Finally here are my results
. Happy reading guys
.
1) INTRODUCTION,THERMAL COMPOUNDS & TESTING PHILOSOPHY
First of all I would like to thank Andrew (IC Diamond) for providing the opportunity to test the IC thermal compounds. When these were first mentioned by the forums dons, it was quite an interesting read and I wanted to try them out but I already had MX-4 at that time which was giving me good results over the Zalman ZM-STG2 which I used previously. So didn't feel the need to buy these new IC products. However now the opportunity came and I went straight for participation
.
For the test, I used MX-4, IC Perihelion and IC Diamond as seen below (Apology as my digital camera is crap lol
)
Each compound was applied for 1 week and then tested. During this time period, computer was used for general use, gaming and little bit of Prime95 was also used to heat up the TIMs. So ample time was provided for each TIM to properly settle down. These are the following application and testing dates for each TIM:
MX-4
05/03/2011(Application Date) - 12/03/2011(Testing Date)
IC PERIHELION
13/03/2011(Application Date) - 20/03/2011(Testing Date)
IC DIAMOND
21/03/2011(Application Date) - 28/03/2011(Testing Date)
These compounds were tested on my lovely
[email protected] which is a 'hot' 65nm chip. It is already at 1GHz overclock and seems to generate plenty of heat especially under load. I have a thread about Q6600 heating up my room so you can imagine it's heating potential
.
My cpu cooler is Thermalright IFX-14 which is the original ancestor of Cogage Arrow and Silver Arrow. The heatsink design is similar to latter two with few differences. Here are the side-by-side pics of IFX-14 and Cogage Arrow:
http://forums.overclockers.co.uk/showpost.php?p=17846578&postcount=22
IFX-14 has a two-screws mounting system and when fully tightened, it exerts tremendous pressure on the cpu IHS and thus on the TIM ideal for gauging TIM performance.I will go more into detail about the cooler later on.
The cpu stress program which I used is the infamous 'Intel Burn Test' or IBT. This program tends to send shivers among the overclocking community as it is widely acknowledged to be the most stressful cpu test program on the planet. Well LinX aswell as both are LINPACK with slightly different graphical user inteface. I like it for it's simple ease of use
.
IBT is a double-precision (64bit) LINPACK program which makes use of Gaussian Elimination Method to solve system of linear simultaneous algebraic equations involving matrices. It uses ram to store those equations and then floating point operations are carried out or Flop (Multi-Add) and the results are outputted in GFlops with the residual values displayed to the right.
The aim is to get as high GFlops values as possible ideally close to cpu's theoretical maximum GFlops at a given cpu speed. This means cpu is performing at greater speed and will generate more heat as a result.
The most important bit during testing is to ensure that in all tests for each TIM, similar values of GFlops are obtained so cpu is performing at same speed for each TIM. This will ensure consistency and testing results will be accurate.
Although prime95 is also a good program for measuring TIM performance, IBT tends to increase the cpu temps by atleast 8-10C more. With all the right components (ingredients
) in place, each TIM will be tested to it's limit.
Finally I have tried to be as accurate, consistent and methodical as possible. However being as accurate as possible means there will always be few minor errors as no experimental test is 100% precise. But I have tried my best to minimise them.
It has been a long time since I wrote any experimental report. The last time I wrote an experimental report was near the completion of my engineering degree. It will be a pleasure again writing such report in this forum
.
2) MAIN HARDWARE
The following were the main components forming my PC system:
CPU
Intel Core 2 Quad Q6600
[email protected] (overclocked) (VID 1.2625v)
http://ark.intel.com/Product.aspx?id=29765
Load Line Calibration (LLC) was enabled
Bios Vcore: 1.3625v
Idle Voltage: 1.328v
Load Voltage: 1.328v-1.312v
Motherboard
Gigabyte GA-EP45-UD3LR Rev1.0
http://www.gigabyte.com/products/product-page.aspx?pid=2951#sp
Bios Revision: F11
RAM
Kingston HyperX 4GB (2x2GB) DDR2 PC2-8500C5 1066MHz Dual Channel (KHX8500D2K2/4G)
http://www.overclockers.co.uk/showproduct.php?prodid=MY-039-KS&tool=5
The ram speed was running at 504MHz (DDR2 1008MHZ).
