Corsair H115i Pro RGB on i7 8700k vs high end air cooler

Hi guys,

I'm thinking about migrating to water cooling from air.

Given the fact there's a some amount of risk related to AIO's malfuctions and/or associated possible leaks compared to the much simpler designed aircoolers,I would like to know prior any purchase if there's actually any significant performance gain in terms of temperatures compared to the cooling I have now (Thermalright IB-E EXTREME in a 3 fan configuration) to justify the associated fuss and cost.

For anyone not familiar with my CPU air cooler,it's almost identical with Noctua's NH-D15 and it's slightly behind Noctua's cooler in terms of performance only dy a couple of degrees C allthough it's significantly louder due to the louder fans.

I'm after Corsair's latest H115i Pro Rgb to cool my i7 8700k overclocked @5.0GHz and I'll greatly appreciate any info from anyone using the specific AIO to cool the same CPU.All reviews out there about this AIO are with almost every other CPU except 8700k.Given the cpecific characteristics of this CPU I think the reviews results aren't as accurate I would need to make my mind up and help me on my decision.

Right now I can overclock at 5.0GHz stable with 1.35Vcore with temps 62-65 Delta T (above ambient) while on idle it hovers around 8-10 Delta T.

Even though I'm good for now,when ambient heat start to rise towards Summer it will be impossible to maintain such settings with ambient anywhere above 25-28 C (now it's around 20 C).Do you think with H115i Pro RGB I'm gonna have considerably better results?

Any help will be greatly appreciated!

Many thanks.

Hey,

In my opinion, and from what I have read, when comparing AIO coolers to a top shelf air cooler like the Noctua NH-D15, most of the time there isn't really 'that' much difference in the cooling result that you will get. You may gain a few degrees at the top end with most of the high RPM AIOs - around 4C degrees give or take - but then the new H115i PRO is more about silence than top end cooling. Most likely you will end up with a very similar result to what you would get with a high end air cooler.

The advantages of an AIO are aesthetics, and fitment. I actually just purchased an H115i PRO (which I still need to install) to replace my failing H110i GTX, so I researched many cooling options. I really considered going back to air cooling but I have four DIMMs of G.Skill Trident Z RAM and the air coolers that I would really be after will not play nice with said RAM in terms of fitment, so in the end I settled with the 115 PRO. It's supposed to be quiet and provide good cooling. It fits with my RAM setup and doesn't look out of place aesthetically inside my Phanteks Enthoo Evolv case.

Maybe someone else can chime and give some additional info.

Another consideration is overall thermal management.

Consider this:

With an air cooler, waste heat from the CPU is dumped into the case. This is in addition to the heat from other components. The result is that your interior temperature may be higher, giving you challenges with other components.

In most installations, radiators are configured as exhaust. In this case, waste heat is dumped directly outside the case, keeping the interior temperature cooler. However, even in cases where the radiator is configured as intake, the exhausted air from the radiator will be around the same temperature (+/- 1-3C) as the coolant. This is far easier to manage and dissipate before it becomes a problem than 60-70C air from an air cooler. It also takes longer for all of the heat from the radiator to be exhausted (liquid has a higher specific heat than air) ... this may seem like a bad thing but it actually makes it easier to manage and provides a more stable thermal environment.

What stress test did you run for the +62-65C value? I will see if I can give a corresponding value for my 8700K, however it is delidded and that will have some influence.

I can say this, my coolant delta with a H115i Pro (280mm) is silly while set at 5.0/1.30v. It runs +4C over ambient with nominal fan speed. My typical peak load coolant delta is +6C and some things are less. Peak CPU core temps are never go higher than 50-52C in a 20C room.

It is possible the delid will overshadow any meaningful comparison, but I can offer this from the last time I did a full air tower vs 280mm radiator test. My old 5820K was a troublesome one and I usually ran it around 4.3GHz and 1.30V in the early X99 days. That corresponded to a 160W draw at full load, not much different than a 8700K at high marks. I did extensive testing with a Corsair H110 and a Dark Rock Pro 3. The H110 was a clear winner with a consistent -10C advantage in peak and average core temps across all load types. I would run the Dark Rock fans hard too. It just couldn't keep up.

