Overclocking limits?

[known12345]

Was wondering, how exactly do you know when you reached the limits of the overclocking potential of your chip, ignoring cooling for now?

I mean how do you know that your chip cant keep going forward as opposed to you having insufficient cooling?

With a q6600 I was able to get to 3.6 at 1.496v to run on prime95 for 10 hours (Antec P182 fans high, ABIT IP35-E, Arctic Freezer 7 Pro, side panels off, roommate's desk fan at high) but not without seeing a high core temp of 65-70C (average of about 68C) and a high PWM temp of 85-100C (average of about 93C). Obviously it would be very foolish of me to leave my settings the way they are as I am extremely stressing my parts (and the fact that the settings are completely overkill) but it made me wonder if I would possibly hit 3.7. More importantly, I wanted to know how high could my chip go AND successfully complete prime95 for 10 hours.

Obviously the higher I attempt to overclock the card, the more cooling I will need to ensure the card doesn't take damage but lets just say cooling wasn't an issue (for now). I know the inability of my FSB setting to POST is an obvious sign that the card has reached its limits but are there other signs that signify I am getting close to my limit?

I set my FSB to 420 and 1.54something v and I was unable to POST. After resetting my BIOS (boy was I terrified for a second), I set the FSB to 405 and 1.5075v and I successfully POST and started up windows and confirmed I had a FSB of 405. At 4am in the morning and feeling very tired, I decided against running prime95 (for now) as I knew the temps were going to be extremely high and I would have to find some creative way to cool the comp which would be easier if I had some rest. Instead I wondered if my cpu would POST at higher FSB so I set it to 410 and 1.5075v and was able to go to windows and check CPU-Z to verify the new FSB. Then I went to 415 and 1.5075v and it again I was able to go to CPU-Z and verify the new FSB. Finally I jumped to 425 and 1.5075v and I was able to get to windows and check CPU-Z but this time the computer felt EXTREMELY laggy as if I was running too many programs at the same time. It took a long time to open CPU-Z, the ABIT BIOS program, Core Temp and various other computer checking programs. At that point I wondered if I had hit the very limit of my CPU.

On a side note, is there a particular reason why I was able to POST at 425 at 1.5075v and not at 420 at 1.54something v? Ive read about situations where sometimes someone is unable to POST at a particular FSB (or range of FSB) and simply passing the number or range (if there is a number or range; they could have possibly hit their limit) may allow them to try their card at higher temperatures.

Anyway, back on topic; when the computer shows extreme lag in an overclock, does it signify that the limits of overclocking has been reached or on the verge of being reached? If that is the case, should I even attempt to try to run prime95 at 425 or just simply attempt to run prime95 at 415 (or possibly 420) and call it a day whether or not I am able to run with no errors for 10 hours? Ofcourse I will try my very best to ensure adequate cooling is available and sufficient vcore is provided but I suspect 3.6 was by far the limit for my card and nothing short of getting a better PWM cooling motherboard, a better cpu cooler, or possibly a water cooling system would alleviate the heat coming from my cpu/motherboard.

Also quick question, ive heard of people having motherboard failures and having to replace them... is overclocking the reason why their motherboard failed?

 

[BillParrish]

Anyway, back on topic; when the computer shows extreme lag in an overclock, does it signify that the limits of overclocking has been reached or on the verge of being reached?

Depends, I am wondering if your CPU has begun to throttle back due to excessive heat because EIST and/or C1E or Thermal Monitor 2 (TM2) have not been disabled.

Somewhat of a moot question, its obvious something is not right and something is on the verge, question is verge of what ? You seem aware you are stressing your components hard (actually the 3.6 seems sustainable) but then continue. They don't call these suicide runs for nothing. Its your stuff and a lot of fun, as long as you can afford to replace what fails.

On a side note, is there a particular reason why I was able to POST at 425 at 1.5075v and not at 420 at 1.54something v? Ive read about situations where sometimes someone is unable to POST at a particular FSB (or range of FSB) and simply passing the number or range (if there is a number or range; they could have possibly hit their limit) may allow them to try their card at higher temperatures.

It has to do with the MCH/northbridge chip changing its internal timings at certain frequencies. It may be that somewhere between 420 and 425 your MCH relaxed its internal timings to cope with the higher FSB and thus was able to boot. Or its just one of those weird things. At these high frequencies the circuit board traces act like radio transmission lines with signal reflections, sensitivity to harmonics etc. The "board" is a multitude of electrical engineering marvels that all must work perfectly all the time or it will not run so it is almost impossible to attribute the boot failure to any one specific thing without about $100K worth of test equipment. Eh, stuff happens. Might even boot at that settings today. /shrug.

Also quick question, Ive heard of people having motherboard failures and having to replace them... is overclocking the reason why their motherboard failed?

