[OFFICIAL] Free ICD7 thermal compound for OCF members Results Thread

"Pumpout" low viscosity pastes - (repeat the mantra (quote below) with the Lite Beer Commercial of the 80's in your mind)

"All Fillers - No particles"

A result of this formula is something called pump-out occuring under the very conditions which are at the crux of our discussion. That formula: Heat + Pressure + Time = failure. Of course that xconclusion is forgone given the last value T but we don't want it to occur months after the paste has been applied.

While PC-enthusiats tend to change their CPU's every hour on the hour we are easily taken by pastes which set rapidly giving great temps but deteriorate almost as rapdily. We are immune to such problems simply because by the time the paste would begin to detrerioate we've already changed it out installing the newest core revision.

However we need to think about what makes a great thermal paste. As counterintuitive as it might seem Temperature is not the only attribute which determines success. Of course it's an intregal part of the equation. However; for the company running servers with millions invested in those files, experiencing an overall 5C drop in temps, while sacrificing 18-months of additional covereage, lower temps mean little compared to TIM lifespan. They need long-term stability, consistancy, longevity, and decenbt temps. And for the average PC user who applies a Bill-A-Goop type formula for instant gratification and then a year later wonder why their experiencing random re-booting, BSODs and buggy programs etc. They may suffer most because it never presents itself in a catastrophic failure, but months of frustration trouble shooting what are hardware issues misconstrued as software system errors.


That's the worst case scenario of a "Pyromaniac Ghost in the Machine." What they need is a stable long lasting product. I'm running several new tests for my Thermal Paste Addendum which includes Noctua and IC-Diamond and some new tests. I've just finished testing Tuniq TX-2 under high-heat, high-pressure between the Q6600 and Swiftech Apogee GTX.

EDIT: I almost got confused as to wich paste I applied hehe. But from the next photos I have to get on my PC you'll see the difference betwen IC-Diamond and other pastes. It just doesn't "break down" as easily and the spread pattern look healthy. I am up early today whoo

Below is the paste after the fact, is this pump-out;

IC7 vs. Arctic Silver Ceramique

Hi, I'm new here, but I'd though I'd share my experience with IC Diamond 7. I didn't get the free smaple, but purchased it from Petra's Tech shop after a goodle search directed me to here.

Artic Silver Ceramique vs. IC Diamond 7 Carat

I’m not one that received the free thermal compound from IC, but a quick google search while I was bored at work brought up this forum. It piqued my interest because I ran out of my favorite thermal interface material, Arctic Silver Ceramique (which coincidently, I received free in the past).

I liked Artic Silver Ceramique because of non-conductive, easy to work with material. It also provided a faster curing time than its more famous brother, Arctic Silver 5. Not to mention, it’s usually a bit cheaper than AS5 and can be found at hundreds of vendors. While that's all good, it also showed a slight improvement by an avg of 1C over AS5.

I decided to go with a new TIM for my personal gaming PC (Q6600 B3 @ 3.2GHz). The IC Diamond 7 (IC7) is fairly new, and I could only find it at 3 vendors (Directron.com Petrastechshop.com & Heatsinkfactory.com)

What drew me to IC7 were 2 things, diamonds and thicker bond material. Diamond thermal conductivity rates up better than silver. That’s a very good thing when our goal is transfer heat effectively from the CPU to heatsink. The other is the thickness of bond material. IC7’s site provided a view that thinner grease is susceptible to grease pump out. Mainly, as the material heats up, it expands, and eventually, this action pumps it out to the edges of the heatsink. Due to this pumping action, the amount of TIM between the CPU and HSF becomes a problem of enough TIM to effectively transfer heat and months later, have to be re-applied.

So, with my head filled with IC’s sales pitch, I pulled out my trusty credit card and plucked down $7.50 ($6 + $1.50sh USPS 1st Class) for this diamond TIM. Two days later, I find it in my mailbox. It was the first time I ordered anything from Petrastechshop.com and they shipped out my order the same day.

Opening the small envelope, I find the IC7 TIM. It’s not in any packaging, no fancy blister pack, no plastic sleeve, nothing. That’s fine, anything to cut costs. It arrived safely.

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

Arctic Silver Ceramique:
First up is Arctic Silver Ceramique. I applied this to my HSF/CPU about a month ago (when I upgraded to a TR Bolt-Thru kit for my HSF) so it’s had plenty of time to cure.
It comes in a non-see thru plunger applicator. Apply Ceramique is very easy to deal with. It reminds me of the generic silicon TIM you see everywhere.

