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

[de5_Roy]

^^ it's the same guy with different nicknames. trolls like this makes real amd fans (not fanboys) look bad.

 

[blazorthon]

[citation][nom]mrgaman[/nom]time to stop with the bulldozer crap and you intel fanyboys have to live with the fact that the 22nm tech intel has made out to be 30% bettter with 3D power gates etc etc is crap!!hahahahahahahaahaha now look who is laughing last yes, a AMD FANBOYIVYFAIL FTL...[/citation]

[citation][nom]failmuchmate[/nom]anyone fancy some liquid ivyfail die? haha fkn thats the funniest thing i have seen EVER in Tech history!!5% improvement stock to stock against the 32nm sandy, these new ivyfails even have new tech like 3d power gates to reduce excess power and such!!LAWL INTEL, YOU FAILED[/citation]

[citation][nom]intelfailssinceforever[/nom]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![/citation]

Stupidity is being rampant today. Bulldozer is even hotter than Ivy Bridge at stock. Switch out the paste for better paste and Ivy Bridge suddenly becomes the best overclocking CPUs around. You have to live with the fact that the 22nm process node, like any other process node, is superior to its predecessors and the fact that Ivy Bridge is a tick in Intel strategy, meaning that not only was it not supposed to be a large performance increase over Sandy Bridge, but everyone who cared knew that it would only be a small performance improvement.

At least unlike Bulldozer, Ivy Bridge's CPU die is an improvement on or at least equal to its predecessor in every way. Considering that P4 was actually hot because it used a lot of power, Bulldozer is more like P4 than Ivy Bridge, especially since Ivy Bridge's problem is easily fixed, whereas you can't fix Bulldozer's low performance per core and per watt.

 

[Guest]

22nm is just a tick? lol i thought advancing to the next LEVEL is suppose to be the HUGE ball game, not the revision of it

 

[blazorthon]

[citation][nom]lol ok[/nom]22nm is just a tick? lol i thought advancing to the next LEVEL is suppose to be the HUGE ball game, not the revision of it[/citation]

32nm to 22nm is a die shrink. All ticks are die shrinks. All tocks are new architectures on a previously used process node. Yes, 22nm is currently a tick. Haswell will be the 22nm tock. Why don't you people learn about the subject before posting such idiocy.

[citation][nom]harrypotterpcwizard[/nom]ivyfail is fail, everyone was going on for ages on its going to be 30% better.. and its not.that is fail people...Erm my bulldozer overclocked @1.50volts on air doesnt even get close to the temps of these intel chips at stock.. where do you people get your info from?i think the ivyfail is bad, one its 22nm it should of been better on power/heat and performance.wheres bulldozer is same old 32nm and they only used 4 modules = bottleneck /clap to amd here, but my 8150fx walks all over a i7 2700k @ 4.8ghz+ its cooler[/citation]

I get my numbers from the Bulldozer machine sitting in the house next to me and the Ivy Bridge numbers from tests here at Tom's. You have no idea what you're talking about, let alone what everyone else here is talking about. Only an idiot would blame Ivy Bridge's technology for these heat problems. The problems are completely caused by the paste and are solved by an easy (albeit time-consuming) exchanging of the paste under the IHS with high quality paste. Ivy Bridge is the most power efficient series of CPUs in the consumer market.

If you really want to talk about Bulldozer, then how about the fact that old 65nm Core 2 Quads meet or beat the 32nm Bulldozer CPUs in gaming performance. Until Piledriver comes out, AMD is the fail here. Ivy is an improvement over Sandy, although it's not a great improvement (only ignorant people were expecting it to be a significant improvement).

Furthermore, where the heck did you get this idea that an FX 8150 can ever beat an i7-2700K? Not in ANY workload, except maybe one or two minor victories out of hundreds. The i7 would have FAR greater single and lightly threaded performance while having slightly superior highly threaded performance.

