Corsair H110i GTX Performance Testing and Review

Page 3 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
This test confirms what martin tested with the H110i (0.11 gpm) tho in both cases, it's not experiencing the full resistance of the system as when the test is done, it's no longer closed and pumping thru block + rad.



But that's kinda like child proofing your house and extending your shelves from 4 feet off the ground to 6 feet off the ground. :) What does the extra 2 feet do your you witha 2 year old ?

With today's efficient GPUs, no one with a decent case / cooling arrangement id having performance curtailed by temperatures. The MSI 980 Ti for example starts to throttle at 85C and tops out at 75C at max overclock / max voltage according to TPU.

Using a CLC type GPU cooler might lower that to say 55C, but what does that for for ya ? Let's say that the stock cooler takes it down from 95 to 75 and the liquid cooler from 95 to 55. That means the single rad fan has to move twice as much heat via the air / metal contact as the stock cooler does w/ 2 fans. In order to do that, the air velocity thru the rad must be far greater than thru the air cooler and that means more noise... noise that carries outside the case as compared with the muffling that occurs on a stock cooler from reflections. And you are still left with a hot VRM which is the culprit that is going to limit your OC far before than GPU does.

This is why these new alternative OLCs from EK and Swiftech have so much promise. First off, you don't have to worry about 3 cooling systems (CPU, GPU, GPU VRM / Memory), you have just one. The pumps are powerful enough to handle all your components. You can cover a CPU and a hi load GPU with just one radiator, mounting it on top of the case, with the fans properly installed as intakes for greater cooling performance, creating positive pressure keeping dust out as well as hot exhaust air from PSU and the noise is funneled into the case where it's muted somewhat.

I can't wait for the Predator review, tho it's gonna have to really impress to justify a $60 performance premium over the H220-X.

H110 GTX @ $140 versus Swiftech H220-X @ $140 = "no brainer"

Either of those @ $140 versus $200 for the Predator ? Gotta be a lot more pluses (QD Fittings + ability to do 1 more fan in push pull) than minuses (no reservoir) to account for that price premium.
 
This test confirms what martin tested with the H110i (0.11 gpm) tho in both cases, it's not experiencing the full resistance of the system as when the test is done, it's no longer closed and pumping thru block + rad.
Click to expand...

It's pretty dismal, that's for sure.

As for the Predator, I am overly impressed with it just from unboxing and doing some initial photos. It is exceptionally well put together. It looks professionally made. The radiator is hefty and relatively thick. And...the entire unit holds just around 300mL - almost twice the total liquid volume what the Corsair H110i GTX held - which should tell you how much thermal volume the radiators are able to cool at any one time, by comparison.

The radiator has 2x G 1/4 ports with chrome caps, sealed with o-rings. The pump is 'listed' as a DDC 3.1 on the label. There is a very nifty power distribution block for fans+pump, and both can be PWM controlled, depending on how you wish to set it up.

I haven't started on thermal tests yet, but based on what is contained in the box, it looks to be incredibly promising.
 
Thermal mass does wonders to even out load peaks.

Five things I remember not liking ..... just from looking at it.

a) Physical design allows push / pull but they don't provide enough fan ports.
b) No reservoir
c) No 140mm design, I'm done with 120mm
d) Not sure I can support $200 price tag.
e) Hey I'm 60... I forgot the 5th one :)

I started a radiator volume thread over on OCN but I admit to not keeping it updated. Thot the community would have provided more input....some users were very helpful / enthusiastic but how many rads can each person buy :)

http://www.overclock.net/t/1493339/the-radiator-volume-thread

I did read that the pump is only 6 watts, same as the Swiftech ... when i think "DDC", I think 18 - 24 watts.

I do see that they have the fans installed upside down (exhaust) and the TPU review says it's a PITA to change.

Reviews (EK 240):

https://www.techpowerup.com/reviews/EKWB/Predator_240/9.html
http://www.techpowerup.com/reviews/Swiftech/H220-X/9.html

http://www.guru3d.com/articles_pages/ek_predator_240_aio_liquid_cooling_review,12.html

http://www.nikktech.com/main/articles/pc-hardware/cpu-cooling/liquid-cpu-coolers/5681-ekwb-ek-xlc-predator-240-pre-filled-cpu-xpandable-liquid-cooler-review?showall=&start=6

http://www.modders-inc.com/swiftech-h240-x-aio-cpu-cooler-review/5/

So far I am liking the Swiftech better but there's still a lot I don't know.... both have wins on the various sites which is understandable but that $60 is a 43% price increase and the Swiftech warranty is 50% longer .... the performance per dollar difference can be used in the graphic from top to bottom as it doesn't address water vs air vs expandability. But given the similarities, the difference really hits ya in the graph here.

https://tpucdn.com/reviews/EKWB/Predator_240/images/perf_dollar.gif

So I am anxious to see what ya come up with when letting these two go head to head .... like can ya go G-1/4 on the EK ?

