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

Page 3 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Status
Not open for further replies.

freggo

Distinguished
Nov 22, 2008
2,019
0
19,780
Maybe I am the odd one out, but I don't get all this overclocking stuff.
Wait 6 month and the next generation CPU/GPU will be faster, mostly negating the overclocking efforts.

OC to me is like buying a new car and than spending a load of time on 'fine tuning' the engine etc to make it faster. Hardly anyone in his right mind does that either.
 

halcyon

Splendid
[citation][nom]freggo[/nom]Maybe I am the odd one out, but I don't get all this overclocking stuff.Wait 6 month and the next generation CPU/GPU will be faster, mostly negating the overclocking efforts.OC to me is like buying a new car and than spending a load of time on 'fine tuning' the engine etc to make it faster. Hardly anyone in his right mind does that either.[/citation]
Overclocking is free though...and to some...fun.
 


A well overclocked Nehalem i7-920 can be as fast as or faster than a stock Ivy Bridge i7 which is three generations afterward (Westmere, Sandy Bridge, Ivy Bridge). That's part of the point of doing it. It's not as power efficient, but it is faster. Besides, it's not a very time-consuming activity, especially with new CPUs where you just hop into the BIOS, increase the multiplier setting, and then restart (or in the case of Ivy, just do it with the program instead of going into the BIOS and it's even easier).
 
idle, load power consumptions are definitely lower when using ivy bridge cpu. but how long one uses ivy bridge on load? for example, if someone plays pc games for 2-3 hours every day but doesn't do any cpu intensive works rest of the time. comparing with ivb review, the average power consumption doesn't seem to have gone down substantially. i mean that i think sandy and ivy both have similar average power consumption while ivy has lower load and idle power consumption...
 
G

Guest

Guest
Overclocking used to provide a more significant benefit, back when a PC was obsolete in 3 years, and every new program would only run on the latest hardware. If overclocking brought a 30% performance improvement, that would be the equivalent of saving several hundred dollars. ie you could wait one additional year before replacing the PC.
That argument no longer holds. PC's today can do every regular task demanded of them. My 6 year old dual-core PC is just as fast as my new IB build at everyday tasks - internet, Microsoft office etc. Overclocking brings diminishing returns now (who cares if Powerpoint slides open 1/100 second faster?)
AMD has realized this paradigm shift, and is switching from 'more speed' to 'more efficient'. Intel has gone this way with IB. Overclock enthusiasts have not made the paradigm shift, and thus you see the growing disconnect between their wishes and the new products being released.
Desktop PC's will probably soon be replaced by low-power mobile devices that are as just as fast. That's where the real push is now. CPU OC enthusiasts will become a small minority, since no 'must have' programs require more speed.
GPU's have not reached this stage yet. Games continue to be released that force modern GPU's to their knees, so overclocking definitely still helps there (an OC'd 7850 will easily beat a stock 7870, at $100 cheaper, and it makes a real difference in many games)
 
Overclocking used to provide a more significant benefit, back when a PC was obsolete in 3 years, and every new program would only run on the latest hardware. If overclocking brought a 30% performance improvement, that would be the equivalent of saving several hundred dollars. ie you could wait one additional year before replacing the PC.
That argument no longer holds. PC's today can do every regular task demanded of them. My 6 year old dual-core PC is just as fast as my new IB build at everyday tasks - internet, Microsoft office etc. Overclocking brings diminishing returns now (who cares if Powerpoint slides open 1/100 second faster?)
AMD has realized this paradigm shift, and is switching from 'more speed' to 'more efficient'. Intel has gone this way with IB. Overclock enthusiasts have not made the paradigm shift, and thus you see the growing disconnect between their wishes and the new products being released.
Desktop PC's will probably soon be replaced by low-power mobile devices that are as just as fast. That's where the real push is now. CPU OC enthusiasts will become a small minority, since no 'must have' programs require more speed.
GPU's have not reached this stage yet. Games continue to be released that force modern GPU's to their knees, so overclocking definitely still helps there (an OC'd 7850 will easily beat a stock 7870, at $100 cheaper, and it makes a real difference in many games)

This is an enthusiast site. Try comparing that 6 year old computer to a new, high end computer in some serious workloads and see how poorly it stacks up. AMD didn't realize any shift, they simply failed to compete with Intel in performance for too long, so they are shifting their focus (I think it's a huge mistake because the modular architecture is great, AMD simply refuses to implement it properly for some ridiculous reason). Intel's Ivy Bridge is a tick in their tick-tock strategy and Haswell, the tock, will be a big performance leap over it. Ivy was only supposed to be a slightly improved CPU that uses less power so that Intel could get experience with the 22nm node before they make their next architecture on it. Intel isn't switching anywhere. Overcocking doesn't have any more diminishing returns than it did before, it's problem with Ivy is that Intel used poor quality thermal paste instead of fluxless solder. If not for that, Ivy would be far better than Sandy Bridge for overclocking. Heck, just using high quality paste instead of low quality paste would be enough to let Ivy beat Sandy by a little.

