Ivy Bridge Temperatures Could Be Linked To TIM Inside Integrated Heatspreader: Report

[Krono5]

W/m-k = watts per meter kelvin = the SI unit.

1 W/mK = 1 W/m.C

only difference is that is converted into 0.0001 of an inch from a meter.

To convert that to the same units, simply mutiply/divide by 0.0000254 (which is 0.001 inches in m or vice versa)
= 8.89 W/m-k.

 

[Freddie1980]

It must have been aluminium then, liquid pro is said to be unsuitable for aluminium IHS for that reason.

Is the IHS on IB copper?

 

[-Mike-]

So the AS5 is actually better than the IC diamond?

 

[Krono5]

So the AS5 is actually better than the IC diamond?

For thermal conductivity. However there's other factors such as how close the compound can get atomically to each surface which can also affect heat transfer. There's no point having a highly thermal conductive cmpound that cannot make proper contact with the surfaces so is only transfering heat at it's full capacity on 50% of the surface area.

 

[Krono5]

Is the IHS on IB copper?

nickel plated copper yes.

 

[chippy08]

So the AS5 is actually better than the IC diamond?

I found Thermalright Chill Factor 3 to be better than AS5 on my Noctua NH-D14

 

[jonny512379]

just an idea (probably a silly one!)

Would it not be possible to solder the IHS back on then?
what temp can the CPU take?
possible method: Im thinking if a small blob of solder could be part melted straight onto the core with a soldering iron (enough to melt and cover it), replace the IHS and clamp it tight with something, then heat the IHS with a blow torch (gently!). as long as very carefull would the CPU stand it, as chips can usualy take the temperature of melted solder.

I dont have an IB to test this with (and probably wouldn't if i did) its just an idea for you crazy people that dont mind taking a Stanley knife to your CPU!
Im finding the thread interesting and wondering if INTEL have done this deliberately, so they can be guaranteed of getting lower temps on their next tick/tock

 

[moogle]

It could probably be done in the reverse way people used to remove soldered IHS'

As in have the CPU and IHS upside down with the IHS in a clamp with solder in the centre and a flame/lighter on the underside heating and keeping the solder liquid.

Then from above you place the CPU onto it and remove the flame/lighter from the IHS and let it set.

Probably not the best way and needs practice to get the right amount of solder and alignment but it could be done I suppose.

 

[Jokester]

Is the IHS on IB copper?

Nickel plated copper.

 

[Mik3]

I just did mine, its really simple and took less than a hour.

Liquid Pro applied:

4.4Ghz, 1.160V, MX4, Titan fenrir cooler.

 

[RavenXXX2]

Impressive results, GG.

 

[Sasso]

snip

that's a nice drop in temps

although...can;t see to clearly from the pics but looks a bit odd with the liquid pro... i mean this stuff (form what i've seen) does the thinnest layer ever.. does it make good contact? (probably silly Q as you would have higher temps otherwise xD ) but it seems like the TIM does stick up a bit from the die? just looks a bit odd xD

also mx4... so did you use liquid pro on the inside and mx4 on the outside? i'm a bit confused xD

either way.... good job

 

[Krono5]

That seems pretty conclusive then, Intel cheaped out on the TIM instead of solder (which liquid pro is basically a liquid version of as far as I can tell).

Good work Mik3. I guess this would also help even more at higher overclocks/temps so the difference may be more like 15C at 4.6Ghz and 20C at 4.8Ghz making way for SB level OC's.

 

[Craterloads]

For **** sakes someone attempt 5ghz with either the heatspreader off or with replacement tim!

 

[hanluc]

I just did mine, its really simple and took less than a hour.

Good results Mike, not sure from the pic but it looks likes you've coated both the die and the underside of the IHS with lquid pro, you just need it just on the die, apply the tiniest blob possible and then it will be too much, spread a real thin layer with a cotton bud. There's an instruction youtube video embedded here where they sell it here.

That seems pretty conclusive then, Intel cheaped out on the TIM instead of solder (which liquid pro is basically a liquid version of as far as I can tell).

Good work Mik3. I guess this would also help even more at higher overclocks/temps so the difference may be more like 15C at 4.6Ghz and 20C at 4.8Ghz making way for SB level OC's.

Here's mine at 4.8ghz,

though I'm currently running it at 4.7ghz around 1.35v as I do a bit of crunching and felt the performance per power draw wasn't worth it, temps are no longer an issue though

 

[hanluc]

For **** sakes someone attempt 5ghz with either the heatspreader off or with replacement tim!

I reckon I would need around 1.5v to achieve that so I not going to bother, I would have loved to though being as you asked so nice.

 

[Ollie1132]

Hm give 1.4V a go and see if you can be stable at 4.9? (you are water cooling right?)

 

[hanluc]

Hm give 1.4V a go and see if you can be stable at 4.9? (you are water cooling right?)

No point, I need 1.4 for 4.8ghz (see the screenshot), and yes it's water cooled.

 

[Craterloads]

I reckon I would need around 1.5v to achieve that so I not going to bother, I would have loved to though being as you asked so nice.

lol sorry

So you think 5Ghz is a no go then?

1.5v is way to much for 22nm to handle, A tad dissapointing, was hoping for 5Ghz+ once the heatspreader was off.

 

[hanluc]

lol sorry

So you think 5Ghz is a no go then?

1.5v is way to much for 22nm to handle, A tad dissapointing, was hoping for 5Ghz+ once the heatspreader was off.

Maybe it's just my chip but I seem to need around 1.3v for 4.6Ghz, 1.35v for 4.7Ghz and 1.4v for 4.8Ghz