Official ICD7 Thread

I'm eventually going to fill a bucket with the ICD and then just drop the laptop straight in.

 

Comes in handy syringes, just drill a small hole in the case and keep injecting untill some starts squezing out from between the keys.............

 

At least that might cool the keys over the HD somewhat.

 

I have had GPU heat problems with my P-7811FX since day one, with temperatures always over 100C while gaming. I just finished taking it completely apart and applying ICD7. I immediately started up 3D Mark and my max temperature was 64C without a cooler. Granted, there was a good chunk of dust in my GPU fan, but I've been getting those horrible temperatures from day one. This product is amazing.

EDIT: I don't mean to sound like a fanboi, but this saved my laptop from a long flight down the stairs.

 

I just put ICD on my old XT5000T, running prime 95 on both cores with a TL64 @ 2.2GHz it would run with AC-5 @ 93c. Now with prime 95 running both cores I only see 82c.

The AC-5 install was over 1 year old, when pulled apart you could see where a huge air pocket developed from what looks like boiling out. ICD should not have this issue.............

 

Sounds promising, will give it a whirl.

 

anybody try ICD 7 to eliminate thermal pads from xps 1210 ? I'm really interested to see the results before I try.

 

kluclock,

You didn't happen to document the procedure you used for getting to the GPU? I haven't seen a guide for doing this, only for the CPU but my search skills may be to blame.

Thanks

 

look at the top of this very subforum, you will see a 7805 disassembly guide.

Though i suppose i should change the threads name to Gateway FX disassembly guide because the process is the exact same for all of the models

 

I'm sure it will work much better than a thermal pad. Just have to be sure that if there is a gap that you use enough ICD to fill the gap so it has complete contact on both surfaces.

IIRC people are using ICD on the VRAM and everything. I'll get to that some day..

I'm testing my i7 right now with the ICD, before/after results will be posted in a few minutes

 

how much quantity do you think is roughly needed to fill the gap? Is the 1.5 gm tube sufficient or do you suggest the 4.5 gm one ?

 

ICD7 is plenty if you're just applying to a single CPU. If you're applying to a CPU and other components, you definitely want the ICD24 to be safe and sure that you have enough compound.

 

This was done on an i7 920 with an MSI X58 Pro board, custom liquid cooler
Temps dropped on all 4 cores after applying the ICD, as well as the 3 other sensors on the board ALSO dropped temps. ICD was ONLY applied to the CPU, this test was done with ALL OTHER FACTORS REMAINING THE SAME (Including ambient temp). The changes in before and after were strictly from the ICD.

Before - Hotter



After - Cooler


Even though the After test showed a max temp of 60°C, this was only a peak for a second, unlike the max temp in the Before test. During the stress test the CPU stayed at 58°C-59°C, so comparing the max temp of the hottest core in the before and after tests, there was a consistent drop in temperatures of 4°C (7°F).

These are great results for simply changing the thermal paste! Thank you ICDiamond!

 

Thanks people for taking the time to test IC Diamond. Here is a summary of results to date.

Out of the group so far 19 saw an improvement in temps, 2 no change in before and after temps. and 1 saw an increase of + 2C

 

I have gone and replaced all the thermal pads in my dual 4870 cards in the W90 except for the single spot where there was a double thermal pad.

I have been running like this for a couple of weeks now and have suffered no ill effects. Most unfortunately there is no temp monitor for the voltage handling parts or the ram chips on the cards in my unit but I can very confidently say that ICD is a valid substitute for a thermal pad.

I have cross posted this information in the dell thread about the copper shim modification. It would be both easier and cheaper to use ICD to fill that gap in there cooling mod than to mess with copper shims.

Im sold, I'll continue to use this product from here forward instead of AS5 since it has shown to me that it works at least as well as AS5 but does not have the conductive properties making it safer to use in all situations. Its just a major pain to squeeze out of the tube my hand was about to fall off by the time I applied that whole tube of ICD24.

