Water vs Air clarification
Water vs Air clarification
Hey guys, let me explain. With your i7-6700k at 4.7ghz and stable temps with the 212 Evo, you're already doing well. Water cooling helps by keeping temperatures lower, which lets you push your CPU further without overheating. An H100 could potentially allow even more overclocking, depending on its capabilities.
1) MAY overclock better, however there's no way to know without actually trying.
(I dislike liquid coolers unless absolutely necessary. They are far more prone to fail, and some have about a 5-year lifespan with the tubes. Pump noise can also be an issue to some people.)
2) I would NOT overclock any further though.
If we assume 4.7GHz is full core load, AND assume you can get up to 5.0GHz then what's the MAXIMUM gain (which won't apply to games). 50/47 =
6%!!
Yeah, if you put in a new cooler then you can get 6% maximum for my scenario which is likely about ZERO for gaming, and probably no more than 3% for other demanding workloads. Really NOT worth the effort.
3) You really get 4.7GHz with an EVO cooler? Wow...
If it's NOISIER...
I'm not sure about all the details, but I'll let someone else give a more thorough explanation. The H100 would cool down faster than the Hyper 212, which would help lower temperatures. This happens because heat is moved away from the CPU and released through the reservoir. Similarly, using a large air cooler like the Phanteks PH-tc14 could work well since it adds extra fins to remove heat. Just remember, even with a high-performance CPU, you can't always reach the highest core speeds.
It doesn't.
Improving coolers really helps with battling heat, that's the essence. On every CLC water cooler available with one or two fans (240/280mm rads), there exists a more affordable air cooler that surpasses it. Check out this video at the 23:00 point:
https://www.youtube.com/watch?v=PYKdKVxbnp8
In the clip, we observe the Noctua NH-D15 reaching 70C at 33 dbA. At the same time, the H100i maintains a CPU temperature 3C higher, yet it requires being 12 times louder—comparable to a vacuum cleaner.
The issue lies in CLCs breaking the first two guidelines of water cooling:
Rule #1 – Avoid mixing metals in a loop – CLCs employ copper water blocks and aluminum radiators, which is detrimental. Corrosion inhibitors can slow down this process but lose effectiveness after about 18 months, and CLCs don’t offer replacement options.
https://martinsliquidlab.wordpress.com/2...-explored/
Rule #2 – Ensure a minimum flow rate of 1.0 gpm – The H100i delivers only 0.11 gpm
Rule #3 – Custom loop products outperform CLCs in terms of flow and design. For example, the H240-X offers an all-in-one setup with a copper radiator and a pump for multiple blocks (RAM, graphics cards, etc.). It manages to be 9C cooler than the H100i while being just 20% noisier. Plus, it allows adding corrosion inhibitors and a reservoir, which the H100i lacks.
These tips come from an Asus overclocking video for Haswell with some additional insights. If your CPU doesn’t use AVX instructions, you can move up one category—Skylake is more efficient, so you might safely increase another level.
Up to 1.200v = Very Good Air Cooler (Hyper 212)
Up to 1.250v = Top Air Coolers (Phanteks PH-TC14-PE, Silver Arrow or Noctua DH14) … Dual 140mm CLC / AIO Cooler with 1500 rpm fans (Corsair H110)
Up to 1.275v = Extreme Speed Dual Fan CLC / AIO with 2700 rpm fans (too noisy for most users)
Up to 1.287v = Best air coolers (Cryorig R1 / Noctua DH-15)
Up to 1.300v = Swiftech AIOs (Swifteh H220-X / H240-X)
Up to 1.325v = Custom Loop with 15C Delta T (3 x 120mm / 140mm)
Up to 1.400 = Custom Loop with 10C Delta T (5 x 140mm or 6 x 120mm)
The * indicates a custom loop with water-cooled graphics cards assumed.
5. A note on voltage: when your CPU runs AVX instructions, voltage can spike by up to 0.13. I set my Vcore at 1.375 in BIOs and it breaks at 1.50 for short bursts when AVX is active. This raises a question about the load conditions during testing. P95 and AIDA scores are best avoided; using a benchmark like RoG Real (paired with tools like HWiNFO) will reveal actual voltage under AVX and simulate real multitasking loads that benchmarks miss. Even stable P95 performance at 24 hours might fail on Real Bench.