If you want to get into things like purity of sound, etc., an air tower may still have strong appeal. There is no pump and reliability is extremely high. However, it is going to be a limiting factor when it gets hot. Same problem with the Dark Rock and I couldn't justify running 4.4 in the Summer with it. As mentioned above, RAM height becomes an issue and once I went to the tall Dominators the air towers were doomed. In addition, I can take a AIO down, clean, and change the fans in 5-10 minutes. Putting that Dark Rock in took hours and involved a lot of swearing. Your case and layout may have the final say in the matter, but I never ever think about putting an air tower back in any more, even when I do have an AIO issue.

Hey guys,

many thanks for your input,your info are quite valuable and will take it into consideration to make my mind up.

@c-attack,

I'm using Intel's Burn test V2 stress test and the values are after 20 mins/half an hour on "Maximum" Stress Level.It would be great if I could get a glance of the same CPU readings although yours is delidded as you already mentioned and it will be at least 8-10 C cooler hands down only because of this.A short stress test I made a while ago today showed max 30-31 C at idle (ambient at 20) and during 100% load at about max 84-86 with the same ambient.Needles to say the 3 CPU fans was at max RPM with all the associated noise (48dbA due to the Thermalright TY-143 140mm fans @2500max RPM)

At this point I would like to say that I don't bother for 1000% and "Absolute" stability as I my rig's most heavy usage and loads is while gaming and the specific games I play stress the CPU up to 65-80% max with everything maxed out inside gaming.Therefore I'm perfectly aware that maybe using other more demanding stress tests than Intel's Burn Test like Prime95,Handbrake,Realbench or AIDA64 would be quite probable for my system not to pass but I don't bother much as during several hours of gaming I'm experiencing 100% stability and that's perfectly fine with me.

One other thing is this:What if replacing the stock H115i Pro fans with stronger ones in terms of Static Pressure and Airflow alike like Noctua's significantly more powerful NF-A14 Industrial PPC-3000 or NF-A14 Industrial PPC-2000?The first delivers almost 10 times more Static Pressure and triple the Airflow while the "2000" delivers quadruple the stock fan's Static Pressure and double the Airflow.Noise will eventually be more,especially with the "3000" but wouldn't AIO's performance be considerably better than with its quite silent but rather weaker fans compared to the Noctua ones?

Nick.

One other thing is this:What if replacing the stock H115i Pro fans with stronger ones in terms of Static Pressure and Airflow alike like Noctua's significantly more powerful NF-A14 Industrial PPC-3000 or NF-A14 Industrial PPC-2000?The first delivers almost 10 times more Static Pressure and triple the Airflow while the "2000" delivers quadruple the stock fan's Static Pressure and double the Airflow.Noise will eventually be more,especially with the "3000" but wouldn't AIO's performance be considerably better than with its quite silent but rather weaker fans compared to the Noctua ones?

You can certainly throw on the A14i 2000 PWM version. I would suggest you not take up the 3000 rpm version for anything except sport overclock benchmarking where you are literally at the limit and 1C might matter. For comparison, I have the ML-RGB 140s on the H115i Pro right now. They are identical to the stock grey version, except for the LEDs. The highest coolant deltas I have seen are +6C for stress testing. My 4 hour MCI memory test with 12 instances and 99% utilization also topped out at +6C with a fixed 1100 rpm on the fans. Long multi-hour heavy gaming loads still peak out at 6C. So theoretically, the most I can ever reduce my end CPU temps with any fan at any speed is going to -6C. Sounds like an improvement, but of course 100% efficiency is never really possible. Knocking the delta down to 4C might be the best you could do at 2000 rpm. Maybe 3C with the 3000 rpm, but of course you actually have to run them at 3000 to get it. Like most things, you get diminishing returns as you extend to extremes. There is nothing wrong with picking up a 2000 rpm fan. The A14i and original ML140 Pro (black or single LED) are all good choices. I have 8 of them somewhere. However, you must run them at the elevated speeds to gain the benefits. A ML140 Pro (2000) with have the same performance as a ML140 RGB when both are at 1000 rpm.