@ IC Diamond

I still want to find a way to perform a static test on this compound, one where the modular speed of a laptops cooling fans does not come into play. I'll try to think of a way to set that up in my free time without having to spend any money for a test setup.

I'll need a heat source that outputs a consistent amount of heat, a heat sink, and a fan with a fixed speed while I monitor the temp of the heat source with both ICD and AS5.

 

I got my tube of ICD7 earlier this week and got a chance to try it out on my 7805u FX

Stock the rmal compound
IDLE CPU - 39C/40C
LOAD CPU - 63C/65C

IDLE GPU - 54C
LOAD GPU - 70C

ICD7
IDLE CPU - 34C/35C
LOAD CPU - 52C/53C

IDLE GPU - 50C
LOAD GPU - 65C

 

nice drop in temps their it seems u applied it correctly......happy to see that it worked congrats.....

 

OK so what is the main differences in ICD7 and ICD24? Is 24 really worth 20 dollars more than the ICD7?

I'm looking at temps of (not undervolting currently) 37-39 degrees idle and after an hour on Othros 68degrees max (normally no higher than 64ish) with my x9000 running at 3.0gHz using AS ceremique. Which i prefer due to nonconductivity (guarenteed safe)

Would it actually be worth it to go through all of the trouble to take my entire rig back apart just to apply this stuff? I cant see it bringing my temps down that much further. But if you say it will make it so i wont break 60 degrees then i might try it.

Though i might try it on my GPU (as i've never applied anything there)

Any suggestions?

 

@ Kamin its the same compound. The numeric designation is just referring to the size you get.

 

I'm contemplating applying ICD7 but am somewhat put off about the issue of thermal pad & northbridge chipset.

If i do manage to damage the thermal pad (some of it gets stuck etc.), will the ICD7 be thick enough to cover the gap or will i have to get a replacement thermal pad?

Am i also right in saying that the correct method of application would be to apply a generous globule of ICD7 straight onto the CPU/GPU and then screw the heatsink straight on top to ensure even spreading of compound? Or, to spread the ICD7 with a plastic card first prior to heatsink re-attachment?

thanks in advance for any advice given.

 

Drop it in a pot of very hot water for 5-10 minutes.

I've done this, as have several others. and if you check out the Innovative Cooling website they have quite a bout of testing results posted. I know andrew and I have talked extensively about different testing methodology, and different ways top torture thermal compounds.

I have professional testing equipment that I still use, but we can come up with some ideas here for further testing......start thinking of new ways to stress stuff and let me know, I'm open to suggestions.

http://forums.hardwarelogic.com/f213/thermal-compound-18312.html#post153380

http://www.youtube.com/watch?v=oz5e-vFHQ4M&feature=channel_page

http://www.youtube.com/watch?v=gwwdWWVdIhU&feature=channel
__________________

 

thx capper, first link was most helpful (vids too).

 

So I get an epiphany here of why things work as they do on laptops after reviewing Cappers excellent how to do it guides.

Part of the reason there are increased delta Temps on the laptops versus desktop systems is the contact area. I use two simulated cpu dies in my day to day testing, a copper 1cm square die and 1" inch square copper die.

I have been focused on large contact IHS applications for so long I forgot about the other half of the market.:SLEEP:

I use the 1 cm die when I need a higher resolution comparison That's because for equal watts and due to the smaller surface area I get a larger heat flux, about/ approx double the the 1" die for delta temps.

So small die larger delta between compounds.

For the bulk thermal conductivity we use an ASTM test set up and for TIM comparison we use a constant gap.

But this does not tell the whole story. One compound may have a low thermal bulk conductivity but smaller particle size in it's bulk loading and under heat sink pressure it resolves down to the thickness of the particle size. The other half of of the equation is the bond line thickness or BLT a thinner joint will have better thermal conductivity.

So with the GPU's and the large gap IC diamond is like the ASTM, the high thermal conductivity trumps the compounds that rely on small particle size to get a thin BLT and not lean so heavily on bulk conductivity to obtain performance....