 

[vakuma5000]

Ivy Bridge IS better than Sandy Bridge.
Clock-per-clock Ivy Bridge is faster, uses less power, and generates less heat.
The die shrink was a success.
The ONLY reason people are complaining about Ivy bridge is because it won't overclock as high as Sandy bridge.
This is due to the paste.
And for people that DON'T overclock(they exist), Ivy Bridge is an improvement over Sandy Bridge.
An Ivy Bridge i3 will surely be better than a Sandy Bridge i3.
If Ivy Bridge had the same solder setup as Sandy Bridge had it would overclock better. There are a number of potential reasons why intel may have decided to purposely handicap their Ivy Bridge, and it''s pretty OBVIOUS they PURPOSELY handicapped it. Any company capable of creating a CPU as badass as Sandy Bridge, is smart enough to know that the paste they are currently using in Ivy Bridge is limiting it's overclocking potential. They did it on purpose. We can speculate about they "why", but the end result is -
For an enthusiast that likes to overclock, Sandy Bridge is the better bet. Plus it will be going down in price due to Ivy Bridge, so there is a benefit for OCing enthusiasts from Ivy Bridge.
It's not the benefit we were anticipating, but there is a benefit.
We will surely get awesome overclockability out of Ivy Bridge-E and/or Haswell, and in the meantime let's just enjoy the awesome performance from both Sandy and Ivy Bridge(s).

 

[dreadlokz]

Waiting for the revision on this one!

 

[anartik]

Good review but you overlooked what appears to be the real key to overclocking Ivy... under volting which another site touched on. I was shocked at the temps (high 70's-mid 80's) I was getting at stock running prime and was not having a great deal of luck overclocking the 3570. It looks to me like the default voltages on a number of boards including the Gigabyte UD5 I have are too high. Undervolted I am now running at 4.5 with an old 2x50CFM fan Thermalright Venomous-X and staying low 70C running prime small fft. Playing BF3 several hours the highest temps were in mid-high 60's. Not at system now but as I recall the default is 1.18 vcore and I'm running at 1.14v with loadline set to normal. Still not clear on the full behavior of loadline on this board but if cpuid is to be believed loadline on extreme (also on auto when overclocking) was kicking voltage as high as 1.45v under full load. On normal loadline it looks like the peak voltage is around 1.28 and a bit lower on low. I can go 4.6+ but that point starts requiring increases in voltage and heat starts going through the ceiling. At full idle throttled down to 1600mhz temps are low 30C.

 

[monsta]

Hope Haswell has more to offer than IB

 

[youssef 2010]

[citation][nom]digiex[/nom]as the manufacturing process gets smaller = Smaller die size, supposed to be cooler temperature,but, with small die size = small area for heat dissipation,...an irony that needs to be solved.[/citation]

It seems Intel needs to redesign their heat spreaders for more efficient heat transfer.

 

[blazorthon]

[citation][nom]youssef 2010[/nom]It seems Intel needs to redesign their heat spreaders for more efficient heat transfer.[/citation]

Intel just needs to go back to the fluxless solder that they used in previous CPUs. The IHS is fine (although it's not like I'd be unhappy if Intel also improved the IHS).

 

[gibs0]

[citation][nom]slicedtoad[/nom]Recommending a closed loop liquid cooler? really?They perform worse than decent air coolers. The lower end ones (think corsair's h60) perform like mid-low range air coolers and cost more. The better ones (h100 or antec's 920) perform on par or worse and with more noise than a similarly priced noctua. If noctuas looks too ugly for you, phantek and several others offer similar performing models.The only reason to get closed loop lc is for looks. I admit they do give your build a nice clean look. That doesn't warrant "So, we're recommending a closed-loop liquid cooling setup, at least" though. If you'd changed that to "We're recommending higher end aftermarket coolers for a decent oc", it would've made more sense.Anyway, I'm just nit-picking a single line from the article. All in all, it was a good read. It just makes me upset to hear wrong advice.[/citation]

I have a Corsair H100 on my 3770K, and ran a Noctua NH-U12P SE1366 on my older Core i7.

My 3770K I currently have OC'ed to 4.6Ghz and it idles around 24 and full load, it gets to mid 40s/low 50s. On my older i7 overclocked to 3.9Ghz (from 2.66 stock), it idled in the high 30s and full load, it got to high 60s/low 70s.

Closed loop liquid is definitely beneficial.

 

[monsta]

Got my 3770K at 4.6 so I can't complain , it performs really well.

 

[JacFlasche]

The more I think about what Intel did the more offensive I find it. I have been waiting to upgrade and now I have to wait for Haskel, why -- because they decided to cripple the cpu they released. Is their any other reason? To save a tenth of a cent per unit perhaps? -- doubtful. This in an insult and I truly hope that AMD gives me a reason to ditch Intel.