Timings: 5,5,5,15
DRAM voltage (bios): 2.0V
Cooler
Thermalright IFX-14
http://www.thermalright.com/products/index.php?act=data&cat_id=24&id=153
Cooler Fans
The famous Akasa Viper 120mm fans x 2
http://www.akasa.com.tw/update.php?...type=Fans&type_sub=PWM Control&model=AK-FN059
Graphics Card
Sapphire HD 5850 (stock speed)
http://www.sapphiretech.com/presentation/product/?leg=&psn=000101&pid=338
Power Supply
OCZ ModXStream Pro 600w Silent SLI Certified Modular Power Supply
http://www.overclockers.co.uk/showproduct.php?prodid=CA-031-OC
Case
Cooler Master HAF 922
(one of the largest mid tower case and excellent for air flow)
http://www.coolermaster.com/product.php?product_id=6606
Here are few photos of my case to give better idea of case interior and also to appreciate it's beauty
. I am actually using CM Storm Sniper Windowed side panel instead of the default HAF 922 grill side panel:
Case Interior
Cable Management
Windowed side panel
Beautiful Solid Gaming Case
3) THERMALRIGHT IFX-14 CPU COOLER
As stated before, this cooler which is a dual tower heatsink is the original ancestor of Cogage Arrow and Thermalright Silver Arrow. It's heatsink design as noted above is similar in shape and size compared to the other two high end air coolers.
It doesn't come with any fans and thus can be considered a passive heatsink. However it's main strength and performance effectiveness lies in the fans that are attached to it. So if slower fans are attached to it then it will give performance in similar numbers and if faster fans are used on it, then it will likewise give higher performance.
The Akasa viper 120mm fans are considered among the best 120mm cooling fans and with high cfm of 83.63 they shift a lot of air. At full or high RPM they sound like mini jet engines
.
So with the IFX-14 attached to Akasa Viper fans, it becomes one of the most bad ass high end air cooler which is actually needed to cool
[email protected]. Following are some of the pics of the cooler:
Apparently the IFX-14 base isn't perfectly flat but rather it is slightly convex. This is so greater contact can be made between cooler base and cpu IHS at the centre where most heat is generated and consequently more pressure can be applied at the centre.
Last time before testing, I took off the cooler as I wanted to see the MX-4 pattern and the pattern corresponded with the findings of xbitlabs. For more info see the following post:
http://forums.overclockers.co.uk/showpost.php?p=18506256&postcount=97
IFX-14 has a 2 spring-loaded screws mounting system which will be discussed later on.
4) FAN LAYOUT
The following is the layout diagram of the different fans in my PC Case and which some of you have already seen:
The two Akasa vipers (1900RPM rated) and 140mm CoolerMaster (CM) fan (1200RPM rated) are connected via akasa pwm splitter cable to 4pin 12V cpu_fan header.
http://www.overclockers.co.uk/showproduct.php?prodid=CB-031-AK
The rear 120mm CM fan (1200RPM) is connected to 3pin 12V sys_fan1 header.
The front 200mm CM fan (700RPM rated) is connected to 4pin 5V sys_fan2 header.
The top 200mmCM fan (700RPM rated) is connected to 3pin 12V pwr_fan header.
I conducted a separate short test under similar settings (which were used in actual tests and in which I will go in detail later on
) using HWMonitor as I wanted to find out fan speeds under idle and load for
[email protected]. So EIST and C1E were disabled in bios and Q6600 was running at 3.4GHz even at idle.
