Corsair H110i GTX Performance Testing and Review

[Math Geek]

as a noob looking to get into watercooling as either an AIO or possibly custom down the road, i liked the data but also see the some of what ryan is saying. in a perfect world if this testing was done just for me (by me for my own use even) i'd run it on a cpu at stock settings to start with to get a baseline. then all the sweet testing you did after for the wonderful thermal load graphs.

as someone trying to educate users (like me) this baseline readings of the cooler and what it does on a specific cpu in a specific setting is part of the knowledge. knowing the rough tdp the cpu brings in, paired with the then exact delta numbers you are able to measure once it is broken down and controlled better helps paint the whole picture. some ending analysis of what you were able to read on a cpu and then helping me see how the delta graph you made relates to and can be extended to any cpu is the final exam of the overall lesson.

does this make sense? as thorough as all you did is, it is partially missing the connection to an actual result on a cpu. i know in your mind the connection is there, but as a student, the aha moment of comparing the graph to what we measured on the cpu and seeing how some of the variables effect that delta graph is the icing on the cake. i love the graph and feel i understand it's purpose, but the other variables that i now know mean a lot to actual real world performance could be explained a lot with a real world test and some insight on what is happening. you are the master and sometimes you have to do some initial hand holding and spoon feeding to get the lesson to stick. in science class, most theory is taught with a specific single experiment and then expanded from there to include the theory and effects represented by that single experiment.

either way i love it but i think ryan has a point, though he may not have explained it as well as he may have. remember i am actually an experienced teacher and it took me years to figure out how to take all the vast deep math knowledge i have and present it to a total novice in a useful digestible way. so i offer some insight from that perspective to you

 

[rubix_1011]

That makes sense and while I think I completely grasp what he was saying, coming from a different point of view does offer some new perspective.

Let me see what I can potentially do in the interim; I might have to follow up with additional content and testing.

 

[JackNaylorPE]

I am going to agree and disagree with both sides of this argument. On one hand this is an AIO, one of the things peeps will want from a review is how it compares "as a package" with others that come "as a package". In addition, as has been shown in detail @ benchhmarkreviews.com, the "frostytech approach" of testing CPUs coolers on a "synthetic heat block" is pretty much useless as the heat signature of a particular CPU, will present different conditions to the block that is on a CPU. What rocks on one platform may be so so on another. This is show never more clearly than in xtremesystems CPU block reviews where there is a different "best" for almost every platform:

http://www.xtremesystems.org/forums/showthread.php?285753-CPU-water-blocks-roundup

Here we see ...

- the top 3 water blocks for 3930k being the Koolance 380i, AC cuplex kyros pro and Koolance 370
- the top 3 water blocks for 4770k being the EK Supremacy J2, EX Supreme HF and EK Supremacy J1
- the top 3 water blocks for 4930k being the EK Supremacy J1, EX Supreme HF and EK Supremacy J2

So the different heat signatures of each CPU do in fact affect cooling performance.

On the other hand, these AIOs are a package and just as important as accounting for the above, is accounting for variables in the individual components, so here I agree with Rubix as to providing the data on each component. For example, let's say I want to pick the best cooler available "for me" on a 4790k w/ an anticipated 135 watt load. I want the best performance possible at no more than 40 dbA. We have the following:

Predator 240 under 150 watt load gives me 70.5C
Swiftech H220-X under 150 watt load gives me 70C

At the surface, the Swiftech seems like the better choice....

... but wait, the Swiftech has an advantage because it's using the fans as intakes, if I get the special tool from EK and flip the fans, I pick up 1C and now the Predator is better....

... but wait, the fans on the Predator are spinning at 2200 rpm and the Swiftech's at 1800 rpm and therefore are much quieter. If I make them the same speed, the Swiftech is now better....

... but wait, given the price difference, I could add different fans ... you see where this is going.

As this is an "expandable system" ... we can not only add blocks but also radiators, what radiator to I choose for my 2nd Rad ? How will each pump handle it ?

So from my PoV, I agree with Rubix in that creating repeatable scenarios allows one to weigh the relative strength of the individual parts so that I can perhaps consider modifications or better plan my expansion .... but I agree with Ryan in that in that many folks will just want to know about "what they bought" as a whole.

The other factor to consider is how much effort should be put into a $140-$200 component ? But, from a user perspective, the relative performance if the two blocks could effect which one you buy. Is it really realistic to do tests on 5930k, 4930k, 3930k, 6700k, 4790k, 3770k, 2700k and add in a few more for AMD...?

Picking one may be misleading because of heat signature variation or simply because one does better at 230 watts than it does at 88 watts. But I think if I had a choice, one or the other, I'd rather see results at say 90, 130, 180, 230 watts, then just 1 test on a particular CPU.

Ideally, I'd like to see both .... use a heat block to create the wattage and test the ability of the cooling system, and then test on a 6700k and 5830k to evaluate the effectiveness on the two major platforms. You'd wind up with an evaluation that addressed the relative strength of each component as well as the entire system.

