Overclocking Core i7-3770K: Learning To Live With Compromise

[blazorthon]

Wouldn't it have been easy to have a stock i5-2500 at factory speed included in the chart?

Thus showing if overclocking is actually worth the effort, the heat and possible damage?

Risk of damage caused by overclocking is almost zero and effort for overclocking is nearly zero nowadays. Temperature is also not too high, unless you're pushing past 4.5GHz or so if you buy even a $20 cooler such as the Cooler Master Hyper 212 or Hyper 212 Plus for Sandy Bridge and a temperature for Ivy Bridge can limit you to a slightly lower clock frequency. Really, the only reasons not to overclock at all would be buying a system that is completely incapable of it or if you need the system to operate within a very specific performance and power usage range.

 

[blazorthon]

Its a very very bad option... it instantly voids the warranty and the user has a high risk of killing a $200~300 CPU. If done improperly, the CPU can fail prematurely.

Intel should NOT be going with the CHEAPER option. Even not overclocking the CPU, the better quality thermal material would be better for the long term and for a cooler running chip.

I specifically said that it shouldn't need to be done and that it has to be done in order to get proper advantages over Sandy Bridge is ridiculous, practically demonizing Intel for it. Besides, there's not that great of a risk towards damaging it if you're careful, it's just tedious and probably voids the warranty (although if you don't make it obvious that you did it, you probably won't have much trouble with the warranty unless they reopen the IHS themselves to check). The only ways that you would damage the processor would be accidentally breaking the CPU because of gripping it poorly (unlikely) or by accidentally scratching the CPU with the tool that you use to open it (a little more likely, but still not too likely). It's a time-consuming and tedious job, but it's not difficult.

 

[JacFlasche]

Instead of taking the heat spreader off and using better paste, why not leave it off and submerge the board in cooled mineral oil with a jet that is flowing directly across the chip?

 

[JacFlasche]

Better yet, a chilled mineral oil/nanodiamond slurry like they use in high performance transformers.

 

[JacFlasche]

Better yet a compound made from a thermosetting ceramic matrix and nanodiamonds that would draw out the heat from the chip and then you could flow chilled nanodiamond slurry across that.

 

[JacFlasche]

Or how about a heat pipe bonded directly to the chip with a nanodiamond ceramic? Now that could move a lot of heat if you cooled the other end of it. Especially if you employed about four percent nanodiamond in the internal vapor fluid and metal foam inner walls on the pipes. Probably about a 1000% more than any external conductive or convection system.

 

[JacFlasche]

This kind of system would probably deal well with sudden heat spikes, in my guesstimation.

 

[blazorthon]

This kind of system would probably deal well with sudden heat spikes, in my guesstimation.

The problem isn't too much heat, it's that the heat isn't getting to the IHS from the CPU die fast enough because of the paste. Having better cooling systems won't change problem.

 

[CerianK]

They aren't conspiracy theories, the paste is the reason for Ivy Bridge not overclocking significantly better than Sandy Bridge...

I was suggesting that it is within the realm of reason that limiting the maximum usable voltage, thus overclock, by capping the thermal dissipation to induce soft-faults might be a stop-gap measure to protect from localized voltage/current induced gate-degradation hard-faults (e.g. SATA fiasco). However, without access to accelerated life test data on these processors, and the new gates they use, it is pointless to speculate further. It may seem counter-intuitive, but so is an entire read through Richard Feynman's book QED.

 

[Kurz]

It seems most of the comments show that people didn't take the time to read the part that Intel used TIM instead of Solder between the DIE and the IHS.

Please people read the article.

 

[obsama1]

Good article, but some flaws:

1) You should have had IB/SB/SB-E at the same clock speed. Overclock all the CPU's to 4.5GHz or just compare them at stock speeds. You need a constant.

2) This should have been done with the 3570K, as a lot more people are on the fence on whether to buy the 3570K or 2500K. I am one of those people, and I would like to see this comparison with the i5.

3) No gaming benchamarks? This desperately needs gaming benchmarks.

Also, I must say, I am buying a new build, and I will not buy IB until they get out a new stepping with improved heat efficiency, and solder. Also, we, as enthusiasts, do NOT need the IGP, which takes up 1/3 of the die. There should be IB versions of the 2550K, but it should have better heat dissipation, cooler temps., and be able to overclock higher, and have solder Make it $10-$20 more than the 3570K/3770K, or maybe the same price. I would love that.

 

[JacFlasche]

CerianK wrote

The problem isn't too much heat, it's that the heat isn't getting to the IHS from the CPU die fast enough because of the paste. Having better cooling systems won't change problem.

