New
#2491
2017 Hardware Thread
-
-
-
New #2493
Thermal Paste Round-up: 85 Products Tested - Tom's Hardware
From Conclusion page:
In the end, it doesn't necessarily matter how much you paid for your paste, or what thermal conductivity its manufacturer claims on the package. More important is the how: your application method makes or breaks this process.
Some of the pastes we tested remain classics, and are even relatively inexpensive. It only follows, then, that we'd caution you against spending big on a brand new paste that'll only make a temperature difference of 1 to 2 Kelvin. If that sort of delta plays a significant role in your overclock, you're courting disaster. Sufficient airflow through the enclosure should ensure that you're never dependent on a particularly high-performance thermal compound.
More hardcore overclockers tuning their systems competitively, or quiet computing enthusiasts who spin their fans down as low as possible for better acoustics might want to give more thought to their paste of choice. Even a majority of power users, however, should focus on a paste’s ease of use, leaning toward the softer compounds. A flawed application typically impacts thermal performance much more than the difference between pricey and value-oriented pastes ever could.
-
New #2494
I agree and disagree with that conclusion. :)
If just looking blindly at temps, you don't have enough information for any conclusion at all, except for the temp at the sensor.
Paste quality, application quality, clean surfaces during application, cooler surface quality, plays a huge role in this. That 2K diff can always be compensated by better coolers. But where it really matters the most, what I mentioned plays a HUGE role.
All this can be compared to a stove and multiple frying pans. Get a good quality pan with good touch surface and good material in the layered core and the food get evenly cooked, while with the bad quality ones that are uneven, no multiple layers and bad material, the food gets burned and unevenly cooked. That indicator is more important than how much you paid for the pan.
None of these things have to cost you a fortune though as price is not the leading indicator here. Human error and bad quality products are. :)
-
New #2495
The test results pretty much mirrors some of the better tests I have seen done on thermal pastes. The difference in this test from #1 to #36 is 3°C, which is most likely within the margin of error of application method or temp monitoring. If 3°C makes a big difference to you, you are in big trouble no matter which paste you use.
-
-
New #2497
My point here is that you can get the temp to anything you want. -100.0C - +60.0C (or whatever) at full load no matter which paste you use, but does the temperature move evenly away from the CPU and does the temperature stay as constant as possible on both idle and full load? The paste won't make a difference in the end, as the test states, but many other factors does.
-
New #2498
Paste consists of very fine micro particles of material that conducts heat well, in many cases metal, held in suspension by a medium that hampers heat transfer as little as possible. It's type should be chosen mostly by type of contact surfaces which could be slightly concave/convex, rough/polished or different materials. All that influences heat transfer and if you test different pastes on same setup that doesn't mean it would be same as in another combination of cooler/CPU.
Other consideration for choosing a paste would be it's longevity and resistance to heat. If it melts and squeezes out at higher temps it will be no good even if it's best at lower temps. Other side of the coin are pastes that harden up in relatively short time in loose it's heat transfer capability. Just like stuff AMD used to use with their coolers in the past and many a processor got damage while trying to remove cooler. Once paste looses it's qualities things go downhill fast and falls into vicious circle of more heat > worse it performs > producing more heat.....
-
-
-
New #2500
Well I suppose we need to define some PC user roles....
Average PC user - One who mostly uses their PC for home or work. They're the ones who buy the store PC's and aren't interested in technical details outside of basics of understanding what they’re buying. For them they just need the PC to do the most menial of tasks such as email, net browsing, online shopping and maybe some lite gaming.
PC enthusiast - one who enjoys tinkering with and building PCs. They're highly knowledgeable, technically proficient, and usually hold jobs in the IT world. They're the ones who are usually first in line to try out new software/hardware as they look to increase the performance and efficiency of their PC's. They also tend to beta test new OS releases.
Gamers/Benchmarkers - basically a step above the enthusiast in that they crave performance. For them they are looking to get the most out of their PC's, especially where gaming and benchmarking is concerned. They look forward to new hardware releases, especially where the CPU and GPU, is concerned in hopes of gaining even better performance. They do however tend to be cautious with new OS upgrades fearing loss of performance. That said, like the enthusiasts, many do beta test new OS's. Lastly, this is usually the group hardware vendors market to since they’re the ones most likely to purchase those expensive high-performance parts.
With that, I surmise it is more likely the gamer/benchmarker who'd be willing to pay 1000+ for a CPU, 800+ for GPU, and 600+ for a MB. This however, doesn’t exclude the enthusiast world. And no, the average user wouldn't pay those prices.
BTW I consider myself a middle of the road person between enthusiast and gamer. That while I don't mind paying a premium for high end parts where performance is concerned, there's a limit to which I'd go. Thousand-dollar CPU's, $800 GPU's, and $600 motherboards are a bit too rich for my blood. However, I don't knock those that do because I understand that passion as I too buy some overly priced toys
My two cents.
in the second computer.
Related Discussions