The 1st one to come up with a 420 rad will get some attention as that will cover a hi end GFX card .... in fact... a combo discount $20 off w/ purchase of the company's GPU water block would do a lot to create market share.
 
a) Yes, you could run push/pull, but you're short 1 fan port (3 PWM fan total) but you could run a pigtail or something
b) True. We will see how this impacts 'reassembly' when I'm done (will include in the write-up)
c) True, only 120mm, but it is still most common for cases with dual fan setup; newer cases are more flexible with larger fans
d) $200 seems steep, but when you compare a custom loop for CPU-only, it's not bad, especially considering it's designed to be a gateway watercooling loop - start out with this, gain knowledge, expand.
e) I'm 36 and I have those days all the time.

I've always wondered how I could do something similar, then it kind of struck me that this would be very difficult to ensure everything is tested in the same way, unless they wanted to ship me radiators to test, and ship back. Tedious. I'd even be willing to do this kind of testing and data collection with an agreement to return to a vendor upon completion with my time and shipping reimbursed.
 
a) which kills an advantage it cuda had over the Swiftech which also allows only 3 fans because of a physical limitation ... but the PWM controller does 8 fans which is a big +

b) "Being a worry wart" ... I like the visual indicator. They got the idea here, maybe we see one in Predator 2
http://www.tomshardware.com/news/ekwb-ek-spin-res-3,30475.html

c) I think 120mm is more popular, yes, but only because of installed base ... 140mm gives way more bang for the buck and i think more new builds go this route ... 480 costs way more for rad than a 420 but have same cooling performance and yes save on one fan to boot.

d) I am speaking only in reference to the competition at $140.... Swiftech costs ya 46.67 per year of guaranteed usage (3 year warranty) ... EK costs $100 per year w/ 2 year warranty.

And yes, identical setups is required for testing but volume ? I kinda figured everyone had the same one of these :)

51XVPV836HL._SX466_.jpg


Was surprised about the variances I saw from users, sometimes 100 ml or more..
 
Yeah, it would have been nice to get 1 more fan header, but my guess is that distro block was engineer for both the 240 and the 360 - and assuming push/pull wouldn't be a highly chosen option. Plus, the rad itself is 40mm wide, plus 25mm fans, you're looking at a 65mm wide footprint out of the box...you'd be sitting at almost 10cm thick (95mm) with push/pull fans.

Funny you say that about the EK spin res - they actually have some pages in the manual on alternative loop setups, and that res (along with a cylinder res) are show as suggestions. Also shows the steps to add a GPU into the mix.

As for the performance comparison between the EK and the Swiftech - I'm expecting both to perform about the same at lower thermal loads and the EK to pull away at higher thermal loads. Either way, I would take either over any Corsair unit.

Once I get done with this EK Predator, I'll give any recommendations like I did with the Corsair.
 
That was the surprise c... Swiftech took 7 outta 8 performance tests in the TPU review. Then again, EK won in one of the other reviews.

-OC w/ Prime = Swiftech by 2C*
-OC w/ AIDA 64 = EK by 1C
-OC w/ AIDA 64 FPU = Tie

* Interesting enough, installed as push (intake) it picked up 1C. Given the difficulty of flipping the fans, ya wud think later editions will put the the normal way as intakes since it has better performance. As I recall martin showed fans work better in push up to 1650 rm and better in pull at 1850 rpm and up. So despite being less efficient at 2200 rpm in push, using them as intakes improved performance.

The EK tho was a rather significant 6 dbA louder which is expected with fans spinning 400 rpm faster. I think if you push the CPU up to 4.6 Ghz or more, the EK would therefore do better thermally, but I think a key data point will be both at 1800 rpm and, if nit equal, both at the same dbA.

I think a good move would be to put Tees on the inlet and outlet of the rad and install thermal sensors on the branch
http://www.frozencpu.com/products/10373/ex-tub-620/Bitspower_G_14_Temperature_Sensor_Stop_Fitting_-_Matte_Black_BP-MBWP-CT.html?tl=g30c229s579

radiator
|
|
|---- Thermal sensor
|
|
|
to pump / block

Might even do the same on inlet / outlet to block for sheetz and giggles ... no useful purpose really in single block loop as temps won't chnage after 8" of tubing but the reeven 6 eyes has 6 temp displays and with 2 at the rad, 1 ambient, 1 case interior temp, I can't think of how to use the other two :)
http://www.frozencpu.com/products/25172/bus-393/Reeven_Six_Eyes_II_30W_6_Channel_Fan_Controller_Temp_Sensor_-_Black_RFC-02.html


 
I have 7 waterproof Dallas one wire sensors that I have connected to a CrystalFontz 633 (I have 10 total). I currently have 2 in a tee exactly like you have diagrammed (fixed with epoxy and not interfering with flow) and then I drop at least one into the reservoir. The others are used and averaged for ambient air temps. Data logs out every second to a CSV file which I convert over to Excel to do calculations on each thermal load for both push and pull.
 