I guess you didn't read the article or understand my post. In my suggestions the heat spreader and thermal paste have been removed and the heatpipe is bonded directly to the chip with a nanodiamond/ceramic cement. If you understand this your comment makes no sense.

 

[blazorthon]

CerianK wrote

The problem isn't too much heat, it's that the heat isn't getting to the IHS from the CPU die fast enough because of the paste. Having better cooling systems won't change problem.

I guess you didn't read the article or understand my post. In my suggestions the heat spreader and thermal paste have been removed and the heatpipe is bonded directly to the chip with a nanodiamond/ceramic cement. If you understand this your comment makes no sense.

I must have missed that. Regardless, then you don't have an IHS anymore. That's what keeps the CPU die from being crushed. Losing it is usually not a good idea. Besides, there's no good reasons to have ridiculous cooling on Ivy Bridge because it doesn't generate much heat.

 

[oxford373]

why there is just one sandy bridge CPU,maybe one cpu is enough to give us what we need to know about ivy bridge overclocking

 

[oxford373]

maybe intel intended to cripple ivy bridge overclocking in order to keep sandy bridge-E in safe place

 

[Guest]

Are the new chips more brittle than the old chips ?

Back in the day I and plenty of other people had P3 coppermine chips without shims and the heatsink directly contacting the die. You just had to be careful to not rock the heatsink and chip the corners off. Tualatin and later resulted in IHS covers. The P3 coppermine was 106mm2, so smaller than Ivy. I'm still amazed that a site like Toms (with more sample processors than I could afford) hasn't tried ditching the IHS to see what happens.

 

[blazorthon]

Are the new chips more brittle than the old chips ?

Back in the day I and plenty of other people had P3 coppermine chips without shims and the heatsink directly contacting the die. You just had to be careful to not rock the heatsink and chip the corners off. Tualatin and later resulted in IHS covers. The P3 coppermine was 106mm2, so smaller than Ivy. I'm still amazed that a site like Toms (with more sample processors than I could afford) hasn't tried ditching the IHS to see what happens.

Up to Ivy Bridge, most CPUs that have an IHS have had the IHS soldered onto the CPU for years now (one reason for them not being removed, the other would be that it's a time-consuming thing to do properly). The whole point of the IHS is that it offers protection to the die. Power usage is a little higher than it was back then and we don't want to give people as much of a chance of destroying the CPU, so they have the IHS that allows coolers to have greater pressure on the CPU, improving the heat conductivity between the IHS and the CPU cooler. Granted, with Ivy's paste between the IHS and the CPU die, the improved conductivity is pretty much rendered meaningless, but the added protection is still there.

 

[Guest]

@blazorthon
There is no way that adding an additional piece of metal to the CPU-Heatsink contact improves thermal conductivity unless CPU-Solder-Metal-Paste-Heatsink thermal transfer is that much better than CPU-Paste-Heatsink. Is metal a better conductor, yes... but is it really that much better than paste ?

The protection against chipping off corners and clumsy installs is probably true. I'd be surprised if the pressure against the CPU has increased though. The S370 socket heatsinks I used had the spring-loaded metal clamp mechanism and you had to use a scredriver to lever that down. Every newer processor heatsink has been far easier to install than those.

Still would like to see Toms "burn" one of those samples trying an IHS-less install.

 

[Guest]

Come on Toms... give it a try. I bet your site would be flooded with views.

http://www.overclock.net/t/305443/ihs-removals-how-to-do-it-should-i-do-it-and-the-facts

 

[4Ryan6]

Some suggestions;

What I personally would love to see is tested comparisons done using software programs the majority of the forum users actually have and use, so we can compare our overclocked results with yours on a 1 to 1 basis.

You guys put a lot of work into these reviews and comparisons but if we do not have the programs to compare against, it's kinda take it or leave it results, goodbye.

It would be nice if you guys would setup a pole in the forums giving the users an opportunity to vote on the software you guys test against so we could make our own comparisons, and then all the reviewers use the same software, and testing parameters.

Meaning if you used specific settings, or rendered picture, or file to compress, you also supply that, so we can do accurate comparative results ourselves.

Plus when you say you've overclocked you leave out the details of the overclock as in how you accomplished the overclock, because some of us are getting much higher overclocks than is being advertised as the max stable attained.

Thank You, For all your efforts! Ryan

 

[blazorthon]

@blazorthon
There is no way that adding an additional piece of metal to the CPU-Heatsink contact improves thermal conductivity unless CPU-Solder-Metal-Paste-Heatsink thermal transfer is that much better than CPU-Paste-Heatsink. Is metal a better conductor, yes... but is it really that much better than paste ?