Desktops aren't getting replaced by laptops. Desktops have and probably always would have far greater performance for the money than a laptop and their maximum performance for a generation has and probably always will be higher than that of a laptop. Furthermore, the CPU and such are not what change performance of programs and files opening, the storage does that. Using an SSD versus a hard drive (or even a new SSD versus an old SSD) shows huge differences. As file size gets larger, storage performance will need to keep up in order to continue to provide adequate performance.

Faster hardware will always have huge benefits for high performance computing, a huge market and a big part of the enthusiasts. Content creation and editing has and always will need faster and faster machines to keep up. For many people, programs such as Photoshop are must have programs and I guarantee that they would not be able to do their job nearly as well on a low end machine as they would on a high end machine and as more performance heavy filters and larger files become part of the job, faster and faster hardware becomes necessary. You see those huge prices on high end hardware? Those are there because many people need that hardware in order to do what they do be it a hobby or a job.

For gaming, there are many high end gamers who constantly need faster hardware to keep playing games at the high settings that they like. You mention graphics as not reaching a state of being enough yet. Did you even consider the fact that faster and faster CPUs are needed for high end gaming? Nothing slower than a stock Sandy/Ivy i5 can handle even the middle of the high end graphics, let alone the upper high end graphics systems. This will only get worse as time goes on and more intensive games are released.

When it comes to regular people doing light things such as web browsing and watching video playback, then sure, better and better computers are not necessary nearly as often. However, even then, with GPU acceleration and other regular things coming about in web browsing and such, having a modern system is still always getting more and more important for many people. Try watching even 1080p video playback on a computer with a GMA 950 as it's GPU (a common on-board GPU made from 2005 to 2008 or thereabouts) and see how it goes.
 

quicksilver98

Distinguished
Mar 16, 2010
53
0
18,630
[citation][nom]halcyon[/nom]I haven't shot for anything greater than 4.4Ghz on my 2500K. I wonder if my sad little Corsair H100 would handle 5Ghz well.[/citation]


My 2700K runs at 5Ghz with an H100 cooling it. Most 2500K's are only good for 4.8-4.9 tops
 

CerianK

Distinguished
Nov 7, 2008
261
50
18,870
Rather than supporting conspiracy theories involving the use of cheap thermal compound on the spreader, I wonder if doing so might be intentional to cap the usable maximum voltage, which might otherwise cause damage to the new 3D gate structures (more-so in some areas of the chip than others). Basically, if you're serious about risking burning these up... do it on something other than air. That should keep returns to a minimum.
 


They aren't conspiracy theories, the paste is the reason for Ivy Bridge not overclocking significantly better than Sandy Bridge. Ivy Bridge generates significantly less heat than Sandy Bridge as it uses significantly less power, but the paste is more insulating than the solder, so Ivy Bridge gets hotter even though it generates less heat. This can easily be proven by the experiment that canoeguy1 conducted, let alone the Japanese site that Tom's referred to as having replaced Intel's poor quality paste with higher quality paste and getting dramatic improvements.
 
G

Guest

Guest
The question is whether Ivy Bridge 3770K can outperform Sandy Bridge 2700k on a closed water loop with the TIM swapped out and the IHS delidded. That IB can't handle voltage increases without cooling is already old news.
 

horaciopz

Distinguished
Nov 22, 2011
446
0
18,960
Maybe im gonna get downvote, well i dont care.

But lets see AMD fiasco Bulldozer about 10% faster than phenoms > huge fail in thermal and power consumption.

Intel "fail" IvyBrige processors about 10% faster than SandyBridge > less power consumption but fail in thermal output.

Oh Irony, so much waited Ivybrige to get this kind of disappoint, It cant give a entirely meaningful upgrade for the last generation, as AMD failed. Well overclockers! gotta stick with your "old" SB processors while Haswell comes.
 


There's nothing wrong with Ivy Bridge's thermal output. The problem is that the paste acts more insulating than the solder does, so the little bit of heat that Ivy does generate can't get out of the CPU die fast enough because the paste practically holds it in compared to the fluxless solder and even high quality paste.
 

