CPU OC Wall?

[johngalt]

SOrry for the Double Post, but the board only allows for so many quotations per post....

C1E Enabled

<snip>

C1E Disabled

loco - first off you'll need to decide whether you can live with high temps or not - either you have a very poor cooler, or it was poorly installed. I haven't seen temps like yours with my OCed system *at full use* - In fact, I've never see 55 deg C yet....

As for my mobo being better to OC, according to Fumz, it ain't so good - DFI seems to be the way to go in terms of true OC-ability, but I picked this as a nice balance between enthusiast and run of the mill generic boards.

I might look into another machine in a year (I've been bitten by the bug, I want to build a truly powerful rig this time - hopefully for under $3000 USD)

My point is that heat isn't what kills a cpu quickly. The cpu in question can handle up to 70C. More importantly, it shuts itself off if it gets too hot. Yes, more volts generate more heat; however, the Tuniq is one of the best air coolers out there and yes, it's much better than "a cheaper one": AnandTech: Thermalright Ultra 120 Extreme: Is More Better? But again, heat, per se, isn't a problem, it's volts. A cpu on water being over-volted has a much greater potential to suffer the quick death that alot of over-volted cpu's suffer.

My point, is that the price of the water cooling unit would not justify the gains since he's already near the limits of safe voltage on air and doing just fine. The only thing water would do, aside from cost a lot, is to give a false sense of security that you could safely over-volt the cpu... and this just isn't true.

Great point about the misconception that Water cooling *automatically* allows for better overclocking, unless, of course, it has to do with the Northbridge cooling - that is one of the hottest places on the planet, net to my GPU under load, and I am pretty sure that with a bit of better cooling I think I might be able to break the wall.

I am guessing that even at 3.6 GHz my CPU will not be trying to drain the system of above 1.5 V, but this is theory on my part, based upon the MCP cooling issue these boards have.

And, you're right, Fumz, I should actually be quite happy with a 33%OC, back in the day 15% was astounding, until Intel started binning CPUs heavily.

 

[locoeng]

loco - first off you'll need to decide whether you can live with high temps or not - either you have a very poor cooler, or it was poorly installed. I haven't seen temps like yours with my OCed system *at full use* - In fact, I've never see 55 deg C yet....

I can live with that heat. I'm using an el cheapo Rosewill cooler and I've reseated 4-5 times cleaning off and reapplying the TP each time. While the idle temps are high, under full load they rarely go over 60C. I keep the A/C on 72F, but the PC is in the loft and it's usually 5-6 degrees hotter there as far as ambient temps go. Unless I get a new cooler I doubt I'll see anything much lower.

 

[johngalt]

You can live with the temps - now - but as soon as you OC they'll go up....hence my point.

 

[locoeng]

There is no fan attatched to that Tuniq is there?

Picasa Web Albums - johnlgalt - The Beast

That thing is a monster isn't it.....never mind I found the midship fan ( Tuniq - Tower 120 Cooler).

EDIT: If you want to go for that Scythe/TRUE120 cooler combo, I'll take the Tuniq off your hand for a fair price.

 

[Fumz]

The fan is inside the heatsink... smack dab in the middle.

 

[johngalt]

lol loco - I was looking to replace the fan part only.

That Tuniq is a beast, but for under $50 USD I can't complain at its sheer cooling power for an Air cooler.

 

[skunksmash]

hey fumz...

isn't it amazing how far the clocks can go when the chip is kept cool....
yes its LN2 but he is probably pushing 2.0v vcore through a 45nm chip to get that
so i still stand by the fact that heat plays a bigger part than volts


Andre Yang, a civil engineer doing his PhD. in the US obtained a high-performing Core 2 Duo E8600 processor, he succeeded in overclocking this chip to a stellar 6.70 GHz clock-speed (FSB 672.16 x 10.0). The processor uses liquid-nitrogen (LN2) cooling and surprisingly this processor was overclocked on a motherboard with the Intel X48 chipset, the ASUS Rampage Extreme when previous record-holders across several recent processors made use of either the Intel P35 or P45-based motherboards.

john .....that is still a sweet OC & i was not suggesting you buy a WC setup ....i was merely stating i think you hit your limit on air.

