ChickenInferno

No, running LinX for 10 runs like I said about 20 posts ago. It takes 20 minutes to run 10 runs on average.

This was probablly the most important thing for you to read out of all of this.

__________________
I edit my posts alot. So please reread my posts if I've edited them; It's for your benefit not mine.

The Truth about Temperatures and Voltages
Flash your Nvidia card with a USB Flash Drive
Start Folding Now

tht13

k but im gonna do dis in mornin since i havent bin sleepin wel latel

__________________
'All there is to being a comedian is saying intelligent things stupidly.... or maybe its stupid things intelligently, its not rocket surgery I can tell you that right now'
Rich Hall

TwoCables

In my own personal experience, I have found that it's absolutely best to make sure that your overclock is capable of doing all of the following:

• Approximately 12 hours of the Small FFTs test in Prime95.
• Approximately 6-12 hours of the Blend test in Prime95 (this is the default test selection when you open Prime95).
• Approximately 5-10 passes in Memtest86, Memtest86+, or in the Windows Memory Diagnostic that is built into Vista and 7. This checks for errors; it is not a stress tester.
• At least 30-50 Linpack runs using the maximum amount of available memory. For this, I highly recommend using LinX. It is the best linpack tester because it has the best GUI, tooltips for everything, and a really good and short Read Me file. Be aware that this test will make the load temperatures approximately 10-15°C higher than with the Small FFTs test, but do not worry about it unless that makes the temps exceed like 85°C or something.

I think that I can personally guarantee you that if your overclock can do all this, then it is truly rock-solid stable, especially for the way we use our computers. I mean, every overclock is unstable by nature, so our goal should be to make sure that it's practically 100% stable.

Now, you absolutely do not need to do this kind of testing until you are "done" overclocking. I mean, when you think you've reached a point where you're happy with the overclock, then do what you can to make that overclock stable enough to do the above. But until you're done, just do lighter testing, like maybe the following:

• Up to 5 hours of the Small FFTs test
• 1-2 hours of the Blend test
• A couple of passes in Memtest
• About 10 linpack runs using all available memory.

But when you're ready to stop overclocking for a couple of days, or for several months, then do what you can to make it stable enough to do the heavier testing in the first set of bullet points above.

Now, if you're wondering how you would ever be able to do a 12-hour stress test, then here's a tip: start the test before going to bed on a night when you have to basically get up and leave the next morning for work or school or something like that. This way, you will certainly be away from the computer for at least 12 hours, and probably longer. Also, be sure to monitor the temperatures for about the first 5-10 minutes of the stress test so that you know what the maximum temperature is for the entire duration of the test.

Also, do not multi-task during a stress test. Here's a combined quote from another thread called "Prime95- Doing other things" posted by Fors30:
Take notice what TheSubtleKnife said here. I have done my own testing on this to see if it's true, and as it turned out, it is very true. So by all means, let the computer give a stress test its undivided attention.


Temperature Monitoring and Information

Keep using Real Temp. And, of course, leave Real Temp open during a stress test. And always remember that the whole point in trying to keep temperatures down is to avoid instability. There are many people who would love to have you believe that your maximum safe temperature is like 70°C, or more specifically: 71.4°C. But they don't really know any better. But 71.4°C is the Thermal Specification of the Q8200. Here's Intel's explanation of the Thermal Specification:
And because the TDP was mentioned, I'll post that too:
In other words, the Thermal Specification was measured at the same place where dust would land if the CPU werre just sitting on a desk or something. Therefore, we don't need to worry about it.

Instead, we must be concerned with the Tj. Max. According to the information that ChickenInferno found, the Tj. Max for the Q8200 is about 100°C. So, it's only important to pay attention to the Distance to Tj. Max. However, some people have had nearly instant instability at like 70-75°C, and so this is probably why 70°C is usual recommendation. This is the thread that I'm referring to, and it's also in ChickenInferno's signature:

TJ Max and Temperature Monitoring Programs

Be sure to read that as carefully as you possibly can because it's extremely important. Also be sure to follow the link to Real Temp's Documentation page, which is also known as the Installation & Calibration page.