Isaiah4110

Distinguished
Jan 12, 2012
603
0
19,010
So it sounds like you can't do much to bring down the heat with stock clocks. I wonder how the i5 3450 and 3570K will perform in smaller cases like the . Will the move to low quality thermal paste cause them to run too hot even at stock speeds?
 

SuperVeloce

Distinguished
Aug 20, 2011
154
0
18,690
From when is 1.3Volts+ on Ivy a stock voltage?!

and that thing about fans spinning up to slow just doesn't make sense. It's more of a "heatpipe lag and transfer of heat from die to heat-spreader and then to heatpipes" problem in my opinion. Even at 100% fan speed there would be thermal throttling.
 


You can lower the voltage. The i5-3570K and i7-3770K have been overclocked a little over 4GHz on undervolts and they can have even lower voltage if you leave them at stock frequency (making them also much more energy efficient at stock than they are with stock voltages, let alone compared to even Sandy Bridge). Regardless, they aren't to hot at stock anyway.
 

medic10617

Honorable
May 24, 2012
1
0
10,510
Disable intel thermal throttling in the bios, as well as speed stepping. It stops the downclocking of the processor and allows the OC to remain constant. I'm running my 3570k at 4.3ghz 24/7 @ 1.15v. However if I bump the voltage up to 1.35, disable the above mentioned thermal throttle, and it's prime95 stable at 4.85ghz 24/7. Using a corsair H100, in push/pull with Noctua fans, max temp is 77C. I might push it further, but currently using 4 sticks of ram, so bclk is only at 102.
 

ta152h

Distinguished
Apr 1, 2009
1,207
2
19,285
[citation][nom]iam2thecrowe[/nom]it was never intended to be a huge upgrade, they never claimed anything like that, its just a die shrink with a few tweaks and additional features. And FYI the PIII was a lot faster when things were being written to take advantage of the new SSE instructions it provided.[/citation]

Not so, the Pentium II 450 and Pentium III 450 were the same process; the only difference was SSE. Coppermine was the shrink.
 

sburck

Honorable
Mar 28, 2012
30
0
10,530
So about ripping off the die cover and adding some quality thermal paste... Is there anymore information on this? How is the die cover attached? How dangerous is this modification?
 


The IHS is glued on. It can be removed with a knife, but you need to be very careful in order to avoid causing irreparable damage.
 

horaciopz

Distinguished
Nov 22, 2011
446
0
18,960



It doesn't matter what is the problem, there is a problem! They can't just solve it with a flash bios, change the stock cooler or anything else that a mayor brand can do to solve some situations...

The problem is: Ivy can't overclock as high as SB for X reason, it doesn't matter what the reason is, it can't go that high as an its predecessor. Temperatures go crazy more and more as it is overlocked and for a regular overclocker it is just disappoint because our CM 212evo can't do the trick with as it did with a SB chip. So a 2700k with 4.6Ghz will do the same as a 3770k 4.2ghz. Where's the upgrade? Lower power output? Okay, I wouldn't pay 300+ for a newer chip just to save some Watts that a 600W or more powerful PSU won't notice...


Thats my point, IB was so much waited for a better performance and lower TDP / Heat, but we got not that significative performance boost, lets say 20% or more at the same clocks and at least an equal overclocking headroom... We have just a newer chip with a better IGP and a little more efficient... :sweat: Not fair for a consumer.
 


It isn't difficult by any means, but you must be very patient and have a stable hand. I've de-capped an old Pentium 4 before and it took a few hours to carefully remove the IHS.

My suggestion would be to go out and buy a few older processors like Pentium 4 for $10, and practice on it first.
 


It's an easy fix for Intel and for the adventurous, just switch the paste with better paste or with the originally used fluxless solder and the problem is solved. I did not deny the problem I corrected you about what the problem is. What the reason is does matter because you CAN fix it if you want to. Only ignorant people were waiting to see Ivy Bridge as being truly faster than Sandy Bridge at stock. It was reasonable to expect it to be better at overclocking and it is if you switch out Intel's paste for high quality paste. Granted, you should not have to do this to get the proper performance boost and that you do have to do it is ridiculous, but don't pretend that it's not an option.
 

belardo

Splendid
Nov 23, 2008
3,540
2
22,795
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?
 

belardo

Splendid
Nov 23, 2008
3,540
2
22,795
[citation][nom]blazorthon[/nom]It was reasonable to expect it to be better at overclocking and it is if you switch out Intel's paste for high quality paste. Granted, you should not have to do this to get the proper performance boost and that you do have to do it is ridiculous, but don't pretend that it's not an option.[/citation]
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.
 
Status
Not open for further replies.