 

[johngalt]

I know you weren't, skunk :D

 

[Fumz]

hey fumz...

isn't it amazing how far the clocks can go when the chip is kept cool....
yes its LN2 but he is probably pushing 2.0v vcore through a 45nm chip to get that
so i still stand by the fact that heat plays a bigger part than volts


Andre Yang, a civil engineer doing his PhD. in the US obtained a high-performing Core 2 Duo E8600 processor, he succeeded in overclocking this chip to a stellar 6.70 GHz clock-speed (FSB 672.16 x 10.0). The processor uses liquid-nitrogen (LN2) cooling and surprisingly this processor was overclocked on a motherboard with the Intel X48 chipset, the ASUS Rampage Extreme when previous record-holders across several recent processors made use of either the Intel P35 or P45-based motherboards.

john .....that is still a sweet OC & i was not suggesting you buy a WC setup ....i was merely stating i think you hit your limit on air.

lol...

Nobody argued heat wasn't a factor, it's just not a factor here. It's also true that as you increase vcore heat rises exponentially. Nobody said anything different; but again, heat's not really the issue here as johngalt's temps are well below 70C.

It's also true that you can get some insane sub-zero clocks as you're able to pump tons of volts into the core and keep the cpu cool enough to operate. However, electromigration occurs no matter what. It's brought on first and formost by volts. Heat exacerbates the problem, but it's still the volts that kill cpus. It really doesn't matter how cool you're able to keep a chip, over-volting is going to kill it.

Nobody runs LN 24/7. Chips just don't last that long under that kind of voltage. While it's a neat feat, it's wholly unrealistic as an option for the normal user. Watercooling comes nowhere near phase or LN. As it applies here, since we're already talking about a vcore of 1.45, it's really not worth the cost. He'll maybe shave 5C off his current temps, which are already very low given the voltage.

 

[skunksmash]

hey fumz...

isn't it amazing how far the clocks can go when the chip is kept cool....
yes its LN2 but he is probably pushing 2.0v vcore through a 45nm chip to get that
so i still stand by the fact that heat plays a bigger part than volts


Andre Yang, a civil engineer doing his PhD. in the US obtained a high-performing Core 2 Duo E8600 processor, he succeeded in overclocking this chip to a stellar 6.70 GHz clock-speed (FSB 672.16 x 10.0). The processor uses liquid-nitrogen (LN2) cooling and surprisingly this processor was overclocked on a motherboard with the Intel X48 chipset, the ASUS Rampage Extreme when previous record-holders across several recent processors made use of either the Intel P35 or P45-based motherboards.

john .....that is still a sweet OC & i was not suggesting you buy a WC setup ....i was merely stating i think you hit your limit on air.

lol...

Nobody argued heat wasn't a factor, it's just not a factor here. It's also true that as you increase vcore heat rises exponentially. Nobody said anything different; but again, heat's not really the issue here as johngalt's temps are well below 70C.

It's also true that you can get some insane sub-zero clocks as you're able to pump tons of volts into the core and keep the cpu cool enough to operate. However, electromigration occurs no matter what. It's brought on first and formost by volts. Heat exacerbates the problem, but it's still the volts that kill cpus. It really doesn't matter how cool you're able to keep a chip, over-volting is going to kill it.

Nobody runs LN 24/7. Chips just don't last that long under that kind of voltage. While it's a neat feat, it's wholly unrealistic as an option for the normal user. Watercooling comes nowhere near phase or LN. As it applies here, since we're already talking about a vcore of 1.45, it's really not worth the cost. He'll maybe shave 5C off his current temps, which are already very low given the voltage.

ok i think its fair to assume both points have merit.......
i DO NOT believe he should buy a WC setup to further his OC
but i do think there is only so far you can go on air......

would you not agree...?

i still believe that in ''domestic systems'' heat is still more of a factor than volts as electromigration will take far longer to have an effect than say someone with a 40c ambient resulting in a 50 or 60c idle...(its just an example)

 

[Fumz]

Of course there are merits to your points. I fully agree with them for the most part.

True, as far as a pure OC, there's only so far one can go on air. However, those pure OC's are done only because they can be done, not for any practical purposes. It's the same reason why guys climb big tall icy mountains; just because. They OC a chip they don't really care about frying in the first place, bench it, and get their pats on the back for a job well done.

Sometimes guys fall off the mountains; sometimes guys fry chips. Meh... such is life.