However HWMonitor only displays one value for the 3 fans connected in pwm config which I believe is for the vipers as they run faster than 140mm fan even at idle. Sadly I have no info on the 140mm fan speed at idle and load. Although previously I did connect 140mm fan to 12v fan header and it span at full 1200RPM give or take 20RPM. So I can only assume that at idle it ran at 600RPM give or take few RPMs aswell.
HWMonitor gives three figures for temperature readings:
Value(operating speed),
Min and
Max. The test was conducted three times noting the three readings under idle and load (IBT) for each fan displayed.
The
Value readings were averaged to give an average operational speed for each fan under idle and load. Then the
Min and
Max readings were also averaged and the difference between
average value and
average max and
average value and
average min readings was used to determine the
error speeds i.e. give or take speeds/RPM from the average operational speed.
For
[email protected]
Fan0 (Akasa vipers + 140mm fan)
Idle speed: 848RPM +/- 30RPM
Error = (30/848) x 100 = 3.54%
Load speed: 1721RPM +/- 25RPM
Error = (25/1721) x 100 = 1.45%
Fan1 (Front 200mm fan)
Idle speed: 392RPM +/- 15RPM
Error = (15/392) x 100 = 3.83%
Load speed: 414RPM +/- 27RPM
Error = (27/414) x 100 = 6.52%
Fan2 (Top 200mm fan)
Idle speed: 742RPM +/- 5RPM
Error = (5/742) x 100 = 0.67%
Load speed: 737RPM +/- 8RPM
Error = (8/737) x 100 = 1.09%
Fan3 (Rear 120mm fan)
Idle speed: 1263 +/- 18RPM
Error = (18/1263) x 100 = 1.42%
Load speed:1255 +/- 15RPM
Error = (15/1255) x 100 = 1.20%
Ofcourse the above readings are all based on estimation and as there were fluctuation in fan speeds represented by max and min values in HWMonitor, one can say that we can atleast get a rough indication of what speeds those fans are spinning at under idle and load. For some reason the rear 120mmfan and top 200mmfan speed decreased slightly under load.
In ideal world we would like the fans to spin at constant speeds. However in real world tests there will always be some variation as indicated by those error values.
Although those error values are very small, they may have minute effect on cpu temps. The other way of looking at it is (analogy) if we had an additional 30RPM (largest error speed) fan placed inside the case. Compared to Vipers and rest of the fans, it will be a very very slow fan. It may have a very tiny impact on the results.
So as you can see this is one example of errors in the experimental investigation.
5) MOUNTING PROCEDURE
The following pics show the IFX-14 mounting system and it's application:
Thermalright IFX-14 Mounting System
Application of Mounting System
As you can see that IFX-14 mounting system is a two spring-loaded screws mounting system. The mounting of the IFX-14 is one of the most critical issue in that if the heatsink isn't placed properly or aligned correctly on the cpu IHS,then this could affect results. However some movements will occur as a result of mounting application as the heatsink has a tendency to 'slide' when placed on top of pea-sized blob TIM.
However the aim is to minimise the movements of heatsink on top of cpu IHS as much as possible and this also means that some human error will be introduced.
In order to minimise human error and to place the IFX-14 on Q6600, the following procedure was carried out:
1) The
back plate was inserted into the back of the motherboard
2) The
bracket mounts were placed and aligned on top of the hollow pillars of the
back plate
3) Four
M3X0.5 screws were then inserted into the hollow pillars to secure the
bracket mounts on top of the motherboard to the
back plate on the backside of the motherboard
(Notice that once the
back plate and
bracket mounts were secured to the motherboard, they didn't need to be taken off again when new TIM was applied to cpu IHS)
4) A pea-sized blob of TIM according to IC instructions was applied on the centre of cpu IHS
5) The IFX-14 heatsink was then placed on top of the cpu IHS carefully and the centre of heatsink base (the top surface of the base has a circular groove in the centre for placing the mounting plate) was aligned with the centre of
bracket mounts where the
UNC 6-32 spring loaded screws would be inserted. This is so that mounting plate can be placed in a straight position on top of the base as seen in the pics above.