 

[blazorthon]

[citation][nom]JacFlasche[/nom]The more I think about what Intel did the more offensive I find it. I have been waiting to upgrade and now I have to wait for Haskel, why -- because they decided to cripple the cpu they released. Is their any other reason? To save a tenth of a cent per unit perhaps? -- doubtful. This in an insult and I truly hope that AMD gives me a reason to ditch Intel.[/citation]

Not that I'm justifying what Intel did with this paste (I despise it too), but if you're feeling adventurous, you can get an Ivy CPU, remove the IHS (it's fairly safe, but it is time-consuming), switch out the paste for highly conductive paste, re-attache the IHS and it will beat Sandy Bridge in overclocking.

Lapping the top of the IHS and the bottom of the CPU cooler will also help a little. Modding can be fun when it pays off

 

[JacFlasche]

http://www.xtremesystems.org/forums/showthread.php?280892-3770k-IHS-Removals-CPU-temp-dropped-from-79C-to-71C

 

[blazorthon]

[citation][nom]JacFlasche[/nom]http://www.xtremesystems.org/forum [...] 79C-to-71C[/citation]

Exactly. There might be even more thermally conductive pastes out there somewhere too and even better results could be had.

 

[JacFlasche]

Just Intels way of getting overclockers to invalidate their warranty I guess.

 

[blazorthon]

[citation][nom]JacFlasche[/nom]Just Intels way of getting overclockers to invalidate their warranty I guess.[/citation]

Probably. Shh, don't tell Intel and problem solved if you can make it look like the IHS was never removed.

 

[oxford373]

[citation][nom]blazorthon[/nom]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.[/citation]

first i only believe information from professional websites like toms hardware, and using cheap thermal paste isn't that cripple overclocking by at least 10 degrees .

 

[blazorthon]

[citation][nom]oxford373[/nom]first i only believe information from professional websites like toms hardware, and using cheap thermal paste isn't that cripple overclocking by at least 10 degrees .[/citation]

Well, since Tom's already told us that the paste is a huge problem for heat transfer and is likely the greatest cause of Ivy Bridge's heat problem, I guess you would be contradicting yourself by not believing it just because I told you too. Also, you don't know as much about paste as you think you do if you don't think that using paste with poor thermal conductivity between the IHS and the CPU die would cause huge heat problems even when there isn't much heat being generated. It's simple physics... Taking something very thermally conductive with something that is not very thermally conductive in comparison means that heat will be more of a problem. Considering that everyone who has tested it has results that contradict your comment and Tom's agrees with this, I'm going to agree with Tom's and the people who tested this.

 

[Guest]

http://www.xtremesystems.org/forums/showthread.php?280892-3770k-IHS-Removals-CPU-temp-dropped-from-79C-to-71C

There we go, see posts 36 and 65 for people running completely decapped (IHS-less) ivy bridge CPUs. They didn't shatter under the weight and it looks like people are saying 10 to 15C drop on their overclocked chips.

 

[Guest]

De-lidded already? I believe "the other chips" require a lot less (cost/effort) to achieve exceptional computing; each CPU Manu provides exceptional computing through its processors a fact lightly talked about amidst the GHz wars. So Intel wants people to pay 500+ for a "high-end" CPU but then take the top off it to compete w/a lesser chip in OC'n? Is that warrantied?? Errrm I got lost

Why is every "Intel Fail" downplayed as "intentional" or "good move" while the competitors(yes Intel has competition) "abnormalities" or "low benchmarks" are catastrophic in the eyes of Intel users. This sentiment seems to be the only viable angle Intel users argue up their cpu's from as if any behavior Intel chips exhibit surely has to be intentional. This process of analysis is phail and biased; 101 DEGREES C!!? And at such a low voltage = poor design. The CPU evidently draws more power than they calculated.... L M A O is all im saying. You won't see this applied in ANY gamer or enthusiast plats because it's uncontrollable-similar to nuclear power. Yea looks and feels good until it "misbehaves" or is "upset". one of these processors melting from constant 100% loads seems feasible and scary. Intel IVY/SANDY are "now chips" so enjoy that fact and remember that "other CPU's" will always have plenty of market share purchasing Intel alternatives. 2600k still has 4years or better as viable at this rate of innovation. BD seems ahead of the game imho: 9GHz..... -_-

