computer users an idea of the time it would take a machine to solve complex problems in linear algebra. The results of the Linpack benchmark can be used to estimate the speed with which a computer is likely to run an individual program or multiple programs.

Because Linpack performs tests in a specialized way (solving sets of multivariable equations), it is more accurate for some programs than for others. In order to gain a true understanding of how fast a computer executes a specific application program, or to compare the performance of one computer with others for use with a particular application, the ideal approach is to test each machine by running the application in question. Linpack is nevertheless useful as an overall benchmark and has gained wide acceptance as a general method of evaluating computer and supercomputer performance."

Heres a quote off of a forum page of another user:

"Remember that in most daily usage of a PC, no one will reach linpack temperatures unless some very hot summer nights coming soon"

"Supposedly much better as 8 mins of this = 40 hrs of Prime95 and up to 22C hotter!
Plan to try it tonight once i get home."

"The primary goal of this benchmark is to measure Gigaflops rating of (Intel) CPU(s). Cluster Linpack version can be used to measure performance of computational clusters, and even huge Super-Computers.

The software solves mathematical problem - arbitrary large set of linear equations, and displays the rate of solving, measured in Gigaflops."


Frequently Asked Questions on the Linpack Benchmark


Intel® Math Kernel Library 10.0 - LINPACK

Intel Optimized LINPACK Benchmark packages can help you in your quest to obtain the highest possible benchmark results for Intel® architecture-based systems. These free packages are implementations of the LINPACK benchmarks which use BLAS and LAPACK software that has been highly tuned for maximum performance on Intel® Xeon® processor-based and Itanium®-based systems. These same high performance BLAS and LAPACK routines are available to software developers in the Intel® Math Kernel Library.


Linpack 1000 benchmark code:


08-18-2008, 02:45 PM
I am going to try something later on and will let you know the results.

08-18-2008, 07:44 PM
O ooooooooooo KKKKKK

I wait for your temp results which i already know what your doing... hehehehe.

And still the conclusion is this program produces more mathematical and memory stressing then Orthos and Prime95 can do.

Remember these chips of today are SMALL and will cool off fast with effective cooling. Its not like an engine that takes hours to cool down.

Anyway no sense talking about it much more as you fail to realize that the cores ... once heat is applied the temps will sore and then cool fast with effective cooling .. thats as simple as it gets.

08-18-2008, 07:59 PM
Well, this is what I tested :

Ran Linpack, let both cores heat up to 80 C.
Started Prime 95 in dual core mode, let it run for a while. Temps were still at 80 C.

Stopped Linpack, temps dropped down from 80 to 60 C in less then a second.

Now I don't know why it does that, but I do know that Prime 95 was still keeping both cores occupied at 100%. There's no way the real temperature can drop 20 C in less then 1 second, when another program is keeping both cores occupied at 100%.

08-18-2008, 09:31 PM
Yes they will drop that 20C cause linpack was stopped (and prime95 was still running) and linpack it was no longer pushing the CPU calculations to the MAX and temps too.

So prime95 pushes your CPU to 60C MAX = less stressing on CPU and memory as well.

The cores are small = tiny and effective cooling will drop them fast within 1-2 secs.

Prime95 does not max it out like linpack does and linpack stress the CPU and memory more then prime95 will or even Orthos.

Like i don't really understand whats your arguement as i stated many times linpack STRESSES more then all other programs will.

The cores are small and when any more heat is applied they will sore up the temp and when the program is stopped it will drop down close to idle temps again within secs as it is no longer being stressed to the max.

How long do you think it should take for 45nm - 65nm cores to drop with good effective cooling? What 2 hours? or 20 MIN.. not going to happen.

Tell ya what SUPER GLUE a socket A heatsink on your CPU and then run the test and see how long it takes to cool down .... LOL

Dude this is funny as your thinking it should take a longtime for these tiny cores to drop temps.

It is REAL TEMPS so it will fluctuate to MAX when stressed and then back down within secs when not stressed.

Linpack uses a more harsher stressing then any Prime program does and thats why the temps will go up at least 20C more then prime95.

When you shut off prime95 or even orthos the temps will drop back down to idle temps within secs too.

08-18-2008, 09:35 PM
By the way does your build pass Linpack standards, run at least 10 loops one instance.

08-22-2008, 07:48 PM
Here you go.

http://img167.imageshack.us/img167/518/80cdv6.jpg (http://imageshack.us)
http://img167.imageshack.us/img167/518/80cdv6.5af0791f02.jpg (http://g.imageshack.us/g.php?h=167&i=80cdv6.jpg)

08-22-2008, 08:38 PM
Very good .. glad to see it does pass ... so stop complaining as it appears to be very very stable like mine and many other OC'ers.

More the loops and Memory used the better for testing, some ppl even run 100 loops.... wow

Also for better accuracy 64-bit OS is recommend and best to run.