The protection against chipping off corners and clumsy installs is probably true. I'd be surprised if the pressure against the CPU has increased though. The S370 socket heatsinks I used had the spring-loaded metal clamp mechanism and you had to use a scredriver to lever that down. Every newer processor heatsink has been far easier to install than those.

Still would like to see Toms "burn" one of those samples trying an IHS-less install.

First of all, with the heat sink, the paste between the IHS and the heat sink should only be enough to fill gaps between the IHS and the heat sink, not to be an entire layer between the IHS and the heat sink. Second, no, paste isn't always that bad, but the paste that Intel uses is THAT bad. It's crap paste and compared to the fluxless solder that previous CPUs use, Intel's paste is practically an insulator. Third, yes, heat sinks can have much greater pressure than they used to. just because they aren't harder to install (all of the time) does not mean that they don't have greater pressure. Furthermore, a heat sink probably wouldn't even fit on the CPU without the IHS.

 

[Guest]

@blazorthon:
Yes, heatsinks fit after decapping as long as they're of a type that can screw down the extra mm or so. (see the overclock.net link I posted or just believe me - I recently drilled some extra holes in the bracket for an Apogee GT GPU block and 1366 bracket to use on my 2500k and just needed to adjust the depth down the extra 0.5 mm difference in IHS height between the 1366 and 1155 sockets). Plenty of others have decapped their CPUs so heatsinks of the right retention mechanism type tend to work.

Have you had experience with the old athlon/P3 IHS-less designs before ? The retention clips on S370s were well known to snap off due to the crazy amount of pressure applied. My old Thermatake SK6 used a metal "I beam" with a bend right at the cpu and you had to use a screwdriver as a lever just to be able to press it down hard enough. Pressure isn't as helpful as you may think as long as there is some.

 

[f-14]

[citation][nom]ta152h[/nom]I'm not sure why the reviewers spent so much time figuring out the shrink made the core more dense, and somehow thought this was significant. I mean, really, this happens virtually every shrink, and this time the power dropped more than normal. Common sense should be telling people this is not the cause, but somehow people aren't understanding this.It's very interesting that replacing the paste makes so much difference. This is obviously something Intel is aware of, since they do plenty of testing, and obviously chose anyway. Would a few pennies be worth it for a processor that is clearly on the higher end of the scale? Probably not.Most likely, they want to keep selling their real high end processors, and it just won't do to have the 3770K beating their 2011 processors, or being very competitive with the successors to that line when they come out. It makes perfect sense. The 3770K is still a great processor, but if you're really looking for the best, it simply will not do. You're forced to buy the more expensive 3960X, and later the even better IB successor to it, which you can bet will have far better paste, and so will overclock significantly better.It's genius. A great product for the vast majority, while leaving more expensive products as the best option for that elite that will actually spend $600 to $1000 for a processor. Well done, Intel. It's not like AMD has anything to say about it.[/citation]
pretty interesting with one flaw, i'll cite as example, remember when intel was working on a 450 mhz cpu and ibm dropped a 1ghz cpu bomb on the market? thing was so fast and ran so hot there was no heat sink that could adequately cool it before they burnt out in a day.
where is ibm now in the cpu market?
if you look at the sales reports, which is selling more and more of every year, desktop cpu's or mobile?
therein lies the gamble amd is betting on.
it's not like intel has anything to say about it.
more people have cell phones these days than computers, and yet they have more horse power than a desktop from 10 years ago.
i'd say the only thing holding back mobile is it's power supply and lack of voice command.
desk tops are rather hard to lug around to business meetings and school.
if your still planning for today, you are ten years behind and on the edge of extinction, just ask ibm and their cpu manufacturing department.
i like intel and i like amd, i however don't like greedy moves and this stinks rotten of p2 vs p3.
all i am waiting on is an amd to sweep the board with another k6-2 to k7athlon type move. the people are speaking with their money and most are buying mobile, so playing the desktop market is only going to live so long. it's as easy to see as the housing bubble and the soon to be healthcare bubble.

 

[Guest]

sounds like the ivy bridge is the pentium 4 all over again, with those kind of pathetic temps!!
throw that cpu in your computer with a few nvidiafail cards and you'd have yourself some sort of mini micromave / heating machine!

 

[SuperVeloce]

mrgaman and failmuchmate, you two failed to understand anything, miserably. Tick-tock is suppose to mean something (look it up, why 5% faster is actually very good, when it's a tick) and because of it manufacturing process and/or architecture progresses faster than it would. And those two has nothing to do with ivy not having fluxless soldering at the moment. Ivy is the most effective cpu out there, with 77w tdp at stock. Bulldozer is good value for money, but better than ivy? Underperforming clock-to-clock and 125W tdp at stock clocks for faster 3 and 4 module cpus? Dont make me laugh.