Reason Behind Ivy Bridge's High Temperatures - TIM Used Instead of Solder

Status
Not open for further replies.
R

RoBoGhOsT

Perambulating
Galvanizer
Reason Behind Ivy Bridge's High Temperatures - TIM Used Instead of Solder

The temperatures of Ivy Bridge CPUs being higher than Sandy Bridge is noted in most reviews and tests . Ivy Bridge CPUs run upto 20°C hotter compared to Sandy Bridge chips when overclocked.

coretemp2700.jpg
coretemp3770.jpg


While many people think that the reason behind this is power density being greater on Ivy Bridge , and some say that Intel is having problems with tri-gate / 22nm tech .

Its right that power density is higher , but that doesn't account for the temperatures being 20°C higher .

According to overclockers.com - the reason behind the high temperatures lies under the IHS of the chips .

ocivydie2640x291.jpg


Intel is using TIM paste between the CPU die and Integrated Heat Spreader , instead of fluxless solder used in previous gen chips . The heat conductivity of TIM is in range of 5W/mK whereas solder has heat conductivity in range of 80 W/mK . This is the reason which decreases the effective heat transfer between CPU die and IHS .

Why Intel is using TIM instead of solder is not sure , or if it will be seen only in Engineering samples or final units as well .

However, there is an advantage in this also , if the IHS of these chips is removed then there is less much less risk of damaging the die , which could help extreme overclockers to get better temperatures for improved overclocking.

Source
 
This is extremely informative and I must say ...... some troubling news too.

I was checking out the source http://www.overclockers.com/ivy-bridge-temperatures and the discussion's therein are also worrisome. Let me quote two of them

from the author

I didn't want to wander too far into the weeds within the article itself, but I want to mention as well about delidding. Many people know it can be dangerous and easy to kill a chip that way. However people often forget a challenge with making contact with the die - if you delid, you may need to modify the socket/socket retention clip to ensure the base of your cooler can contact the die, as without the IHS the chip will ride lower on the motherboard. Just a note for anyone who may follow up with further testing on this.
Click to expand...

Now what is this delidding - is not clear to me ?

and further ..............

#hokiealumnus

It's also important to note that those who air cool their CPU with direct-contact heatpipe coolers do NOT want to delid their CPU. You'll have heatpipes that don't touch the heat-producing die surface and be worse off than when you started.
Click to expand...

So does this mean that 2ns party - direct-contact heatpipe cooler's are useless ?

Lastly , are the above cooler's same as "air-coolers"
 
terence_fdes said:
This is extremely informative and I must say ...... some troubling news too.

I was checking out the source http://www.overclock...ge-temperatures and the discussion's therein are also worrisome. Let me quote two of them

from the author

Now what is this delidding - is not clear to me ?

and further ..............

#hokiealumnus

So does this mean that 2ns party - direct-contact heatpipe cooler's are useless ?

Lastly , are the above cooler's same as "air-coolers"
Click to expand...

Delidding - see image in first post.

dellidding is removing the chip metal top. the retention brackets are made with exact dimensions of the metal top. cpu coolers are designed to touch the top of this metal.

so if you remove the top there will be a small gap between the die and the cooler. cooler will only be effective if its in contact with the die in this case. so unless you do something (like design your own cooler) to get cooler and die come into contact with each other you shouldn't remove the metal cover.
 
terence_fdes said:
So does this mean that 2ns party - direct-contact heatpipe cooler's are useless ?

Lastly , are the above cooler's same as "air-coolers"
Click to expand...

Now what is this delidding - is not clear to me ?
Well this is what delidding a processor means Sire -- you remove the shiny metallic top you see on the processor, you do know that the actual chip is the green wafer. The shiny metal cap on top is a Integrated Heat-spreader, for a breakdown of the processor --

Intel-Core-i7-980X_Gulftown-Processor_Exploded.jpg
  1. The metal cap with Intel emblazoned on it is the IHS of the processor.
  2. The pinkie stuff is the actual processor.
  3. The black-green at the bottom is the wafer and the pin area for the processor, basically what allows it to be stuck in a particular socket.