I would say no, the root cause of motherboard failure when overclocking is the end user not understanding and reducing/eliminating the risk of component failure before undertaking the exercise. Active or extra cooling on MCH/northbridge chip ( it will thermal trip to save itself like the CPU). The PWM CPU voltage supply circuitry and the PLL clock generator chip would probably be the next two major areas needing attention. A lot depends on the design margins of a lot of the board components as far as voltage and current and temp ratings. This is why "good" boards cost more (well it should be why expensive board are expensive) they are designed with better quality parts that cost more and have labor intensive extra cooling like heatpipes etc. and basically are designed to handle with long life the extra stress extreme OCing places on the board. Unfortunately many times a so-so design has a fancy heatpipe slapped on it, a few extra USB ports are strapped on, a couple of options added to the bios, and the price is jacked up for the suckers. Only just recently do manuf bother to talk about quality Mosfets and solid caps. Informed consumers give most businesses nightmares and imo that why you do not find ANY real technical information available on motherboards and most people have to rely on reviews by non-engineers. A circuit schematic - the most basic source of information to enable an accurate assessment of the design - forgetaboutit. /rant off.

 

[MaxTheLimit]

You sir are far more bold than I when you overclock. Personally I like to keep a safe margin between me and thermal throttling. You tread a thin line sir.

 

[known12345]

Depends, I am wondering if your CPU has begun to throttle back due to excessive heat because EIST and/or C1E or Thermal Monitor 2 (TM2) have not been disabled.

I though I disabled C1E and EIST but now that I think about it, I enabled C1E and EIST because I was running at 3.0 and when I set it to 3.6 and it POST, I completely forgot about C1E and EIST (was awfully late and I was dead tired) and probably forgot to disable it. That would explain alot...

Somewhat of a moot question, its obvious something is not right and something is on the verge, question is verge of what ? You seem aware you are stressing your components hard (actually the 3.6 seems sustainable) but then continue. They don't call these suicide runs for nothing. Its your stuff and a lot of fun, as long as you can afford to replace what fails.

Ahh I completely forgot about the dangers of overclocking!!! Iono if it was because it was really late or because overclocking had gone well without any major problems that I thought that problems/limits were nonexistant... Now that I am back down to Earth I think I will just try to find a way to lower the temperatures at 3.6 instead of pushing for 3.7 because there is no way I can replace what fails =(.

It has to do with the MCH/northbridge chip changing its internal timings at certain frequencies. It may be that somewhere between 420 and 425 your MCH relaxed its internal timings to cope with the higher FSB and thus was able to boot. Or its just one of those weird things. At these high frequencies the circuit board traces act like radio transmission lines with signal reflections, sensitivity to harmonics etc. The "board" is a multitude of electrical engineering marvels that all must work perfectly all the time or it will not run so it is almost impossible to attribute the boot failure to any one specific thing without about $100K worth of test equipment. Eh, stuff happens. Might even boot at that settings today. /shrug.

Hmmm I see. Setting FSB at 420 and having it not POST could have simply just been one of those "weird things" and having it POST on 425 was normal procedure... or it could have been the other way. Either way, I think ill just stay satisfied at 400 so I dont have to worry alot about weird things that start to happen when the verge of something is in reach.

I would say no, the root cause of motherboard failure when overclocking is the end user not understanding and reducing/eliminating the risk of component failure before undertaking the exercise. Active or extra cooling on MCH/northbridge chip ( it will thermal trip to save itself like the CPU). The PWM CPU voltage supply circuitry and the PLL clock generator chip would probably be the next two major areas needing attention. A lot depends on the design margins of a lot of the board components as far as voltage and current and temp ratings. This is why "good" boards cost more (well it should be why expensive board are expensive) they are designed with better quality parts that cost more and have labor intensive extra cooling like heatpipes etc. and basically are designed to handle with long life the extra stress extreme OCing places on the board. Unfortunately many times a so-so design has a fancy heatpipe slapped on it, a few extra USB ports are strapped on, a couple of options added to the bios, and the price is jacked up for the suckers. Only just recently do manuf bother to talk about quality Mosfets and solid caps. Informed consumers give most businesses nightmares and imo that why you do not find ANY real technical information available on motherboards and most people have to rely on reviews by non-engineers. A circuit schematic - the most basic source of information to enable an accurate assessment of the design - forgetaboutit. /rant off.

Gotcha... so EXTENSIVE cooling on the motherboard (particularly North Bridge, PWM CPU voltage supply circuitry, and PLL clock generator chip) is a must to ensure the motherboard doesn't reach high temperatures and overheat/take on too much stress. Yeah, im looking at my motherboard manual and the motherboard's website and detailed descriptions (or the names) of various parts/components of the motherboard is non-existent. I have to rely on a book that gives a very general view on the motherboard and various informed posts by individuals such as yourself to find out about the different components of my motherboard.

You sir are far more bold than I when you overclock. Personally I like to keep a safe margin between me and thermal throttling. You tread a thin line sir.

I was simply more ignorant and foolish at the time (hopefully). I think i'll stop trying to push the cpu (for now) and focus on getting very good cooling then continue my efforts.