6. Your CPU's performance is limited, mainly due to AVX forcing higher voltages. During testing, we aim for temperatures below 80C under RB, which translates to mid to high 60s during gaming. "Under 80C" is quite high for regular use.
7. Two main constraints exist: temperature and voltage. A better cooler will help address the first issue. However, once your voltages—especially real ones—exceed 1.45, you should consider the long-term health of the CPU. I can't predict this definitively... we won't know until more time passes and failures become common.
8. My advice is to run a stress test tool that includes AVX (such as RoG RB). Pay close attention during the third test (there are four modules, each taking about 8 minutes) to observe the maximum voltage under AVX. Use this value as a benchmark for acceptable levels. If it's within range, you can push the cooler further and increase the overclock slightly.
Mikeandike:
Hey everyone, I was trying to understand something. I have an i7-6700k at 4.7ghz and 1.385 volts, and with a 212 Evo I keep temps under 80°C. My main question is, how does water cooling help with overclocking? In my setup, would an H100 push me further, or am I already at the limit?
A liquid cooler gives you a significantly bigger radiator surface area.
Of course, you can't attach that big radiator straight onto the CPU or motherboard.
You need to install it somewhere else—like on the case wall or top.
The challenge then is... how do you move the heat from the CPU to that larger radiator?
With a regular air cooler, you use rigid copper pipes, but if you mount the radiator elsewhere, you can't do that.
Enter liquid cooling. It just moves the heat energy from the CPU face to the radiator fins.
This bigger radiator has a much higher fin area, which lets it release more heat into the room.
That's why smaller 120mm liquid coolers don't always perform better—they're sometimes worse due to the fin area.
1) It might be possible to improve performance further, but we can't be sure without testing it first.
(Liquid coolers aren't my preference unless necessary. They tend to fail more often, with tubes lasting only about five years. The pump noise is another concern for some.)
2) I wouldn't push the overclock any higher.
Assuming 4.7GHz is at full core load and you can reach up to 5.0GHz, what's the biggest increase possible (excluding games)? 50/47 equals about 6%.
That would mean a maximum gain of around 6% in my case, which is likely near zero for gaming, and probably under 3% for more demanding tasks. It doesn't seem worth the effort.
3) You can still achieve 4.7GHz with an EVO cooler?
It's impressive, but if it makes the fan louder than desired (which would suggest high fan speed), I'd recommend another air cooler.
A good option could be a Noctua NH-U12 or better.
*In fact, I swapped the fan on my EVO with a $20 Noctua model and it was much quieter while maintaining the same cooling.
You can find more info here:*
http://noctua.at/en/which_fan_is_right_for_me
So it was the NF-F12.
This is the fan:*
https://www.amazon.com/Noctua-NF-F12-PWM...B00650P2ZC
You can generally replace it without changing other settings, and noise should decrease slightly.*
Clarifications ....
1. Pump noise, like fan noise **can** be a problem if you make poor choices. On CLCs, you don't have a choice tho.
2. Why overclock beyond 4.7 Ghz ? Because we can ! It's part of the fun.
https://www.youtube.com/watch?v=CmYDgncMhXw
Photonboy shared his thoughts on overclocking possibilities and cooling solutions. He mentioned he prefers not to overclock further, citing reliability concerns with liquid coolers and their short lifespan. He also noted that while a higher core speed might be achievable, the gains would likely be minimal for gaming purposes. He suggested considering a cooler upgrade if noise levels become an issue, recommending specific models like the Noctua NH-U12. He shared his experience of replacing a fan with a quieter alternative and provided links for further information.
Wow... that's a big difference... I've never seen a delid with a Hyper 212 before... the TIMs are as expensive as the cooler.
Check out the AS5s home page for details on curing and capacitance problems—like a 200-hour cure time longer than normal use and capacitance concerns. If you can apply conductive liquid inside the CPU, you shouldn't worry about those issues.
I'm still wondering what makes a "12-hour tough test" (including AVX).
I recently switched back to Air Cooling after using Custom loop Watercooling for five years.
Water cooling demands significantly more maintenance.
I installed the Noctua D15 with Dual Fans.
I upgraded my GPU to an MSI 1080 Gaming card.
Factual R5 Case
MX-4 Thermal Paste.
The computer runs quieter, is smaller, and maintains good cooling performance like my custom loop.
I didn't need to adjust my overclock at all from the stock 4.5Ghz voltage (2600k on i7).
Since I plan to upgrade soon, having a quiet, compact system is beneficial.