Pressure loss thru a T is minimal. Friction coeffiecient is 0.54 (x V^2 / 2G)

If I can remember my notes from days as a fluids instructor at 1.0 gpm, we have a flow of about 2.9 feet per second or about 0.07 feet of water or 0.03 psi. I wouldn't bother but I admire your attention to detail.

 
I like detail, especially when it comes to scientific methods. I know that given my setup and cost of equipment, my several hundred dollars worth of gear is nothing when it comes to several thousands or tens of thousands of dollars worth of equipment. But, if I can replicate my testing methods and identify potential areas where environmental variables can be eliminated, I should be able to collect data in a way that allows repeatable test scenarios for anything I put on the test bench (hopefully).

This detail oriented method is also the same reason I enjoy brewing beer on weekends - it requires repeatable methods and exacting times, temperatures and volumes.
 
Without repeatability, scientific investigation is meaningless. And I agree, if ya not in a position to quantify the impact of a decision, then it's best to go to the extreme to make sure a decision has 0 impact. If I kno that a rock weighs 375 pounds, I'll use a 500 pound hoist to lift it.... if I don't know how much it weighs, then i'll use the biggest mutha I can lay my hands on .

As for the beer, repeatability is important tho improvements sometimes come by way of mistakes. My niece's hubby just took home some awards at a recent brew competition and the one they served at their wedding was one of those. he told me what it was tho as I had sampled a few before he told me, I can't tell ya what it was. I have some samples in my fridge I gonna drink on their anniversary.
 
Sometimes you can get too scientific, and totally miss something very important, like, how well the individual water blocks perform actually mounted to a CPU?

The thermal load testing is OK, but we aren't cooling aquariums!

It all seems completely moot, and IMO a half completed review, if you never even mounted the CPU water block on any CPU at all?

The AIOs do cool CPUs and are designed to mount on CPUs so where is that data?

Secondly this is the Overclocking Section of Toms Hardware, so Overclocking data is missing as well?

CPU temperatures totally MIA.

 
You're missing the entire point. I don't need to mount to a CPU. I don't need to overclock a CPU. I can test exacting thermal loads to compare what ANY CPU could possibly load into the cooler. The testing methods that I am using allows me to account for any possible heat load these coolers could potentially see. For 'overclocking results' - this is about coolers, not specific CPUs or specific CPU overclocks. Thermal load in watts in various, increasing amounts accounts for overclocking; that's made clear in my testing methods. Otherwise, I would need 5-6 CPUs to try and replicate my load tests, and even then, it isn't as accurate because I'm still limited to testing and reporting based on MY exact CPU setups; not something that can be applied to everyone.

Every single one of these cooler reviews goes the route of "mount cooler, test cooler with CPU load test, overclock CPU, test cooler with CPU load test, give X-number of stars". Great, but this is not what I wanted to do. These tests and reviews are limited in knowing exactly what loads are being tested and dissipated by the coolers based on the 1 or maybe 2 setups they have available. This is where I am trying to draw the line of testing difference - I am removing the variables of any specific CPU speed or voltage, motherboard stability, RAM stability, wondering if 100% actual load is applied, etc. I have the tools to measure exact thermal load for each test.

You and I both know that the biggest problem with cooler reviews is that Joe User doesn't understand the difference in how thermal load and ambient temperatures account for cooling a specific CPU at a specific speed or thermal load. If I say that a i7 4790k running at 4.8 GHz at load reports temperatures of 55C, for example, someone is then going to say, "Great, but what about my i5 3570k running at 4.0GHz? What if I live in a hot climate? What if I use the +25% overclock in my BIOS?" ... you've now created a testing and overclocking review for only a cooler on a specific CPU because you cannot tell Joe User what to expect on his CPU. If you can TEACH Joe User how to account for his own CPU loads and provide him with a graph of data that represents a thermal curve, he can simply connect X,Y axis and find his expected cooling capability.

Of course, you're always welcome to do this overclocking testing and result reporting if you'd like to supplement. For someone who has been supportive of this testing to this point, I find it strange that you'd suddenly do a 180 and respond with this, and while I understand your point, my response is the entire reason why I did not go this route. The web is full of these kinds of reviews, why would Tom's want to add yet another when we could be the first to try and teach our members the how's and why's of cooling and show them how to determine what to expect in their own specific systems?
 
I am not missing any point rubix_1011, and you are side stepping the issue!