 

[JackNaylorPE]

For a little distraction from the philosophical approach... thot I'd try and list a perfunctory spec list, well a start anyway:


OLC AIO
Manufacturer: Joe's Badarse Coolers
Model Name: JBC 2-120-1800
Included in Box: Whole Bunch of Good Stuff
Price: $149.99
Warranty: 10 years

Radiator
Model No.: JBC 45-120-2-12
Configuration: 2 x 120
Material: Brass tubes, louvered copper fins
Dimensions: 250mm x 125mm x 45mm
Connection Type: G-1/4 Barb Type
Fittings: Swivel 90° elbows, 3/8" (10mm) OD
Removeable (tubes / fittings): Yes
Hardware: Standard: (8) coarse thread fans screws Alternate: (8) 6-32 x 1 3/16 (30mm)  Philips screws  
FPI: 12
Design: Split Fin
Screw Protectors: No
Connection Ports: 2
Bleed Ports: 1

Pump:
Model No.: JBC-1.5-6
Flow Rate: 0.50 - 1.25 gpm
Pressure Head: 4.2 psi max
Speed Control: PWM
Speed Range: 1200 ~ 4500 RPM
Voltage: 12 VDC
Nominal Power: 12W
Redundant Fail Safe Calls you on smartphone if there's a problem and shuts down system
Bearing Type: Ceramic shaft, Unobtainium bearing
Power connector: SATA
RPM & PWM connector: Mini 4-pin
MTBF: 100,000 hours
ROHS Compliant: Yes

Fans:
Model No.: JBC-120-SP
Quantity: 2
Mounting Direction: Intake
Dimensions: 120mm x 120mm x 25mm
Speed Control: PWM
Speed Range: 600 - 2400  RPM
Airflow rRange: 20 ~ 80 CFM
Static Pressure Range: 0.50 ~ 2.50 mmH20
Noise Level Range: 15 - 35 dbA
Nominal Voltage: 12 v
Nominal Current: 0.2 - 1.0 A
Bearing Type: Z-axis Modified Rifle Bearing
Connector: Mini 4-pin
MTBF: 60,000 hours
ROHS Compliant: Yes


Water Block:
Model No.: JBC-I-1151
Base Plate Material: C110 copper
Cooling Engine: 0.25mm x 0.25mm  micro-channel
Connection Type: G-1/4 Barb Type
Fittings: Swivel 90° elbows, 3/8" (10mm) OD
Removeable (tubes / fittings): Yes
Default Hardware: For Intel LGA 1150, 1155, 1156 (pre-installed by default) Back-plate (4) Spring loaded screws & washers
Alternate WorkStation Hardware: For Intel LGA 2011  (4) Spring loaded screws & washers
Alternate AMD Hardware: For AMD AM2, AM3, FM1,FM2, 939 (2) Pre-assembled brackets & spring loaded screws, Bracket installation screws
Optional Hardware: Optional back-plate available for Intel legacy socket 775 and 1366

Tubing & Coolant
Material: PVC
Color: Black
Dimensions: 5/8" x 3/8" (16/10mm)
Coolant: Pre-filled with Tequila
No. of Clamps: 4
Clamp Material Black anodized aluminum
PWM Splitters: 1
PWM Ports 4
RPM Signal Port: Channel 1
Power Connector: SATA
PWM / RPM connector Mini 4-pin
Installation Hardware Peel-off sticker, w/ my picure on it,  mounting screws & nuts
ROHS Comliant: Yes
TIM Provided: 1.5g

Reservoir:
Provided: No
Material: Acrilic w/ clear window
Level Indicator: Yes, Visual
Flow Indicator: Yes, Paddlewheel
Critters: (2) Seahorses

Cool Stuff:
T-Shirts (1) "Liquid Control Phreak"
Door Hanger: (1) "Shhh ... it's quiet in here, were water cooled"
Extras: Real girl's phone number

 

[rubix_1011]

That was well worth reading in its entirety.

 

[JackNaylorPE]

Worth reading ? .... Are you trying to solve an insomnia problem ?

After proof reading i wanted to take a nap

 

[LoneWolf_53]

For what it's worth I realize this is an old thread, however I thought I'd add an observation that may be helpful to some.
I am running three of these H110i GTX coolers all cooling Xeon X5680 CPU's (yes I know they are older but they work).
These computers run a distributed computing project thus are on 24/7/365 under full load.
After running these for about a year I have noted that the cooling efficiency of these things has dramatically degraded.
The CPU's that were initially cooled to around 60°C while under full load with ambient room temp in the mid 20's, are all now showing 65-70°C and that with the CPU load reduced to 50% from the full 100% before.

I have read in several places other uses complaining of similar, and or overheating issues.

Corsair says these are closed systems that do not require maintenance but I'm a bit suspicious of this claim.

I know the pumps are all running fine, and good quality thermal paste is in play, so the only thing that leaves is a loss of coolant being the culprit.

I have to rebuild one of the computers in the near future and will be attempting to do what someone else did which is remove the copper bottom plate and fill the system up again, which apparently for him returned the performance to the great cooling capabilities seen when the unit was new.

Just out of curiosity has anyone else opened one of these up to top up the fluid?

 

[rubix_1011]

You should never have to fill these, but I know that there are many people who have encountered the same thing. My recommendation would be disconnect tubing from the radiator to fill, but please realize this will void your product warranty if you choose to do so. Actually, 'opening' the unit at all voids it, but it sounds like you're OK with that.

 

[LoneWolf_53]

Thanks for your input.

Yes I'm aware opening up will void the warranty but I'm OK with that. I just want to regain the performance they are capable of.