I have not explored how much more I can reduce coolant temps on this 8700K. I have had lower RPM fans on from the start as it is Winter here. However, I did years of testing of with my Haswell-E platforms and 280mm radiators with scores of fans. Those set-ups delivered a peak wattage of around 205-215W (more than we can do on 8700K) and the sweet spot for fans was around 1500 rpm. That is certainly noticeable, but also kept the coolant delta to the same +6C under max load, long duration. In practical terms, I never needed more than 1000 for use and the sweet spot for noise was coincidentally right around 1200 rpm, where the max for the ML-RGB and ML-stock fans resides. ML fans are very nice for inverted installations. My black ML140 pro models can be run at 1100 before they really become noticeable. That is a very subjective value, but the A14i usually gains my attention about 200 rpm earlier. It is after all an "industrial fan" and that is how I would characterize the underlying tone. I like both fans to the point where the decision about which one to use comes down to the case and build. I love the ML140 black in my Air 740. The A14i in the white Air 540. Just a square/rounded thing. The Chromax 'new release' black models are also an alternative. Besides the colored corners, it is a 1500 rpm max fan. I am curious to know if it uses the same motor as the standard A14 instead of the A14 industrial. That would lean out the noise spectrum even further.

The 3000 rpm A14i is the one of only two fans I have ever sold on within a week. Normally I trade, loan, or stockpile these things. That one was just too much. Note it has an 800 rpm minimum run speed and the current draw is drastically larger than the 2000 rpm model. The H115i Pro can handle that draw on its fan controller with room to spare. NEVER plug two A14i 3000 into a splitter on one motherboard header. You will exceed the 1.0A limit by some margin.

I'll see if I can load up IBT this afternoon. I just bumped my memory up to 3733 and need to sort that first. Frankly, if you are hitting the 80s on air with IBT, I would say you are OK. No normal program should approach that level of CPU hardship.

EDIT: So I am not sure if this will make you happy or sad. My IBT Max run was rather favorable. That was the first time I have run it on this machine. I'm on 5.0 GHz, 1.30v adaptive and tuned well, 17.5C ambient. On the first cycle I had AIDA up to line graph (and thus Link closed) and forgot to max the fans. Core and package temps ran right at 53-55C with +6C coolant delta. On the second and remaining passes, I closed AIDA, put Link back up, and maxed the ML-RGB at their radiator restricted 1150 rpm. Knocked about 1.5C off the coolant and the resulting cores temps were steady 50-53. Surely, the overwhelming majority of the core temp differences is down to the delid. We can add in another 3-4C for the ambient temp difference, but that is still about a 30C gap. You don't get a 30C gain from a delid, so a portion of that is the H115i Pro vs the air tower. I would estimate it to be about the same 10C I previously saw at similar wattage on the X99 platform. That 10C may be worth something come Summertime. Also note I got 99% of the same performance with the fans following their preset curve in the 700-900 range versus the max 1150. However, I will say this regarding fan ranges. Nearly every fan becomes noticeable louder as you pass the 90% threshold. These fans are rather diffuse in sound at 900, but very clearly heard at 1150, typical for any fan with a 1200 max. However, this does mean there may be additional utility in a 1500 or 2000 rpm fan. I know my ML140 Pro is just a bit softer at 1100 than this. So in addition to a small performance advantage with slightly higher speeds, it is possible you may gain some noise reduction in the middle ground if you like to operate near the 1000-1100 rpm range.

C-attack,

for Noctua's "Industrial" fans I was already aware of the differences between 3000 vs 2000 however after your last post I made my mind up at least for which fans are the best for this AIO.

As far as concern the whole idea replacing my present twin tower air cooler with H115I Pro, I think I'll follow that road finally as your stress test results are more than I would need to know in order to turn to CPU water cooling.Besides this, I also decided that a deliding will be unfortunately inevitable if I really want to stay cool during Summertime with ambient room temperatures hovering around 33-34 most of the time and some times even worse and that's for at least 4 months period here.I think a combination of both AIO+delid is the best possible option.Otherwise I must back down my Vcore and hence my maximum attainable CPU clock and I definitely don't want that.

Many thanks everyone and especially c-attack.Your info were more than enough to what I would need to know!

Best regards,

Nick.