As for the practical limits of an OC, water isn't that much better than a Thermalright 120 or the Tuniq Tower these days. Air cooling really has improved a lot over the last few years. 3 or 4 years ago water was able to demonstrate much better results over air; now, not so much. Heatpipe technology has improved to the point where financially speaking, it's hard to justify water's cost. To get really good performance on water costs a bit now. You finish filling out an order at Danger Den and you may come to realize you could have just as easily bought a Vapo-Chill.

If, for example, it costs you $50.00 to get to 4GHz on an E8400, then spending $250.00-$500.00 to get to 4.5+ isn't a good return on your dollar, right? That was my only point, not that keeping temps down isn't important.

I think the example is a bit on the extreme side and forces me to agree with you? Yes, lol, the guy with an ambient case temp of 40C who idles at 55-60C (15-20 degree rise over ambient) is most likely going to have have heat issues, but then again, since we're talking 131F, he probably shouldn't be turning on his machine during the hottest time of the day in the middle of the desert. :D

As cores reduce in size, the more careful you have to be about volts. These 45nm cores are apparently very picky and going above 1.4 can lead to failure very quickly.

 

[skunksmash]

Of course there are merits to your points. I fully agree with them for the most part.

True, as far as a pure OC, there's only so far one can go on air. However, those pure OC's are done only because they can be done, not for any practical purposes. It's the same reason why guys climb big tall icy mountains; just because. They OC a chip they don't really care about frying in the first place, bench it, and get their pats on the back for a job well done.

Sometimes guys fall off the mountains; sometimes guys fry chips. Meh... such is life.

As for the practical limits of an OC, water isn't that much better than a Thermalright 120 or the Tuniq Tower these days. Air cooling really has improved a lot over the last few years. 3 or 4 years ago water was able to demonstrate much better results over air; now, not so much. Heatpipe technology has improved to the point where financially speaking, it's hard to justify water's cost. To get really good performance on water costs a bit now. You finish filling out an order at Danger Den and you may come to realize you could have just as easily bought a Vapo-Chill.

If, for example, it costs you $50.00 to get to 4GHz on an E8400, then spending $250.00-$500.00 to get to 4.5+ isn't a good return on your dollar, right? That was my only point, not that keeping temps down isn't important.

I think the example is a bit on the extreme side and forces me to agree with you? Yes, lol, the guy with an ambient case temp of 40C who idles at 55-60C (15-20 degree rise over ambient) is most likely going to have have heat issues, but then again, since we're talking 131F, he probably shouldn't be turning on his machine during the hottest time of the day in the middle of the desert. :D

As cores reduce in size, the more careful you have to be about volts. These 45nm cores are apparently very picky and going above 1.4 can lead to failure very quickly.

agreed....lol

 

[johngalt]

Sigh. Agreed. I will test some more independently but I am afraid that I have, indeed, reached a wall - either because of massive MCP heat or because the CPU simply "can't do no more".

Unless loco wants to swap CPUs with me? :P

 

[locoeng]

Unless loco wants to swap CPUs with me? :P

*cough*<!-- google_ad_section_end -->

 

[johngalt]

Tell ya what, I'll give you my CPU *and* my Tuniq for your CPU.... :P

 

[locoeng]

I'll take this for my Q6600 and you can keep the cooler

EDIT: The E8400's have come way down since I built my rig...

http://microcenter.com/single_product_results.phtml?product_id=0281097

I was originally going to get this chip on sale for $200, but they were out of stock in most places and the sites that had them were anywhere from $230-$260...so I figured I could get two more cores for the same price and got the Q6600 off ebay for $230 shipped.

 

[johngalt]

Hmmm - your CPU-Z shows a Quad Core CPU lol....

And if I were to buy that you most definitely would *not* get it from me....I was originally trying to buy that CPU but everyone ran out of stock when I was building so I settled for this 6600 instead.....

 

[locoeng]

I was originally trying to buy that CPU but everyone ran out of stock when I was building so I settled for this 6600 instead.....

You are talking about the 9450 aren't you? I built mine the same time as a guy on another forum, but he did get the E8400. It's a good chip, but I can still keep up with him with the programs we use (ConvertXtoDVD, DVD Rebuilder etc.) with mine at stock and he's trying to O.C. like mad. <!-- google_ad_section_end -->

 

[johngalt]

Yes, the 9450 - still getting an eBay notification because I saved a search on that CPU....

 

[locoeng]

Hmmm - your CPU-Z shows a Quad Core CPU lol....

That is what I have...originally was gunning for the E8400, but didn't see paying the same price for it vs. the Q6600 so I got the Q6600 instead.