(Step 5 is the most tricky step and when trying to align the centre of IFX-14 base with the centre of
bracket mounts, 2-3 instances of movement did occur in the case of each TIM)
6) The heatsink was then held in that position for about 1 minute by both hands and during this time frame, pressure was applied by both hands, so that TIM could spread a bit more and that heatsink would be somewhat glued to the TIM. This also ensured that heatsink would stop 'sliding'.
7) Then the
mounting plate was placed on top of the heatsink base, with my index finger holding onto the centre of mounting plate and with the other hand inserting
UNC 6-32 spring loaded screw into the
bracket mount and giving it few turns until it was secured enough. Same thing was repeated with the other screw.
(If I had held onto the heatsink with one hand while using the other hand in trying to tighten the screw, the mounting plate would have flipped. So you can see it isn't one of the user friendly system. You have to be very careful)
8) Once both
UNC 6-32 screws were secured enough, then I used both my hands to turn both screws further until I could go no further due to heavy friction.
9) Finally I used the screw driver to give 1-2 turn to each screw in alternate manner until they were fully tightened. While tightning screws to the max, I was also observing them from side to ensure they went fully down and that there was no further room left for them. (The pressure exerted on cpu IHS is tremendous and even though I have no software to check pressure, I can safely say that compared to my older Thermaltake Big Typhoon cooler which has 4 screws mounting system, the amount of exertion required to tighten the screws fully down was a lot more in the case of IFX-14)
The above procedure was repeated for each TIM
.
6) TESTING PROCEDURE
On the respective day of test, each TIM was tested at three different times:
afternoon, evening and night to get broader overview of how each TIM performed under different environmental conditions.
Throughout each day of test the room window was slightly open by about 0.7inch from the closing position and the room door was fully open to simulate normal room environment. This was also done to allow for some circulation of air inside the room. Also generally as the day progressed into evening from afternoon and then to night, the room temperature fell and hence the ambient temperature was also affected by small amount.
As you will see later on that almost all the ambient values from the tests were within 0.8C of each other which are very close to each other. It is impossible to get exact value as room temperature and hence ambient temp is in state of fluctuation through out the day and even on different days.
Hence why the TIMs were tested at 3 different times to get better overall view.
The other testing scenario would have been to conduct tests in closed room. This would mean no fresh air and during tests the same air would have been heated and then cooled when tests finished. So air temperature would have still varied and dropped as the day progressed. Also within the same room, different points will be at different temperatures and hence why the ambient temperature is needed which differs from room temperature.
Ambient temperature is the temperature inside the PC case which slightly varies from room temperature as electronics also exhibit heat during operation and is usually higher than room temperature. In my case if the room temperature is around 17C then ambient temperature will be around 1-1.5C higher.
I think the ideal way of testing these TIMs would be in vacuum where there are no external temperature influences.
However as long as the ambient values are close to each other and not like 10C apart, then we can take into account the temperature rise or delta T of each TIM which is crucial in determining which TIM performed better. With delta T (will be discussed later on), ambient temperature is taken out as a factor and then concentration is solely placed on the performance of each TIM. For more details about what I have stated above please see the following posts:
http://forums.overclockers.co.uk/showpost.php?p=18661898&postcount=239
http://forums.overclockers.co.uk/showpost.php?p=18663496&postcount=240
http://forums.overclockers.co.uk/showpost.php?p=18692617&postcount=245
http://forums.overclockers.co.uk/showpost.php?p=18694864&postcount=248
http://forums.overclockers.co.uk/showpost.php?p=18702133&postcount=252
1)Before testing, C1E and EIST were disabled in bios. So Q6600 was operating at 3.4GHz even at idle and secondly since those power saving options were turned off, this meant that nothing would be holding back Q6600 from performing under IBT and thus heat generation would also increase.
2) Thermometer was placed underneath the first intake Viper fan next to the graphics card as this is the best position I could find.
3) I used coretemps to monitor Q6600 temperatures at idle and load. Note I will be only taking into account Idle
Low and Load
High. i.e the lowest the coretemps can go and the highest they can reach. So in other words the extreme readings.
So once booted into Windows7, 15 mins later
idle values were noted and at the same time
ambient temp was also recorded. Because straight after IBT (load) was started.
4)For
Intel Burn Test, the following settings were used:
-Custom Stress Test/Level
-Ram value 2800MB (70% of my total 4GB ram)
-Threads: 4 (Manually set)
-Times to run: 30
The beauty of 'Custom Test' is that we can tweak memory values and see how high GFlops values we can get. I always use 'Custom' now and 2800MB always gives me 45GFlops out of a possible 54.4GFlops (Theoretical maximum) for
[email protected]. I believe this is due to IBT storing same no of equations and hence the residual value will also be same. This also means that cpu is always performing at faster rate and thus generating more heat in the process.
5) Once the IBT test was completed, the readings were noted and computer was turned off. The side panel was also taken off and a fan in my room was used to cool the components for 1 minute. Afterwards the computer was left in that state for 2-2.5 hours to allow for adjustment in the room temperature.
I believe the benefit of conducting test in open room was also to allow all the heat that was generated during IBT testing to escape from the room quickly and thus we reached normal room temperature quickly before the next test.
6) This then took us into the second test i.e. evening test and the above procedure from
step3 to
step5 was repeated. Then same thing for the third test(night)
The above procedure was carried out for each TIM.
7) RESULTS
Here are the results
.
Before posting results, one or two things to mention:
As Intel Burn Test is a double-precision (64bit) LinPACK program, the theoretical maximum for
[email protected] is 54.4GFlops in double-precision which is calculated as follow:
No of cores x Flop per core x cpu speed = 4 x 4 x 3.4GHz = 54.4GHz
Here is the intel GFlops data sheet for stock speeds.
http://www.intel.com/support/processors/sb/cs-023143.htm#3
If you carry out the calculation mentioned above for those speeds, you will get the same number for GFlops
.
As stated before I manage to get 45GFlops which is about 83% of the theoretical maximum as there is always hardware limitation such as ram latencies and essential windows background programs running that can't be terminated. However this is still a very high value.
Secondly Q6600 is not a native quad core architecture cpu but rather two dual cores E6600 that are joined together on one die. So it is very likely that both chips can produce different amount of heat under loading. This actually happens in my results as cores0,1 are more sensitive to loading than cores2,3 and generate more heat. I also suspect that cores0,1 consist of one dual core chip and cores2,3 consist of the other.
MX-4
-Tends to spread easily but is more sensitive to the physical design of heatsink base.
5mm Pea-sized blob
TEST1 (Afternoon)
Ambient 19.5C
TEST2 (Evening)
Ambient 19.3C
TEST3 (Night)
Ambient 19.1C
Spread Pattern after Tests completion
Spread Pattern on IFX-14 Base
IC PERIHELION
- Very thick compound. I used a bit more of 6mm as I wasn't sure if it will spread enough. IC Diamond has also stated that it is better to apply a bit more than less.
6mm Pea-Sized Blob
TEST1 (Afternoon)
Ambient 18.1C
TEST2 (Evening)
Ambient 18.7C
After the 2nd test I wasn't happy with the coretemps I was getting as there was big difference between core0,1 and core2,3. So before carrying out the third test I decided to re-apply the IC Perihelion and thus re-mounted the IFX-14 using the procedure mentioned above. Then prime95 was used for 1 hour and then gaming for the 2nd hour to heat up the compound. Afterwards computer was turned off for about 30 mintes to let it cool down and also using my fan for about 1-2 minutes.
Then the 3rd test was commenced to see if it made any difference.
Spread Pattern after 2 Tests completion
Spread Pattern on IFX-14 Base
TEST3 (Night)
(Re-applied TIM and re-mounted IFX-14)
Ambient 18.9C
For the third test I decided to apply 5mm of pea-sized blob to see how the spread pattern would turn out and whether slightly less application would help.
5mm Pea-Sized Blob
Spread Pattern after 3rd Test
Spread Pattern on IFX-14 Base
IC DIAMOND
- Thinner than IC Perihelion but thicker than MX-4
-When it came to testing IC Diamond, the clocks in UK then have already moved by 1 hour forward. So although for example the first test was completed at 3.45pm, the original time was still 2.45pm.
5.5-6mm Pea-Sized Blob
TEST1 (Afternoon)
Ambient 19.3C
TEST2 (Evening)
Ambient 19.3C
TEST3 (Night)
Ambient 19.0C
There are no pics of IC Diamond spread pattern as I am still using it since completing tests on it
. However I did apply a pea-sized blob of 5-5.5mm and placed IFX-14 on it while applying the pressure with hands. After 2-3 minutes I took off the heatsink to see how the spread pattern was and to my surprise it spreads just as nicely and smoothly as IC Perihelion with thin layer spread across the cpu IHS
. Afterwards the IHS was cleaned with isopropyl alcohol and the TIM was re-applied for testing and whose results you can see above
.
Looking at results, IC Diamond has clearly given the lowest load temps
. As can be seen that as the testing days progressed into night from afternoon, the load temps became noticeably lower for each TIM except in the case of IC Perihelion whose load temps in the evening were slightly higher than in the afternoon.
However in the case of idle temps, none of the TIM gave any significant advantage although as we will see later on in analysis, there were slight difference in the idle temps between the TIMs.
As seen from IBT testing time, each TIM was subjected to about 1 hour of Intel Burn Test at high GFlops in all the respective tests. This is extremely stressful for any cpu as 1 hour of IBT load with high GFlops values will push any cpu to it's limit and thus the TIM applied.
Normally IBT tends to fail on the first run if not enough bios vcore is supplied to cpu or the heatsink is inadequate. Hence in my opinion IBT is the perfect program to test any thermal compound.
Looking at the load temps again, there is atleast 9-10C difference between IC Diamond and IC Perihelion under Intel Burn Test. So IC Perihelion didn't perform aswell as I was expecting it to be even with the high end cooler attached, temps went upto 80C easily. Probably this is to do with it's ceramic properties which aren't as efficient as diamond properties in heat transfer.
MX-4 in all the tests came in 2nd place and there is atleast 3-4C difference between IC Diamond and MX-4 under load.
In terms of ambient temp, almost all the readings were within 19.5C - 18.7C = 0.8C of each other with the exception being 19.5C - 18.1C = 1.4C which is still a small value. So the ambient readings were very close to each other.
However it is due to the difference between ambient readings that we have to take into account the
Delta T or
Temp Rise or the
Difference between average idle,load and ambient as the
main performance measure to determine which TIM performed the best overall as stated in post 248 in section 6).
This will be dealt with in the next analysis section where results are analysed and presented in a more methodical manner so a better view can be gained of each TIM performance as the above posted results are all raw results
.
Finally the spread pattern of TIMs varied even under same mounting procedure. MX-4 seems to be more accomodating of the IFX-14 base design which is slightly convex underneath. So although it spreads more and is more 'wet' it is also more sensitive to base curvature and hence is thinner in the centre. However overall it spread itself into a fine thin pattern across the cpu IHS.
IC Perihelion and IC Diamond tend to spread the most beautifully with an even layer of fine thin pattern and aren't affected by the base design of IFX-14. I would say that IC Products exhibit the ideal spread pattern and I was impressed
.
However as seen in the case of IC Perihelion 6mm was more adequate than 5mm in spreading further. However in both cases the centre of IHS was nicely filled which is the most important part. Finally as for the third IC Perihelion test, it didn't make much difference as load temps were more or less similar although slightly lower as we were in night time
.
8) ANALYSIS
The following are the tables of analysed results for each thermal compound to gain better understanding of each TIM performance.
Looking at the the tables we can see that when temperature differences are considered, IC Diamond still tops the list followed by MX-4 and then IC Perihelion. Converting these tables of results into graphs we get the following information:
GRAPH1
In the
graph1 we can see that in terms of average idle temps, there wasn't much difference. However overall average idle temps were higher in case of IC Perihelion while lower in the case of IC Diamond with MX-4 in the middle. Similar pattern is also observed in average load temps aswell.
GRAPH2
In the
graph2 the ambient temps were included and it can be seen that they are more or less at similar level with slight variation as discussed previously.
The light bars represent the difference between average idle and ambient while dark bars represent the difference between average load and average idle for each TIM. Again overall IC Diamond comes on top followed by MX-4 which was followed by IC Perihelion.
GRAPH3
In
graph3 we subtract the ambient from average idle and average load in each test for the thermal compounds. So we are only left with temperature rise in both cases of idle and load. With the ambient factored out we can visually see which thermal compound has been the most effective or given best results in reducing the coretemps.
At load we can see the performance gap between all three thermal compounds and this gap increases further in the case of IC Perihelion.
In the above table the
overall delta T idle and
overall delta T load were calculated for the three tests combined by averaging the respective values in
graph3.
For example MX-4: Overall Delta T Load = (51.0 + 49.95 + 49.90)/3 = 50.28C
Presenting those results in the final graph:
GRAPH4
There you have it
.
IC Diamond has beaten both MX-4 and IC Perihelion.
I will let you guys work out the cooling/performance differences between the three thermal compounds
.
9) CONCLUSION
Until IC Diamond thermal compound was introduced, I considered MX-4 to be the best thermal compound as it is successor to MX-3 which itself is a top product. But having gone through the experiment, it has been an excellent learning experience and IC Diamond is truely an impressive product.
As previously stated in any experimental investigation there are bound to be errors and those errors were in the form of some fan fluctuations, some movements when placing IFX-14 on top of Q6600 and slight variation in ambient temps which were taken care of by making use of
delta T calculations as we are conducting tests in real world dynamic atmosphere.
However as seen in this investigation I have tried to be as through as I can and tried to minimise errors to the best of my ability.
The three most important components in testing were:
Q6600 which is considered to be one of the hottest cpus but is a legendary chip. Having it overclocked to 1GHz over the stock speed of 2.4GHz was no simple matter itself but was an excellent factor in determining the thermal compounds performance.
IFX-14 heatsink A dual tower high end air cooler whose mounting system having convex base allows for greatest contact in the centre of cpu IHS where most heat is generated and whose mounting system exerts immense pressure on the cpu and thus on the TIM.
Intel Burn Test The dreaded cpu stress test program that increases the cpu coretemps by atleast 8-10C more compared to Prime95 and which is designed to stretch the cpu to it's limit and the heat generation during IBT testing is considerable.
IC Perihelion on the hand didn't perform as well and couldn't compete with either MX-4 or IC Diamond under IBT. However for moderate overclocks, IC Perihelion will do well.
On the other hand for high end or extreme overclocks, IC Diamond is the new king. Although MX-4 competes very well with IC Diamond and performance gap is only about 2-3C. So MX-4 is still also a high end product.
Overall here is my rating :
IC Diamond : Best
MX-4: 2nd Best
IC Perihelion: 3rd Best
Again I would like to thanks Forum Dons for introducing IC Products and Andrew(IC Diamond) for providing the products.
WingZero30
