• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

FRONTPAGE Intel i7 3770K - Ivy Bridge - CPU Review

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
Ok, so who has the....intestinal fortitude.... to delid a working IB, and put the solder on the die and relid to really test this theory out?

I believe it to be true, it is logical, however Im curious to know exactly how much difference it makes.
 
Ok, so who has the....intestinal fortitude.... to delid a working IB, and put the solder on the die and relid to really test this theory out?

I believe it to be true, it is logical, however Im curious to know exactly how much difference it makes.

I'm sure you could get one of those sponsored guys who gets a tray of CPUs to bin to do it :p
 
Remember, this isn't ordinary solder - it's a special fluxless blend made by and for Intel. The temperature to melt is (most likely) far lower than your typical lead-based or RoHS solder. If any action is to be taken, it should probably be de-lidding to directly cool the die. Trying to solder the IHS back on is soldering-iron-assisted-CPU-suicide.
 
Just take the bloody lid off and file the height of it down so it fits tightly on the DIE cap...
 
Was curious after reading that Anand article about undervolting where they did 4.5 at 1.1v so tested for myself. More like 4.3 for me. Average temp after an hour of load = 46C.

Lundervolt1.png
 
Was curious after reading that Anand article about undervolting where they did 4.5 at 1.1v so tested for myself. More like 4.3 for me. Average temp after an hour of load = 46C.

Lundervolt1.png

Not bad... What motherboard? Chip is always a factor (VID etc..), but the voltage regulation also makes a difference.
 
"The hottest core measurements are 46% higher with Ivy Bridge."

Arghh....No they are not. The Ivy Bridge core maxes out 7.8% higher in temperature than Sandy Bridge. You CAN NOT USE CELSIUS (or Fahrenheit) for percentage comparisons in temperature. You MUST use Kelvin. Anything else is flat out wrong. I am seeing it all over the place with these IVB reviews and forum posts.

Would you say when it's 1C and it rises to 2C that the temperature doubled? What about 33.8F to 35.6F...is that a 5.3% increase in temperature? Because, well...those are the exact same temperatures. (1C to 2C is 33.8 to 35.6 in F). Of course, the real answer is 1C to 2C is a 0.3% increase. (274.15K to 275.15K).

How about 0C to 4C? Wait, that's an INFINITE increase in temperature right? What about -3C to 6C?

I'd expect better from a technical article.

Although its rarely helpful just to post "^This", I'm really tempted right now. I'm glad there's someone else out there who understands what % of temperatures actually mean (or rather don't). It's like when someone calls Internet Explorer "The Internet" or calls the monitor "the computer" and the box thing under the table "the hard drive". It's obvious what they're trying to say but the wrong thinking behind it just makes the teeth grind of everyone else.

Other than that, really good article! Thanks.

EDIT: Whoops. I see I'm not in the majority on this. :eek: Well "bah!" I say! We shouldn't leave bad maths lying around where kids might read and accept it! Wont someone please think of the children!
 
Last edited:
Although its rarely helpful just to post "^This", I'm really tempted right now. I'm glad there's someone else out there who understands what % of temperatures actually mean (or rather don't). It's like when someone calls Internet Explorer "The Internet" or calls the monitor "the computer" and the box thing under the table "the hard drive". It's obvious what they're trying to say but the wrong thinking behind it just makes the teeth grind of everyone else.

Other than that, really good article! Thanks.

EDIT: Whoops. I see I'm not in the majority on this. :eek: Well "bah!" I say! We shouldn't leave bad maths lying around where kids might read and accept it! Wont someone please think of the children!

When discussing cpu cooling by air or water, it is delta T that is relevant. The starting point is defined as AMBIENT temps when discussing cpu cooling, not 0 kelvin, delta T is core temp - ambient. Without ambient, cpu temps are not defined. Not only do you need to be mathematically correct, you also need to be technically correct for the field being discussed. It is not technically acceptable to define a delta percentage based on 0 degrees kelvin when discussing delta core-ambient, unless 0 kelvin is your ambient, in which case you have larger concerns. I have less of an issue with Hokies terminology than yours.
 
Just to reiterate - the chip is not producing 46% more heat, thus the Kelvin thing is not useful regardless. It is likely producing a very similar amount or even less heat actually. The issue is core temperatures, not heat. We also have a likely reason for the temperature issue.

The CPU itself should be running a bit cooler as far as actual heat production goes. It is going to run at a higher temperature though due to the new transistor design, the increased power density and the change from solder to TIM under the IHS. Temperature was my point, and remains my point.

If rge is correct and I should have used delta, things change dramatically for the worse. Unfortunately I did not measure the ambient temperature at the time, but in that room (except in the dog days of summer) it is within two degrees of 23°C, most of the time right on it. So, we can extrapolate from there.

Delta from ambient to hottest core
2600K: 33°C (56°-23°C)
3770K: 59°C (82°-23°C)
All +/- 2°C

Thus, the percentage increase of the delta is actually much worse than using zero as the starting point. Best case (Sandy's measurement being +2°C and Ivy's being -2°C), you're looking at 62.9%. Worst case in that deviation (Sandy being -2°C and Ivy being +2°C) would be 96.8%. Right in the middle is 78.9%, much worse than starting from zero.
 
Historically questioning: does Intel bond the CPU and IHS differently on overclockable vs. non-OC chips?

I do not believe so, intel chips have been overclockable for quite some time, it wasn't until SB that they had the different 'unlocked' vs 'locked' chips and with the small variation in Bclk it made it so that there actually was a chip better for OCing [minus the extreme editions in years past]. There have been different models (I believe it was said that C2D e4*** chips used TIM while e8*** [and maybe 6***?] series used Solder.
 
Looking through the most obvious segments have not yet found any info on what the DIE to HIS materials are, no mention of solder or thermal compound, nothing at all.
 
Back