Basically with IC Diamond the design points were a high bulk thermal conductivity in combination with the optimal particle size.

------------------------------------------------------

About testing- I would epoxy a thermocouple to the side of the die. Then use the old AMD method of drilling a hole in the sink and epoxying the thermocouple so it is positioned directly or near the center of the die(Have to work around the heat pipe somehow) flush with the base of the sink. This will measure the temperature drop across the compound.

I prefer thermocouples as an on system resolution is +/- 1 C CPU plus you need a high quality ambient temp reading, easy to fluctuate a couple of degrees on that account the built in margin of error falls with in the total performance range of the compounds tested. This is why I Avg. the results from the forum testing. I estimated wattage's on a sample group and corrected it to overall results and surprisingly the averaged results mirror what I get in the lab within a couple of 1/10ths of a degree when get above a sample size of around 30 + or so

On a one inch synthetic die I see a delta of about 1C and on the smaller die about 2c both @100 W with IC Diamond

The weakness in real world testing from my point of view for absolute numbers is the platform only forms a partial picture.

For example total watts through the CPU is an estimate and can fluctuate and be different cpu to CPU/MB.

Notebooks have notoriously poor cooling so some portion of those watts are pumped back into the board so most of the results are fine comparatively when done on the same machine but between models and vendors things start to fog a little ( Note the forum results Chart)

In what I do I have to be able to get repeatability - I can pick up a reference sink I tested 5 years ago, test it today and get the same number today as I recorded 5 yrs ago.

 

Help lol... I am trying to buy some ICD 7 in the 7 carat size... but the main site is sold out, and the only seller on ebay is in europe and is overinflating the price!

 

http://cgi.ebay.co.uk/Innovation-Co...3.l1177&_trkparms=240:1318|301:1|293:1|294:50

Seller is in Canada; bought the 24 carat tube from him....it's in the post right now. Should get to your location a lot quicker in comparison to the UK.

 

Here is one closer to home.

http://cgi.ebay.ca/Innovation-Cooling-7-Diamond-CPU-Thermal-Paste-Grease_W0QQitemZ260425128833QQcmdZViewItemQQptZLH_DefaultDomain_2?hash=item3ca28c1b81&_trksid=p3286.c0.m14&_trkparms=65%3A12|66%3A2|39%3A1|72%3A1215|240%3A1318|301%3A1|293%3A1|294%3A50

I have dealt with this individual before, and its all good.

 

erm, it's the same dude lol.

 

Yes it is. I did not check the first link, noticed it was for eBay in the UK, so posted the Canadian eBay sale.

 

PM me your address - I will send you a tube - Gentleman's agreement post your result

Andrew

IC

 

Hmmm, interesting. Why did you use ICD7 on the northbridge, but OCZ Freeze on the CPU? I currently have some OCZ Freeze, but if ICD7 is better, then I'll have to try some out. Just curious as to why you used two different thermal pastes.

 

ICD7 is better....long term testing has shown this. These thermal compounds with a high liquid content tend to dry out after 4-6 months (You hear people refer to their thermal compounds turning almost to powder, or having gaps). I Do a LOT of testing, and as such am consztantly switching out thermal compounds based on what I need, and to see the differences in performance between different compounds. OCZ Freeze does perform very well, short term as well as the ICD7 and MX-2 and TX-2.......but like those compounds, its liquidity is a problem long term.

 

Resurrecting a stagnant thread probably wont win me many friends but i just got to clarfiy this issue please:-

received my 24 carat tube of ICD7 in the post today. I've read that it would be advisable to "boil" the applicator syringe in water for ease of application. My question is whether this would actually be detrimental to the TIM compound, given that (i presume) the syringe is not waterproof. Do i fully submerse the tube or just let it float on one side?

the last thing I want is water mixing with ICD7 which i then unwittingly apply on my 2 week old X9100 p7805u.