 

[blazorthon]

[citation][nom]DigiRevII[/nom]De-lidded already? I believe "the other chips" require a lot less (cost/effort) to achieve exceptional computing; each CPU Manu provides exceptional computing through its processors a fact lightly talked about amidst the GHz wars. So Intel wants people to pay 500+ for a "high-end" CPU but then take the top off it to compete w/a lesser chip in OC'n? Is that warrantied?? Errrm I got lost Why is every "Intel Fail" downplayed as "intentional" or "good move" while the competitors(yes Intel has competition) "abnormalities" or "low benchmarks" are catastrophic in the eyes of Intel users. This sentiment seems to be the only viable angle Intel users argue up their cpu's from as if any behavior Intel chips exhibit surely has to be intentional. This process of analysis is phail and biased; 101 DEGREES C!!? And at such a low voltage = poor design. The CPU evidently draws more power than they calculated.... L M A O is all im saying. You won't see this applied in ANY gamer or enthusiast plats because it's uncontrollable-similar to nuclear power. Yea looks and feels good until it "misbehaves" or is "upset". one of these processors melting from constant 100% loads seems feasible and scary. Intel IVY/SANDY are "now chips" so enjoy that fact and remember that "other CPU's" will always have plenty of market share purchasing Intel alternatives. 2600k still has 4years or better as viable at this rate of innovation. BD seems ahead of the game imho: 9GHz..... -_-[/citation]

The CPU does not draw more power than advertised... The paste between the IHS and the CPU die acts like an insulator compared to fluxless solder; the paste slows down the transfer of heat substantially. That is why people report huge gains when they ditch Intel's paste. That 9GHz overclock on BD was done on an 8150 with all but one module disabled, so it was not with all cores active and furthermore, the old Netburst Celerons were able to hit upper 7GHz and lower 8GHz numbers, should we consider them ahead of the curve since they held the GHz crown up until BD came out? No.

BD is not ahead of any game just because it can clock very high. AMD gets mocked whereas Intel doesn't because even with how hot Ivy can get at too high of an overclock, it is still far faster than BD and is even more power efficient than Sandy Bridge. Ivy is still an improvement over Sandy in every way besides overclocking performance at stock and with a moderate mod, it is then also better than Sandy at overclocking. BD can't say the same.

Having high clock frequency tolerance means nothing in practicality when those frequencies aren't practical. It is an overclocking record, nothing more, and nothing less. Even with that record, it would not beat a Sandy or Ivy Bridge CPU overclocked with the same liquid nitrogen or hydrogen cooling. Heck, cheaper phase-change cooling could get a Sand\y or Ivy well beyond the performance of a 9GHz BD CPU in both single threaded and highly threaded performance. Bringing the Ivy up past 6GHz would let it take the single threaded crown away from Bulldozer at 9GHz and the Ivy could go even farther if we started disabling cores on it to level the playing field.

These overclocks aren't about performance; these overclocks are about simply getting the highest frequency that is stable enough to take a screenshot of CPUz and maybe a quick benchmark. This is suicide overclocking, not practical overclocking. Intel wins in practical, but AMD whens in suicide overclocking where the greatest frequency is wanted (at this time).

 

[jimmyd1964]

Maybe Intel needs to produce a LN2 Home Kit. Whoa....I just had a P4 Flashback. Sort of like a Old Lady Hot Flash. What did they put under the heat spreader? Peanut Butter. ha ha
Someone wasted a lot of money building the Hadron,
News Flash: Xi_b^*0 will decay quickly into several different products of lesser mass. The researchers at Intel found, when 2V are applied to the Ivy Bridge processor.

 

[eriko]

[citation][nom]mayankleoboy1[/nom]Nice review!1. Are there plans to release any K CPU's without the HD4000? will they OC higher?2. Any chance of intel releasing a second stepping of K-series IB chips?[/citation]

Indeed, I just purchased a 3920XM, and I've got HD4000 graphics, which I'll never use - even in my desktop replacement / laptop.

I've done 4.4GHz at stock voltage, and wonder how it would have gone without the necessary graphics on the die.

I mean really, who buying an Extreme processor wants / needs integrated graphics?