IC Diamond 7 Carat:
This comes in a rather large plunger applicator. My first impression was “Wow! That’s a lot!”. Then I decided to peel off the label. Yeah, them my reaction was, “Hrm. That doesn’t seem like a whole lot”.







Reading about how thick this was on the Internet, I was wondering how my experience was going to go. Surprisingly, it wasn’t as thick or as hard to apply as I thought it would be. It was a bit thicker than Ceramique, and did require a little bit more “oomph” to push down the plunger, but if I didn’t have that to compare it to, then I really wouldn’t have even thought about it. (I actually had a more trouble dealing with Cooler Master TIM)

TEST BENCH:
Intel Q6600 (B3) @ 2.4GHz, 3.0GHz, 3.2GHz)
Kingwin RVT-12025 Heatsink w/ 120mm fan (max rpm)
*Note: Using a Thermalright 775 Bolt-Thru Mounting Kit
eVGA 650i motherboard
eVGA 8800GT 512MB Video
Antec 900 case (closed)
AMBIENT ROOM TEMP: 21C
AMBIENT CASE TEMP: 24C

A note about the setup. The Kingwin RVT-12025 (also known as a rebadged Xigmatek HDT-1283) is a unique HSF. It incorporates what’s called a “Heat-pipe Direct Touch” system. Basically, it uses the 3x 8mm heatpipes to come in direct contact with the CPU to transfer the heat rather than the heatpipes buried into a solid copper or aluminum base. This makes applying any TIM a bit different than what manufacturers recommend.

I used the same process for both TIMs tested here. Artic Silver recommends a single line between the cores on the CPU and a thin layer rubbed on the HSF. IC7 recommends a “pea-sized” glob on top of the CPU. I feel neither of these methods work with a HDT design. What I did was apply a thin line on each of the heatpipes.

After applying three lines to the HSF, I found I used up about 1/3 of the available material.





TESTING:

CASE FAN CONFIGURATION:
I’m just a simple end-user. There’s no special set-up, it’s what I use 24/7 as my gaming platform. My system is bottled up in an Antec 900 case and kept with case covers on. Max RPM was set on the HSF fan for both TIMs.





RESULTS:
Ambient Room Temperature was kept at a CONSTANT 21.0C. Load tests were conducted with PRIME95, Small FFTs for about 15 minutes each. Tests were conducted at speed/volts: 2.4GHz/AutoV, 3.0GHz/1.312V, 3.2GHz/1.343V

Artic Silver Ceramique:
2.4GHz - Auto Volts
IDLE: / LOAD:
Core1: 36C/ 57C
Core2: 36C/ 57C
Core3: 36C/ 54C
Core4: 36C/ 54C

3.0GHz – 1.312V
IDLE: / LOAD:
Core1: 37C/ 62C
Core2: 36C/ 59C
Core3: 36C/ 59C
Core4: 37C/ 62C

3.2GHz – 1.343V
IDLE: / LOAD:
Core1: 37C/ 63C
Core2: 36C/ 64C
Core3: 36C/ 64C
Core4: 37C/ 63C


IC Diamond 7 Carat:
2.4GHz – Auto Volts
IDLE: / LOAD:
Core1: 35C/ 54C
Core2: 34C/ 51C
Core3: 35C/ 54C
Core4: 35C/ 51C

3.0GHz – 1.312V
IDLE: / LOAD:
Core1: 36C/ 58C
Core2: 35C/ 55C
Core3: 34C/ 55C
Core4: 36C/ 58C

3.2GHz – 1.343V
IDLE: / LOAD:
Core1: 36C/ 61C
Core2: 36C/ 58C
Core3: 35C/ 58C
Core4: 36C/ 61C

To my surprise, the IC7 trumped Ceramique. Usually, I was expecting a 1C degree difference in either direction between these 2 TIMS. What I got was a –4C difference using IC7 over Ceramique at 3.2GHz under load. That’s an impressive improvement between TIMs. At all speeds under Idle/Load, the IC7 wins out.
IDLE: .75C Average Difference (2.4GHz, 3.0GHz, 3.2GHz) IC7 Wins
LOAD: 3.333C Average Difference (2.4GHz, 3.0GHz, 3.2GHz) IC7 Wins

The graph below represents the Average Temperature of all 4 Cores.


I tried my best to simulate a constant environment condition between the two TIMS, from keeping the same ambient temperatures to the way I applied the tims. The bottom line is that I’m using it as any other end-user would use the product. There was no special test bench setup I used, no tweaks, no special measuring equipment (other than Core-Temp and Prime95).