So delidding will be the process to remove the IHS and revealing the actual processor.

For more on the topic --> http://forums.extremeoverclocking.com/showthread.php?t=351971 / http://www.overclock.net/t/372102/delidding-cpu-on-new-boards.

Now all heat-sink fan packages in OEM production and third-party channels are made to contact with the IHS of the processor, then carry away the heat, so if the IHS is removed and the piddling processor is 'naked' the problem is two fold --
  • how to depress the contact area low enough to make contact with the cooler heat-pipes / surface.
  • How not to end up short-circuiting the processor in this method.

Hope this clears all your doubts #terrence_fdes. Cheers!!
 
6pack said:
Delidding - see image in first post.

dellidding is removing the chip metal top.
Click to expand...

noob Q again - But Pray why would one want to remove "chip metal top or the IHS" ? What purpose does it serve ?

Thanks #Alpha17 - I posted my querry before I saw your reply.

I also check out the links that you posted. I can only conclude that such an excercise (of delidding) - is fraught with horrendous results. Has anyone at TE ever tried this out
bye22.gif
 
terence_fdes said:
noob Q again - But Pray why would one want to "chip metal top" ? What purpose does it serve ?
Click to expand...

Well it reveals the actual processor in itself.

Most early processors upto Pentium 3 did not have the IHS, you could see the actual processor. The only advantage I can think off is modding a cooler / processor and try to improve the thermal efficiency of the cooler used.

How the above can be practically achieved is anyone's guess? I mean you de-lid the thing, I do not know what next. Please note all GPU's chips are also without a IHS [if I am not wrong] and they have direct contact heat-pipes to cool them, and you can try that approach on the processor as well.

Yes see here, no IHS this is how an actual processor is in its primeval state of production --

GPU_chip_big.jpg


Now I slid of the answer and am confused myself, Cheers!!
 
what purpose? some do it just because they can
<


in reality, hardly any one would do this. only the over enthusiast over clockers who want to go that extra mile with liquid nitrogen cooling etc, would take the pains.

removing the lid is not easy i guess. that black goop like sticky thing around the die is quite tough to pry apart.
 
6pack said:
what purpose? some do it just because they can
tongu23e.gif


removing the lid is not easy i guess. that black goop like sticky thing around the die is quite tough to pry apart.
Click to expand...

That is why I mentioned I slid of the answer, I do not know any practical purposes for this act of de-lidding the processor.

Yeah I concurred that from a forum post on Over-clockers.forum, where a guy was considering something on the lines of a blow torch to soften the solder inside a Phenom IIx4 processor. And I was like
scared14.gif
.
 
Those using LN[sub]2[/sub] would be a small set of people. Instead, Intel should have made a special version for those and for the rest of us, with the same TIM used in SNB.
 
Great #[member='vivek.krishnan'] & other's ......... As I see it now delidding is (How to beat the Summer Heat in hot & humid Mumbai) like peeling off the scalp off my fore-head and placing the best ice-pack on it to stay cool and perform better
happy55.gif


Ok, jokes aside - getting back to "High Temperatures of Ivy-bridge"

What solutions do we have ?

- Is it then advisable to stick to the stock-cooler [If no OC'ing} is done - but keeping the system on 24x7 for say a few days at a stretch or Is Sandy a better bet in such a case ?

- Would installing a Noctua (though best among the air-coolers) EVEN if one does not want to OC the IB - Say go for the 3770 (not the K version) - be a better option than the stock cooler ?

- I raise these issues because I would want the IB proc to be rock-solid & stable to last for at least 5-7 years.

Is it too early to pose these questions ?
 
terence_fdes said:
What solutions do we have ?

- Is it then advisable to stick to the stock-cooler [If no OC'ing} is done - but keeping the system on 24x7 for say a few days at a stretch or Is Sandy a better bet in such a case ?

- Would installing a Noctua (though best among the air-coolers) EVEN if one does not want to OC the IB - Say go for the 3770 (not the K version) - be a better option than the stock cooler ?

- I raise these issues because I would want the IB proc to be rock-solid & stable to last for at least 5-7 years.