Do you really think I am the only one seeing this?

I am trying to get you to realize what I am saying before you test the others and have compromised the original shipped to you, integrity of the coolers.
 
I'd invite 'everyone who is seeing the issue' to include their opinion on the subject. What exactly would you like to see in these write-ups?

Compromised the integrity of the remaining coolers, how? The remaining coolers for this group are intended to be expanded.

Maybe I am not following you, here. There are literally dozens of write ups on the web about what you are talking about; so you are suggesting I just do the same as everyone else?
 


I am saying you should not exclude that direct mount CPU data but provide it as well, along with exact ambient room temperature the direct mount CPU test is conducted in.

Because you are really not even testing the water blocks, only the radiator cooling capability.

 
I understand your point, and I realize what you are saying. But this is exactly what is done on all other reviews, so I would just be providing the same information that exists so many other places on the web. Playing devil's advocate with myself, I can see the argument of 'we should provide that data here, as well' but at the same time, at the end of the day I'm only just providing data of that cooler based on my own, specific system that just represents what happened to me, but might not happen for someone else - I am then introducing not-repeatable results that would essentially have an asterisk that says:

(*Results seen may not be results or experience seen by anyone else. This is a generic application; installation and temperature results are not indicative of all systems)

In doing this, I have contradicted my own testing and results by introducing dozens of variables into a result for people to read and depend upon. They will overlook the load test results and skip right to "My CPU saw load temps of 55C using Prime95" and therefore undoing the teaching opportunity and understanding that each system is unique and different. I would also debate that far too many assumptions are made from reading reviews that return result data in this manner.
 


I think your thermal load testing is great for overclock estimation but IMO you should have done your own direct CPU mount tests before that, even before taking any cooler apart, so your test not only covers the CPU direct mount load testing basics but goes way beyond other reviews including your thermal load tests.

IMO since the H110i GTX was not expandable it should have been isolated in it's own test group, and not disassembled.

But that's water under the bridge now, the expandable lines to be tested should IMO be thoroughly tested before disassembling with direct CPU mount testing.

Once you take it apart you have compromised the initial factory shipped state!

It would be interesting to see how the expandable line fares running the competitions water block, but Oh My, that would require mounting it to a CPU.

 


I believe I said something like compromised the shipped factory integrity, you know the sealed integrity that is shipped out of the factory door, that anyone buying the cooler you review could expect to receive when they buy one.

 
It would also be interesting to see how the AIO fares against your custom water cooling setup?

Because you are not just an AIO CPU cooler tester, you are a custom water cooler, you bring so much more to the table than some of the other testers do.

That could bring some illumination to some Toms users that are trying to decide going high end AIO or a custom build.
 
It would be interesting to see how the expandable line fares running the competitions water block, but Oh My, that would require mounting it to a CPU.
Click to expand...

I have decided to not do any cross-vendor testing on the units. If it was "a bad idea" as you mentioned with the H110i and using, say, a DDC pump just to see what happens, doing something like this is also a bad idea - you're testing a scenario that doesn't offer value, so to speak. I have since accepted your advice in this regard.

If you would prefer to supplement the data yourself, please feel free to do so on your own accord. I also don't appreciate the mockery and sarcasm; it does nothing for the debate overall, but that is your choice.

It would also be interesting to see how the AIO fares against your custom water cooling setup?

Because you are not just an AIO CPU cooler tester, you are a custom water cooler, you bring so much more to the table than some of the other testers do.
Click to expand...

Considering I have a full cover GPU block in my loop, the only way to compare load results is to simply add the TDP load totals from my system and run it against the expandable loops. This is why I choose to test how I do - I don't have to build or buy components just to test with. I can use the engineering specs of components and load the loop based on this value total.

This was something I had planned to do with the XLC coolers - simply because that is what they are marketed as.
 


So actually you ask for input to give reasons to disregard it.

Why should I do the testing, when you have the hardware?

My Bad on the mockery and sarcasm, it is good for you I actually like you! :)

We have a completely different point of view, and my mockery and sarcasm comes from testing AIO CPU coolers that are not actually mounted to a CPU?

To me that's preposterous, it's like buying a vehicle on stats alone and never test driving it?

That is absolutely not going to happen in my home!



 
You are incorrect with your assumptions about me being open for input - it is just that I disagree with the methodology you've mentioned and choose to continue as I have.

I have the right to disagree and defend my own system without having to concede to all outside ideas. Asking for input does not automatically mean that I am also required to do what is suggested, so please remember this when someone mentions they are open to input. I have elected to take some of your advice but I have also respectfully declined some, as well. The issue here is understanding that both are done with the same level of professionalism, but it is interpreted differently based on the decision.

Like I have stated, I chose to test in the way that I have so that I can repeat all tests in the same manner.
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom