Thermal Paste vs. Reusable Graphite Thermal Pad Benchmarks (IC Diamond Pad)
video description
Date: 2020-06-24
Related videos
Comments and reviews: 10
Talishar
I've always wondered but your DeltaT over ambient values always seem to be grossly off from what should be expected. Where are you getting these values from? Even at 95 watts, your thermocouple would be reading well over 100C and beyond the thermal limits of pretty much every CPU out there. By this chart, I wouldn't touch either product with a ten foot pole because even at 95W, the deltaT values are ridiculous. Getting a value over 5 when you are normalizing a DeltaT by ambient would be a red flag for electronics unless your ambient temperature was under 10C. At STP of 21C a computer wouldn't see a normalized value over 3.8 as the computer would have either thermally throttled or BSOD to save itself. The final set at 270W is nearing natural gas turbine idle temperatures as at 21C for ambient, your final set would mean that your thermocouple read around 460C which would have most PCBs a molten mess at that point.
While i agree that using celsius when you want to be more accurate in showing heat capacity is bad, the normalized values should make sense. For charts like these, you should be directly defining what value you are subtracting from Tdie or whatever and what your ambient is. If you don't just want to report what the Tdie or whatever direct sensor value your reading off of, you should be giving numbers in units of heat [J]. While the graph gives a good visual of the basic general relationships between the different power settings and thermal mediums, the kinda crazy high value at even the lowest power setting kinda says that whatever is being used to measure everything is malfunctioning or off and the rest of the data is then inaccurate or can't be trusted.
reply
I've always wondered but your DeltaT over ambient values always seem to be grossly off from what should be expected. Where are you getting these values from? Even at 95 watts, your thermocouple would be reading well over 100C and beyond the thermal limits of pretty much every CPU out there. By this chart, I wouldn't touch either product with a ten foot pole because even at 95W, the deltaT values are ridiculous. Getting a value over 5 when you are normalizing a DeltaT by ambient would be a red flag for electronics unless your ambient temperature was under 10C. At STP of 21C a computer wouldn't see a normalized value over 3.8 as the computer would have either thermally throttled or BSOD to save itself. The final set at 270W is nearing natural gas turbine idle temperatures as at 21C for ambient, your final set would mean that your thermocouple read around 460C which would have most PCBs a molten mess at that point.
While i agree that using celsius when you want to be more accurate in showing heat capacity is bad, the normalized values should make sense. For charts like these, you should be directly defining what value you are subtracting from Tdie or whatever and what your ambient is. If you don't just want to report what the Tdie or whatever direct sensor value your reading off of, you should be giving numbers in units of heat [J]. While the graph gives a good visual of the basic general relationships between the different power settings and thermal mediums, the kinda crazy high value at even the lowest power setting kinda says that whatever is being used to measure everything is malfunctioning or off and the rest of the data is then inaccurate or can't be trusted.
reply
Jon
Hello, Steve;
FANTASTIC!
THANK YOU! THANK YOU! THANK YOU!
Thank you for all you do.
IF you liked the video LIKE the video.
I REALLY appreciate this!
I've been waiting for some time for this review but it raises a question or 2.
I'm using a CM HYPER 212 EVO on an AMD FX-6350.
Both of them were lapped on 'fine' wet-dry aluminum carbide lubed w/ WD-40 on a piece of window glass.
The reason for doing this was to smooth the cooler base and the IHS for maximum heat transfer.
On my Asus M5A97 R2.0 using the 'stock' overclock to 4.2GHtz I get normal temps in the high 30sC (usually).
I know you guys have a LOT to do & you'll still trying to catch up on a backlog projects.
BUT!
When you finish the cooling testing you're doing, could you consider taking like the top 2 products, 'lapping' the cooler base & the IHS ?
Then compare that w/ the previous results?
Do you see what I'm after here?
I want to know how much, if any, difference lapping makes.
So if you understand & want to find out too, DO IT!
BTW, my CM cooler base was seriously uneven so 'lapping' it smoothed/leveled/evened it a good bit.
[No 'measurements', sorry.]
Have a GREAT day, Neighbor!
PS
Keep Up the Good Work!
reply
Hello, Steve;
FANTASTIC!
THANK YOU! THANK YOU! THANK YOU!
Thank you for all you do.
IF you liked the video LIKE the video.
I REALLY appreciate this!
I've been waiting for some time for this review but it raises a question or 2.
I'm using a CM HYPER 212 EVO on an AMD FX-6350.
Both of them were lapped on 'fine' wet-dry aluminum carbide lubed w/ WD-40 on a piece of window glass.
The reason for doing this was to smooth the cooler base and the IHS for maximum heat transfer.
On my Asus M5A97 R2.0 using the 'stock' overclock to 4.2GHtz I get normal temps in the high 30sC (usually).
I know you guys have a LOT to do & you'll still trying to catch up on a backlog projects.
BUT!
When you finish the cooling testing you're doing, could you consider taking like the top 2 products, 'lapping' the cooler base & the IHS ?
Then compare that w/ the previous results?
Do you see what I'm after here?
I want to know how much, if any, difference lapping makes.
So if you understand & want to find out too, DO IT!
BTW, my CM cooler base was seriously uneven so 'lapping' it smoothed/leveled/evened it a good bit.
[No 'measurements', sorry.]
Have a GREAT day, Neighbor!
PS
Keep Up the Good Work!
reply
itech
In my seat of the pants home testing when I bought an IC Diamond pad I new that I was going to get a EVGA CLC 240 to put on top of my 2600X. I wanted to compare. After running both through the Time Spy benchmark for 30 minutes I found a 1-2 degree F higher variance running a all-core 4.2 ghz overclock which was not entirely stable but would complete the benchmark reliably. It was 95% percent stable I would say, it would have occasional blue screens on a game I have that is in alpha that is unoptomized and threw too much workload to the cpu. The pad has come in handy when I had to do an unexpected teardown and I did not have any on hand and no computer store open at 10pm. Later I used that cpu to test a motherboard for a family member and you guessed it, no paste needed, no cleanup. It has been in place for a year now and it still has not changed in performance. I will note I have not used any torque measuring devices only using a criss-cross pattern like you would a car wheel and what I would call a very firm torque for the final torque with my calibrated elbow. LOL
reply
In my seat of the pants home testing when I bought an IC Diamond pad I new that I was going to get a EVGA CLC 240 to put on top of my 2600X. I wanted to compare. After running both through the Time Spy benchmark for 30 minutes I found a 1-2 degree F higher variance running a all-core 4.2 ghz overclock which was not entirely stable but would complete the benchmark reliably. It was 95% percent stable I would say, it would have occasional blue screens on a game I have that is in alpha that is unoptomized and threw too much workload to the cpu. The pad has come in handy when I had to do an unexpected teardown and I did not have any on hand and no computer store open at 10pm. Later I used that cpu to test a motherboard for a family member and you guessed it, no paste needed, no cleanup. It has been in place for a year now and it still has not changed in performance. I will note I have not used any torque measuring devices only using a criss-cross pattern like you would a car wheel and what I would call a very firm torque for the final torque with my calibrated elbow. LOL
reply
Reggie
Wow, can't argue with those results. A really great setup!
I do think I might see one minor detail you might (but probably not) have overlooked, esp since it's not the the most common type of bearing in use in PC fans. I'm pretty sure that fans like Corsairs Maglev fans run into problems from a reduction in the air gap (or air cushion) formed in the maglev bearing and it's electrical field when run at reduced voltages. I've always understood it that a maglev fan really needs to be RPM controlled via PWM due to how the Maglev bearing floats in it's air gap. Unless I've miss understood how those particular fan's work. Please, it's not a knock on the setup at all (and as I noted, maybe I miss understood how Corsairs Maglev bearings work). This setup is all about creating a controlled & repeatable environment (and it does that with out any doubt) and it is exemplary work. The results of the tests between Thermal Pase and Graphite Thermal pads clearly show that.
reply
Wow, can't argue with those results. A really great setup!
I do think I might see one minor detail you might (but probably not) have overlooked, esp since it's not the the most common type of bearing in use in PC fans. I'm pretty sure that fans like Corsairs Maglev fans run into problems from a reduction in the air gap (or air cushion) formed in the maglev bearing and it's electrical field when run at reduced voltages. I've always understood it that a maglev fan really needs to be RPM controlled via PWM due to how the Maglev bearing floats in it's air gap. Unless I've miss understood how those particular fan's work. Please, it's not a knock on the setup at all (and as I noted, maybe I miss understood how Corsairs Maglev bearings work). This setup is all about creating a controlled & repeatable environment (and it does that with out any doubt) and it is exemplary work. The results of the tests between Thermal Pase and Graphite Thermal pads clearly show that.
reply
wishus
The only variable I can think of is steves hair cooling off everything within a 4 foot radius every time he flips it back over his shoulder.
I use a graphite pad in my retro builds when I just want to grab some things off my shelf and throw a system together for a couple days. I have reused one probably 20 times at least. Performance wise they do the job well enough. I don't overclock my vintage hardware at all. The systems are always open bench so i figure the actual CPU temps will be similar to it using thermal paste inside a case, so I am happy with this. The hottest my setups can get is either a Phenom 9950x4, qx6800, 1100T and maybe an 8 core xeon based sandy bridge. And it works well with all of them on stock settings. The big advantage I see is no mess. It is much quicker and just less fuss for this particular scenario. For my case builds which will be more permanent, or my personal system I use good paste of course.
reply
The only variable I can think of is steves hair cooling off everything within a 4 foot radius every time he flips it back over his shoulder.
I use a graphite pad in my retro builds when I just want to grab some things off my shelf and throw a system together for a couple days. I have reused one probably 20 times at least. Performance wise they do the job well enough. I don't overclock my vintage hardware at all. The systems are always open bench so i figure the actual CPU temps will be similar to it using thermal paste inside a case, so I am happy with this. The hottest my setups can get is either a Phenom 9950x4, qx6800, 1100T and maybe an 8 core xeon based sandy bridge. And it works well with all of them on stock settings. The big advantage I see is no mess. It is much quicker and just less fuss for this particular scenario. For my case builds which will be more permanent, or my personal system I use good paste of course.
reply
Neonmirrorblack
Still going to stay with the pads. The temperature difference is real, but it's not enough to actually matter unless you're trying to squeeze the most out of an overclock that you can. I'm currently at 4.9Ghz on a 9600K and well under a throttling threshold for gaming . Also using an NH-D15. Paste is obviously superior for thermals, but it's annoying when trying to remove a heat sink for cleaning, or transportation and also having to reapply it every few years sucks as well. After using paste for twenty years, I'm done with it since the pads are way more than sufficient for gaming loads, and they also don't wear out like paste does, unless you end up removing them and reusing them multiple times.
reply
Still going to stay with the pads. The temperature difference is real, but it's not enough to actually matter unless you're trying to squeeze the most out of an overclock that you can. I'm currently at 4.9Ghz on a 9600K and well under a throttling threshold for gaming . Also using an NH-D15. Paste is obviously superior for thermals, but it's annoying when trying to remove a heat sink for cleaning, or transportation and also having to reapply it every few years sucks as well. After using paste for twenty years, I'm done with it since the pads are way more than sufficient for gaming loads, and they also don't wear out like paste does, unless you end up removing them and reusing them multiple times.
reply
anti
I commend the effort to make things scientific. Torque wrenches and screw drivers have a lot of user difference that come along with them. Often results are different from one person to the next or even one hand to the other. There could even be thread defects that change things thread screw to screw. Even thread percentage would change friction. Hardware is not infinitely reusable either. Weight on the coolers would be a much more repeatable method. You'd just have to consider the weight of the blocks. Maybe using a actuator or hydraulic system would be easier to apply the same measurable force.
You can not torque something to one thread when there is no force holding it down. It's just not screwed.
reply
I commend the effort to make things scientific. Torque wrenches and screw drivers have a lot of user difference that come along with them. Often results are different from one person to the next or even one hand to the other. There could even be thread defects that change things thread screw to screw. Even thread percentage would change friction. Hardware is not infinitely reusable either. Weight on the coolers would be a much more repeatable method. You'd just have to consider the weight of the blocks. Maybe using a actuator or hydraulic system would be easier to apply the same measurable force.
You can not torque something to one thread when there is no force holding it down. It's just not screwed.
reply
mrdali67
I am thinking, If I am to buy a Water cooler like the corsair Icue H150, is the good paste like Hydronaut or Kryonaut gonna give a significant difference in performance compare to the standard pad/paste that is usually preinstalled on these aio's ? Going to build a system soon with the B550 Tomahawk and most likely an R7 3800X, in a Corsair Icue 465X case, as I can't wait untill xmas waiting for the new cpu's. The 3800X could propably do with the stock cooler, but its more if I am to buy a significan more beefy cpu later when we see how the new Zen3's performe, and if I get the H150 I am gonna have 6 rgb fans in there .. I shouldn't be needing any AC nor lighting in the room :)
reply
I am thinking, If I am to buy a Water cooler like the corsair Icue H150, is the good paste like Hydronaut or Kryonaut gonna give a significant difference in performance compare to the standard pad/paste that is usually preinstalled on these aio's ? Going to build a system soon with the B550 Tomahawk and most likely an R7 3800X, in a Corsair Icue 465X case, as I can't wait untill xmas waiting for the new cpu's. The 3800X could propably do with the stock cooler, but its more if I am to buy a significan more beefy cpu later when we see how the new Zen3's performe, and if I get the H150 I am gonna have 6 rgb fans in there .. I shouldn't be needing any AC nor lighting in the room :)
reply
RedBearAK
Because these pads supposedly have enormous capacity for horizontal thermal conduction, what I want to see someone do is take a giant sheet and clamp it to a passive heatsink the size of an ATX case side panel, with the end of a thin strip of the pad then clamped onto the CPU. Theoretically this material should be capable of wicking massive amounts of heat away from the CPU, with the heat traveling horizontally through the narrow strip to be dissipated by the huge passive heatsink outside the case. I want to see a passively cooled threadripper. No fan. A sheet that size might cost several hundred dollars but if it works it would be amazing.
reply
Because these pads supposedly have enormous capacity for horizontal thermal conduction, what I want to see someone do is take a giant sheet and clamp it to a passive heatsink the size of an ATX case side panel, with the end of a thin strip of the pad then clamped onto the CPU. Theoretically this material should be capable of wicking massive amounts of heat away from the CPU, with the heat traveling horizontally through the narrow strip to be dissipated by the huge passive heatsink outside the case. I want to see a passively cooled threadripper. No fan. A sheet that size might cost several hundred dollars but if it works it would be amazing.
reply
Tony
I got a thermal pad for my build after watching the Linus video cuz of the easy application (was my first build) and so that I wouldn t have to take off the cooler to reapply thermal paste sometime in the future in case I keep my current computer for a long time (I built a nice future proof computer that should last at least another 5 years with maybe only needing a graphics card upgrade). I m pretty lazy so those two points were good enough for me to use the thermal pad over thermal paste. I would be interested to see the long term durability of the thermal pad. Thanks for the video :D
reply
I got a thermal pad for my build after watching the Linus video cuz of the easy application (was my first build) and so that I wouldn t have to take off the cooler to reapply thermal paste sometime in the future in case I keep my current computer for a long time (I built a nice future proof computer that should last at least another 5 years with maybe only needing a graphics card upgrade). I m pretty lazy so those two points were good enough for me to use the thermal pad over thermal paste. I would be interested to see the long term durability of the thermal pad. Thanks for the video :D
reply
Add a review, comment
Other channel videos