Is it too early to pose these questions ?
Click to expand...

Even if you are not over-clocking but want to keep your system on for great stretches of time Sire, I recommend that you put a cooler of atleast the Cooler Master Hyper 212 EVO's calibre.

If you don't mind I would recommend the Noctua NH-U12P SE2. Reasons are multitude and I will not mind repeating them --
  • Better overall quality of the package, comes with two really good fanns.
  • The cooler is compatible right of the block on everything from the Intel LGA 775 --> Intel LGA 2011 [even ASUS's special socket-X], same is the story for the AMD line-up.
  • The Secu-Firmâ„¢ 2 is a solid kit of components that makes putting on the cooler really easy [although a longer process] and it doesn't feel fragile OR flimsy, it puts the cooler down and stays down.
  • Apart from the above a super extended 6-year replacement warranty.
  • A really quick service system from Austria, they actually shipped me a AM3 socket adapter when I asked for one. All you require is purchase proof.

Really I cannot comment further how much better Noctua is, it might not be the top brand. But it does treat you like a true customer.

Apart from the above Sire, I do think it is premature to discuss about cooling Ivy-Bridge if Intel improves its package in the next stepping / batch of processors you might not face much heating problems. But just to be on the safer side and knowing about its history of under-performing stock HSF, it might just be the right idea to invest in a Cooler Master Hyper 212 EVO atleast.
 
#[member='terence_fdes'] AFAIK the stock cooler is useless even in SNB overclocking. You will need to go for an aftermarket cooler anyhow.

Also, seeing that the performance vs temp of the current set of chips is bad, Intel might go back to fluxless solder for the next batch/stepping. Plus overclocking is not as good it should have been. Unless you need the CPU now, would suggest to wait for a month to see whether Intel plans to change anything in the equation.

As for keeping the machine on for 24x7, unless its a rare case, say once a month, max, would suggest to get a low power rig unless its unavoidable. Ensure that there is god cooling.
 
[font=helvetica, arial, sans-serif]#[/font]terence_fdes

sire, even at voltages like 1.5v, no one has yet seen ANY degradation on the i7 920's. mind you most of them run 24/7 for over 4 years now. the effects of voltage degradation on chips is a very less understood topic of and unless you specifically set out to kill them, todays chips are practically bullet proof and unless you specifically set out to do so, its quite hard to kill them
<
 
cyberwarfare said:
[font=helvetica, arial, sans-serif]#[/font]terence_fdes

sire, even at voltages like 1.5v, no one has yet seen ANY degradation on the i7 920's. mind you most of them run 24/7 for over 4 years now. the effects of voltage degradation on chips is a very less understood topic of and unless you specifically set out to kill them, todays chips are practically bullet proof and unless you specifically set out to do so, its quite hard to kill them
happy19.gif
Click to expand...

I agree that the damage might be negligible, but still nobody likes to willingly damage his own equipment / peripherals, eh... Sire. I mean I had absolutely no justification to jump to a Noctua cooler, the stock HSF was going beyond the call of duty and would not have minded if I would've jumped the processor ~50 -->100MHz over stock without pushing the voltage. The only reason I cited was because it was being exceptionally whiny not the best reason, but the upgrade has been justified as now I can sleep in the study room where my PC is generally torrenting through the night [the few LED fans not withstanding].

terrence_fdes is in a similar quagmire, the stock cooler just migt cut it if he doesn't over-clock but it isn't reassuring to see your temperatures gradually rise. I can vouch for this as I saw Core i5 2400 regularly cross 80°C when rendering [in Autodesk MAYA] after re-TIM and reseating the HSF, the temperatures dropped ~70°C -->75°C. I mean anything can happen, for the count they were using a NZXT Tempest EVO cabinet.

My friend undertaking a similar operation on his AMD Phenom IIx6 1090T with the same cabinet [NZXT Tempest EVO] and on stock AMD HSF wasn't crossing ~65°C. This case clearly shows that the Intel stock HSF is pure bunkum vis-a-vis AMD counterparts.
 