All of this information basically means one thing: don't worry about the idle temperatures because they're not accurate anyway. Once the core temperatures are below approximately 50°C, then all we can say is "hey, they're below 50°C, and that's all I know even though Real Temp says 38°C". The thermal sensors were designed for only one thing: to tell the CPU to shut off if the Tj. Max is reached. Therefore, the only time the sensors are 100% accurate is when the Tj. Max has been reached. But as the temperature gets farther and farther away from the Tj. Max, the accuracy decreases.

In other words, our temperature monitoring methods are pretty inaccurate, crude and rudimentary. But Intel has left us with no choice. Thankfully, they're not so inaccurate that they're absolutely useless. But even so, Intel has recommended that we do not use these sensors as veritable thermometers, and they most certainly recommend that we don't use them to report idle temperatures. But speaking of which, if you had idle temps that were exceeding like 50°C, then it would be a good time to pay attention.

But thankfully, Real Temp can be calibrated to report temperatures that are likely pretty accurate. That is the reason why I said its Documentation page is also known as the Installation & Calibration page.


Vdroop

I recently learned from the AnandTech article below that it's a bad idea to reduce, eliminate or try to even reverse vdroop. So, whatever you do, do not make any attempts at changing your vdroop. Oh, and vdroop is when the actual voltage droops on full-load, like with the Small FFTs test.

Here's the AnandTech article. Please carefully read the page linked and then the next page. It's just 2 pages of this very long article.

http://www.anandtech.com/cpuchipsets...spx?i=3184&p=5

Now, I know that my sig rig happily states that I have performed the vdroop pencil mod on my motherboard, but I first saw this article yesterday, so I am no longer happy about my vdroop mod now that I know the truth about it.

__________________

tht13

ok thanks for this advice, atm my target is to get to 1600fsb and then once ive reached that ill see how much higher i can go.
atm im at a 1373 fsb and testing. my plan right now is to run a linx test 10 runs all mem after each oc
and then overnight run a prime95 test overnight.
ill keep you updated

__________________
'All there is to being a comedian is saying intelligent things stupidly.... or maybe its stupid things intelligently, its not rocket surgery I can tell you that right now'
Rich Hall

tht13

ok the overclocking has been goin ok i got to fsb 1453 on normal voltages and 1463 on just over but then trying 1473 i just cant get it even when nugging up each of da main volt (vcore,fsb and spp) these are at versy high but its still not working. wats wrong?

__________________
'All there is to being a comedian is saying intelligent things stupidly.... or maybe its stupid things intelligently, its not rocket surgery I can tell you that right now'
Rich Hall

TwoCables

Which voltages did you try this time around?

__________________

tht13

i tried vcore 1.35 fsb 1.25 dimm1.7 spp1.35 and mcp 1.5
but i think i must have set these be4 i went to bed and they seem to work but i still think they are to high for such a low frequency

and it turns out this was for fsb 1483 not 1473 but still from 1473 the vcore and vsb had to jump up from vcore 1.25 to 1.35 and fsb from 1.15 to 1.25 and only for 10mhz fsb

__________________
'All there is to being a comedian is saying intelligent things stupidly.... or maybe its stupid things intelligently, its not rocket surgery I can tell you that right now'
Rich Hall

tht13

ok i have good news and bad
the good news is that i was fed up of having to go to rly high voltages just to get a 10mhz increase fsb so i decided just to try jumpin to 3.0ghzclock(3.00107 to be accurate 1715fsb) at norm volt didnt work i then raised volt nd it works!!!!!!! runni linx atm
bad news is the volt were vcore 1.45 fsb 1.40 dimm1.7 spp1.40 and mcp 1.6
and with these high volt i have a load temp on core 0 of 58c core 1 54 core 2 54 and core 3 58 peaks so im gonna try to lower them, since i dont know how long ill last on these temps. even with my cooling (i think my xmas present this year ll be windows 7 nd liquid coolin)

__________________
'All there is to being a comedian is saying intelligent things stupidly.... or maybe its stupid things intelligently, its not rocket surgery I can tell you that right now'
Rich Hall

TwoCables

Wait, are those load temps from at least 10 minutes of the Small FFTs test in Prime95?

__________________

tht13

no load on 15 min on linx

__________________
'All there is to being a comedian is saying intelligent things stupidly.... or maybe its stupid things intelligently, its not rocket surgery I can tell you that right now'
Rich Hall