I'll post before/after results, assuming all goes well!

thanks

 

You do not boil the water with the syringe in it. Boil the water, then put the syringe in for a couple of minutes. Leave the cap on the syringe and you should get no water.

 

man, i feel like a complete idiot.....

thanks for the info.

 

Gidday guys

This sounds good, I have just ordered a tube. One question though, I have seen some posts here saying "apply a pea-sized glob", others saying this is too much. How much should I apply to a small core CPU, mine is a T9300 with the heat sink stright on to it, no thermal pad. I want to get it right first time. I am currently using AS5 so it will be interesting to see any difference.

Regards to all.

 

Pea is for a desktop processor which has a die that covers the entire processor... on a laptop you use about the size of a grain of rice

 

That would be the response if you had not asked about it and just forged ahead. There is nothing idiotic about asking questions to clear up uncertainties. Hope things work out okay.

 

You should definitely use more than a rice grain sized gob (and a very good general guideline)....thats the general guidance for your usual thermal paste, one that is liquid based and spreads easily. Personally, I'd recommend twice as much ICD7, replace the heat sink, then remove it to see how it spread. ICD7 is non conductive, and it wont hurt anything if it spreads over the edges and touches the pcb (It would take a lot for it to do this).

Thats the perfect answer, the only dumb question is the one not asked. There are a bunch of really great people here who are willing to help, take advantage of their generosity.

 

You know since I found out that heating the compound in hot water makes it less thick and easy to spread, I wonder about using it as a thermal pad replacement.

You in some cases have a very thick application of it, under high heat is it possible it will lose enough viscosity to melt down and lose its bridging connection?

I ran OCCt on my W90 the other day and my GPU went to 117c

Its possible my cards have an issue running that program as there is rumor that the desktop varients of the 4850 and 4870 also run much hotter than they should with that test. But other users are only like 90c with GTX280's

My real game test was the crysis benchmark looped like 10x with a heavy OC and the cards did not even go to 80c so thats fine with me in real life performance, but none the less the event had me thinking about what happens when the compound is not spread thin like a cpu install and is really thick for a thermal pad replacement and then really high heat hits it.

 

It doesn't lose viscosity, after an initial heat up, it cures and solidifies into a wax-like texture. That was one of the first things I tested before recommending it to others. I've found that while it does "soften up" a little, it really doesnt change dramatically.

Also, as long as there is even pressure I've found that it doesnt bleed out like other liquid based compounds......only when there is little or no pressure applied, and the compound is initially set in a vertical position doers its spread change at all, and even then its minimal.

as to OCCT, itys a great desktop stress tester.....I've found that no one program is great for every environment, and I run several different apps to test temperatures and stress, including hardware testing equipment......

 

Thanks for the responses, but we still seem to have a couple of different application options for a T9300, grain of rice size, twice grain of rice size, pea size, what's the verdict. T9300, small exposed core?????????????

Regards to all.

 

well, I applied some ICD7 on my X9100 & northbridge yesterday and I used a GENEROUS globule of the stuff....size of a M&M/Skittle on each. Followed Capper's vids to the letter.

On a side note, Arcticlean thermal remover smells quite nice too....like zesty citrus/oranges. I also found out that my X9100 is an engineering sample too.

To be quite honest, i havent noticed much difference in idle temps and when the X9100 is stressed with OCCT 3.1 and Core Damage (8hours+), TZS0 and TZS1 temps reach a max. of 80C and that's with undervolting @ 1.05V using RMClock + external cooling from 12V overvolted Cryo LX & NB-MA1 USB fan extractor. CPU core temps level out at around 70C.

Perhaps the ICD7 needs time to be "broken in"? I'm kinda disappointed at present!

 

The advantage of ICD 7 over other thermal paste's is in it's short curing time, longetivity, non conductive and viscosity... I would say if you were using OC freeze or AS5 before your results are going to be similiar

 

Putting the X9100 in the 7805 is big mistake, now you have got electric heater inside your laptop and not decent CPU. Such processor produces twice amount of heat than P or T series processor does.