ALPHA17 said:
I agree that the damage might be negligible, but still nobody likes to willingly damage his own equipment / peripherals, eh... Sire. I mean I had absolutely no justification to jump to a Noctua cooler, the stock HSF was going beyond the call of duty and would not have minded if I would've jumped the processor ~50 -->100MHz over stock without pushing the voltage. The only reason I cited was because it was being exceptionally whiny not the best reason, but the upgrade has been justified as now I can sleep in the study room where my PC is generally torrenting through the night [the few LED fans not withstanding].

terrence_fdes is in a similar quagmire, the stock cooler just migt cut it if he doesn't over-clock but it isn't reassuring to see your temperatures gradually rise. I can vouch for this as I saw Core i5 2400 regularly cross 80°C when rendering [in Autodesk MAYA] after re-TIM and reseating the HSF, the temperatures dropped ~70°C -->75°C. I mean anything can happen, for the count they were using a NZXT Tempest EVO cabinet.

My friend undertaking a similar operation on his AMD Phenom IIx6 1090T with the same cabinet [NZXT Tempest EVO] and on stock AMD HSF wasn't crossing ~65°C. This case clearly shows that the Intel stock HSF is pure bunkum vis-a-vis AMD counterparts.
Click to expand...
.

I completely understand what you are saying. There are a few threads on XS and oCN where chips have been put through hell and results recorded. But yes we all love to keep our chips cool. Even I, hate to run anything over 65c just coz I'm anal about stuff like that O:-)

Sent from my GT-N7000 using Tapatalk 2
 
ALPHA17 said:
I agree that the damage might be negligible, but still nobody likes to willingly damage his own equipment / peripherals, eh... Sire. I mean I had absolutely no justification to jump to a Noctua cooler, the stock HSF was going beyond the call of duty and would not have minded if I would've jumped the processor ~50 -->100MHz over stock without pushing the voltage. The only reason I cited was because it was being exceptionally whiny not the best reason, but the upgrade has been justified as now I can sleep in the study room where my PC is generally torrenting through the night [the few LED fans not withstanding].

terrence_fdes is in a similar quagmire, the stock cooler just migt cut it if he doesn't over-clock but it isn't reassuring to see your temperatures gradually rise. I can vouch for this as I saw Core i5 2400 regularly cross 80°C when rendering [in Autodesk MAYA] after re-TIM and reseating the HSF, the temperatures dropped ~70°C -->75°C. I mean anything can happen, for the count they were using a NZXT Tempest EVO cabinet.

My friend undertaking a similar operation on his AMD Phenom IIx6 1090T with the same cabinet [NZXT Tempest EVO] and on stock AMD HSF wasn't crossing ~65°C. This case clearly shows that the Intel stock HSF is pure bunkum vis-a-vis AMD counterparts.
Click to expand...

This is precisely the point #[member='cyberwarfare'] ...... Why skimp on something that I consider so very essential - a robust & silent cooler (like Noctua) for the IB - OC or Not. Today, anyone who wishes to go for an Ivy Bridge system is definately going to be spending upwards of 50K -70K.

Every weekend my old AMD Athlon system runs for over 36-45 hours ..... torrenting (there is only 1 Fan in the Cabinet!!!). And for this I have to keep my ceiling fan running continuously in our Hall Room where my system is...... adding up to my monthly electrical bills. I would rather invest in a Noctua (Rs 3900/-) ........ and never bother at all (apart from a good Corsair or NZXT Cabinet with a min of 4 fans too).
 
^not true, an IVB system can be had for way less than that amount. Heck you can get a cheap ass barebones system for as little as 30k or even less (excluding monitor, speakers etc).

chips today have a tj max of 105c. this means that they can run at that temp for 3 years 24/7 and not show any issues at all. and incase something does happen, intel will have your back coz they cant deny warranty since you DINT use an after market cooler and stuck with theirs.
 
#[member='terence_fdes'] If for torrenting you keep running it continously, why not go for a pogoplug / seagate go flex net? It will allow you to run it 24x7 then.
 
Status
Not open for further replies.
Top Bottom