I like your application method, makes sense. Good stuff on the ambients room temp can fluctuate 2-3C in a couple of hours under HVAC conditions. Makes a big difference in results when your potential error encompasses most of the expected range of results. Great job

Very thorough, nice photos excellent dpcumentation. Your a great tester!

Can you provide a few pic's of the IHS and heatsink base fater removal? I'm very interested to see how these newish direct to heatpipe coolers work.

This is not a criticism towards you or the company, I just wonder if the heatpipes could have been angled and joined at the bottom resulting in an all copper/heat-pipe base?

The interrupted specifc heat capacity and conductivity from copper heat-pipe to aluminum sections in between must produce certain "dead zones" allowing a certain amount of heat to build there.

Anyway my opions aside there's no doubt you've done an excellent job.

Liquid3D said:

This is not a criticism towards you or the company, I just wonder if the heatpipes could have been angled and joined at the bottom resulting in an all copper/heat-pipe base?

Click to expand...

I think it is probably structural, copper is soft to begin with. My bet is there is not a lot wall thickness left after sanding the tubes without the AL supports the whole set up buckles under clamping.

I wonder how it tests on a solid synthetic die vs a CPU/IHS or perhaps rotating it on the IHS, changing the cross section of coverage?

The AL support plates probably are fairly good heat spreaders, transferring the heat with a pressfit on the sides

It looks like the sides of the heatpipes have also been "fattened and flattened" with copper like the bottom, so the AL probably has a good-sized area of contact with the pipes ...

I should have taken a pic of the ASC tim after I pulled off the HSF. I used the same technique (three lines of TIM on the heat-pipes). The cover was uniform and complete (spreading across the copper and AL spacers).

JoeC is right in that the copper is soft. Here's a pic after a quick initial wipe off of the ASC TIM.

As you can see, previous installation of the HSF has dented the copper a tiny bit around were the IHS fits. (hard to see without the light, but it's on all four corners.

Looking at it closer, I should have probably put the TIM a little closer to the center of the base on the 2 outside heat-pipes. Putting the TIM in the dead center of the outside pipes puts the TIM a little too close to the edge of the IHS.

Perhaps I'll re-apply it at a later date. If I do, I'll be sure to take some pics before wiping off the TIM.

The AL support is quite big, and the pipes are squared off inside. Seeing how much they sanded down the 8mm pipes to make a flat surface does make me wonder how thick the copper tubing is and the inside diameter.

I really doubt this HSF is ideal for a CPU without an IHS. I would think The pipes rely heavily on the amount of contact surface. Without a solid base on the HSF, the heat transfer from the die would be facilitated by the center pipe and perhaps the AL spacers that would come in contact with the die. That and the fact that such a small surface area would really wreak havoc on a small area on the copper tubing. The weight distrubution on the center would probably really damage the copper.

Liquid3D said:

"Pumpout" low viscosity pastes - (repeat the mantra (quote below) with the Lite Beer Commercial of the 80's in your mind)

"All Fillers - No particles"

A result of this formula is something called pump-out occuring under the very conditions which are at the crux of our discussion. That formula: Heat + Pressure + Time = failure. Of course that xconclusion is forgone given the last value T but we don't want it to occur months after the paste has been applied.

While PC-enthusiats tend to change their CPU's every hour on the hour we are easily taken by pastes which set rapidly giving great temps but deteriorate almost as rapdily. We are immune to such problems simply because by the time the paste would begin to detrerioate we've already changed it out installing the newest core revision.

However we need to think about what makes a great thermal paste. As counterintuitive as it might seem Temperature is not the only attribute which determines success. Of course it's an intregal part of the equation. However; for the company running servers with millions invested in those files, experiencing an overall 5C drop in temps, while sacrificing 18-months of additional covereage, lower temps mean little compared to TIM lifespan. They need long-term stability, consistancy, longevity, and decenbt temps. And for the average PC user who applies a Bill-A-Goop type formula for instant gratification and then a year later wonder why their experiencing random re-booting, BSODs and buggy programs etc. They may suffer most because it never presents itself in a catastrophic failure, but months of frustration trouble shooting what are hardware issues misconstrued as software system errors.


That's the worst case scenario of a "Pyromaniac Ghost in the Machine." What they need is a stable long lasting product. I'm running several new tests for my Thermal Paste Addendum which includes Noctua and IC-Diamond and some new tests. I've just finished testing Tuniq TX-2 under high-heat, high-pressure between the Q6600 and Swiftech Apogee GTX.

EDIT: I almost got confused as to wich paste I applied hehe. But from the next photos I have to get on my PC you'll see the difference betwen IC-Diamond and other pastes. It just doesn't "break down" as easily and the spread pattern look healthy. I am up early today whoo

Below is the paste after the fact, is this pump-out;

Click to expand...