 

I bought the p7805u preconfigured with the X9100 and 640GB RAID0 + Microsoft Office Ultimate 2007 pre-installed (full version) from ebay UK for £1152 (~$1860) which I think is a pretty sweet deal, considering other UK vendors/sellers still price the X9100 (on its own) at a ridiculous £599 (~$960) *cough* Dell UK*cough*

At the time of purchase, i wasnt aware that Gateway bios was locked to prevent OCing past 3GHz nor of the fact that upgrading the CPU on the p7805u is a piece of cake.

Otherwise, I would have opted to get a base unit shipped from the US and bought an OEM T9800/T9900 from China and performed the upgrade myself. The only mods I've had to do to my p7805u is break the RAID0 array for 1 128GB G.Skill Falcon SSD + 320GB WD scoripo HDD and slap on some ICD7 on CPU & northbridge.

Currently considering swapping out standard optical drive for blu-ray but @ 1440x900 res, i know it isn't really worth it but it's good fun to tinker - upgrade will probably set me back approx. £50 (~$80).

 

Well, that explains everything. Yeah, it's a good idea to swap the present drive for Blue-Ray one. However imo the Blue-Ray burner would be better than only reader drive. You can always hook your lappy to the TV set and watch movies in high definition.

Cheers...

 

That is a major exaggeration. The X9100 is MORE than a decent CPU. It does not produce TWICE the amount of heat of a P or T series.

 

Come on, you know what I'm talking about. First of all Hello Moto got this CPU already installed in the machine, but if he could select other I'm sure he would get the T one. He can still swap it. And if you would have a choice in selecting the processor would you pick this X one. If so what's the point in having this X cpu if you can't overclock it beyond 3GHz in P series notebooks. Right?

You have T9900 which runs at 3GHz and produces less heat than the X one. You are right that the X9100 doesn't produce twice amount of heat, but still much more than the X one in my opinion.

I have the t9600 that runs at 62-65C max temp and I don't need any coolers and/or undervolting. Probably I would need these with the extreme processor.

 

I know what you're talking about, just wanted to point out that it was a major exaggeration.

A 3 GHz Core2 processor is more than a decent CPU, and it doesn't produce twice the heat of a T series or P series if you compare it to near identical speed cpus.

Obviously the other series produce less heat, but they are also more expensive.

He has many options, he can definitely leave his setup the way it is, or he can swap CPUs, or he can even get a diff. laptop with a bios that will allow him to overclock his X9100 and utilize its full capabilities that the other comparable high-end processors don't have, etc.

 

The problem isnt that the X9100 is a bad processor, its a very good processor. The problem is that its a wasted processor in the 7811 FX because its an EE processor with an unlocked multi, a feature not supported by our notebooks.

As to the thermal issue, its a 45W processor, which does generate more heat than the 25 and 35W T and P series processors...... and puts a lot of strain on the very basic HSF found on the 7811 FX (Which was built around a 25W P series processor. Then, take into consideration that its an engineering sample....which while great when you are talking desktop processors, is not ideal for notebook processors, because EE processors are for testing and not held to the same quality control as retail or OEM processors.

Think about a stock Intel HSF on your desktop system with an Extreme Edition processor, and then using great thermal compound. Expecting it to provide excellent results is not very realistic, because that basic HSF cant displace heat fast enough for the thermal compound to make any difference. If your temps are after 8+ hours of stress testing, thats honestly not bad..... considering you are putting your notebook under some ungodly stress. I understand its a gaming notebook....but its a notebook none the less, and 8+ hours straight of 100% load....using a 45W processor with a hsf built for a 25W processor, and being disappointed in the thermal compound is silly.

The heat sink is the limiting factor here, as many using the X9100 will point out, it runs extremely warm, and pushes the heat sink in the 7811 FX to its edge. I've talked to Gateway, and Acer, and they strongly recommend against using a 45W processor in this chassis, and even question using a 35W processor in this chassis.