I remembered your problem with the push pins and saw this on the hardwareLogic forum results we may make special note of it in our troubleshooting guide because of the small sampling size of 50-60 tests 2 people had problems that we are aware of.

How about the rest of the Pump out pics? I am anxious to see them. When I experienced it it was on a 1cm die and I did not have a good view of it or are you saving them for your review?.

I am bumping this one as a reminder to the 100 or so that still have not posted their results.

Many thanks to those that took the time and effort to test and post results. Much Appreciated. Great Job!

also we are at the 3 month mark and would like to get some feedback on long term stability.

And for those that have an interest Anandtech giveaway is starting to post their results.

http://forums.anandtech.com/messageview.aspx?catid=39&threadid=2164479&enterthread=y

just tried the ic7d on my backup rig.
[email protected] 1.25v
asceramique
idle 36/37c
load 45/46c(orthos small ffts)

ic7d
idle 34/35c
load 40-42c
very impressed with the drop in load temps.
main rig will be tested next week sometime(waiting for new mb).
ive also tried it on several of my clients rigs and have seen on average a 10-12c drop.maybe not the fairest of comparisons going from the crud tim that was on them but its still pretty darn impressive imho.

I'm sorry but I won't have any long-term tests. The systems I tested the ICD7 on ended up getting different heatsinks when they were moved off the test bench to a case ...

JoeC said:

I am bumping this one as a reminder to the 100 or so that still have not posted their results.

Many thanks to those that took the time and effort to test and post results. Much Appreciated. Great Job!

also we are at the 3 month mark and would like to get some feedback on long term stability.

And for those that have an interest Anandtech giveaway is starting to post their results.

http://forums.anandtech.com/messageview.aspx?catid=39&threadid=2164479&enterthread=y

Click to expand...

That's why I didn't sign up for the free sample. I simply didn't have the time to do the test in a reasonable time period.

All you guys who got the free samples need to step up and finish your tests

I too, like QuietIce, have no long term test results. Both the pc's I was testing the ICD7 on are both gone .

However I have a new build in the works and will be retesting the ICD7. New build specs:

DFI BloodIron P35-T2RL
Intel XEON E3110
2GB kit of Crucial Ballistix
Swiftech Storm waterblock.

My new test results will be posted here.

I've swapped HS's

Sorry I forgot... been so busy with college and work I have even had time to crack the cover off

Hopefully in the next few weeks ill set it up since you reminded me.

Ditto , sry Ive yet to put up my results .

Been crazy busy with family matters (health related) .


I promise some results to be posted soon , specialy since Ive got a Q6600 and E3110 to test out .

*Update*

About 1 week running the new setup (sig) Using ICD7.

Current vcore: 1.28v @ 4.0GHz Water cooled with a Storm.

Temps are:

In Hardware Monitor:
Idle: Core0= 42c? Core1= 36c
Load: Core0= 49c Core1= 49c

In RealTemp:
Idle: Core0= 32c? Core1= 25c
Load: Core0= 43c Core1= 43c


For some reason core0's temp is somewhat pegged and will not drop below either 42c in HWM and 32c in RT, but once I put a severe load on the chip, both cores seem to respond equally and both temps raise

Will update in 30 days.

I have no reliable means of measuring my ambient temps in my room but it's safe to say they've hovered in the 20C range this winter. On my Socket A Athlon XP-M at 2.4GHz, my computer has seen an average of a 3C drop in temps. My temps used to be in the 52-55C range on a broken in AS5 application, depending on ambients changing during the daytime. My new temps on IC7 are in the 49-52C now which results in about a 3C drop on a CPU in the 100W range on a Thermalright SP97 with a Panaflo L1A 80mm at full speed. All of this is on BOINC SETI 100% load.

I run my own PC business and I'd like to get IC7 in large tubes. 15grams would be the size I'd be picking up every few months depending on amount of work I've got. Someone drop me a line and let me know who to get in touch with.

BTW finally bumped up my FSB to 210 with only a 1ºC jump for an extra 10Mhz FSB very sweet stuff.

AmbientFiction said:

I run my own PC business and I'd like to get IC7 in large tubes. 15grams would be the size I'd be picking up every few months depending on amount of work I've got. Someone drop me a line and let me know who to get in touch with.

BTW finally bumped up my FSB to 210 with only a 1ºC jump for an extra 10Mhz FSB very sweet stuff.

Click to expand...

We are going to be selling a larger size in the next couple of weeks.

24Carat size

No takers on the pressure test?