Jump to content
Corsair Community

h100v2: two issues


garyd9

Recommended Posts

A quick background: I haven't done any type of water cooling for at least 12 years. (The last time I did water cooling, there were no such thing as "kits" for it, there were no IHS's on the chips, etc.) That being said, I picked up a h100i v2 the other day to try and overcome some space issues I've been having.

 

However, I'm noticing two oddities that I thought I'd ask about.

 

First, when my machine is idle, I'm noticing that the H100i v2 water block/pump/water temp is always at least 2 degrees HOTTER than my actual core temps. For example, while I write this, Corsair Link is showing the H100i v2 temp as 28.7C. My core temps (reported by the i7-6700k) are 23C / 23C / 25C / 26C. My Asus m/b is reporting a CPU temp of 26C. The machine has been idle for hours. This seems backwards to me: it seems that the water temp should always be LOWER than the core temps (else the water block will heat the core.) What is Corsair using to measure temps (and is it calibrated whatsoever?) Is there a way to calibrate (or offset) the block temp reading?

 

Second, I'm wondering if Corsair's mounting system is all that great. When I stress this processor for a long period of time (true linpack running on 24GB memory blocks for over 30 minutes) using default voltage/speeds, the processor will go up to 72-75C (depending on ambient), but the water/pump/block temp never gets over 35C (and it takes at least 30 min to get to 35C.) Once it gets to about 35C, the water temp stops going up.

 

Raising fan and/or pump speeds have NO impact whatsoever on either the core temps or the block/water temps (but both have an impact on the noise level, obviously.)

 

From my experience, this seems to indicate that either the temp measurement in the block is completely useless, or that there's poor thermal contact between the processor and the water block. Previously, this same CPU was used with Noctua NH-U14S, and wouldn't get over 60C under the same voltage/speed conditions.

 

My suspicion, based on reading various forums, is that Corsair has tried to "idiot proof" the waterblock mounting, and has erred on the side of too little contact instead of crushing something. When mounting the block, after threading the spacers to the back piece, there was a LOT of wiggle. I realize that some have stated that "wiggle is okay", but with no springs to ensure good contact, how much wiggle is too much?

 

Are 50% temp differences normal with this water block? Should I try to remount the block with NH-1 and (at my own risk) use washers to force additional pressure between the block/IHS?

 

Thanks

Gary

Link to comment
Share on other sites

You would think the cores would match the coolant temperature step for step when in a true idle state, but remember they are not submerged in liquid and heat is transferred through the CPU interior, heat shield, cold plate, and then into the stream. The cores themselves are far too small to have a direct thermal relationship with the coolant temperature. Even when the cores are resting at 1000 MHz or whatever, the CPU is producing some measure of heat and the cumulative heat from the CPU heat shield is transferred into the coolant. They heat the local area, which then hopefully transfers to the shield, then the cold plate, etc., but there will always be thermal differences between the individual pieces in the chain. How does your coolant temp in this state compare to the "CPU" or CPU package temps? Generally the "CPU temp" will mirror the coolant temp when at idle. Also, the core temps are never really at rest. It is a very dynamic value and you have to be careful making assessments using it or assuming the motherboard is able to report such a dynamic variable to scientific standards. It certainly will change faster than the usual 1 second polling rate (or worse) most programs use to monitor temperatures.

 

I understand your concern and I have been using a X62 for about a month. It has similar behavior, only worse. I can wake it from sleep in cold 18C room and the damn coolant temp reads 27C in the 1 second it takes for the monitor to turn on. Obviously the CPU did not warm 500-750ml of coolant by 9C in 1 second. In this case, I suspect it has a lot to do with the location of the coolant temperature sensor. I would say my probe is too close to the contact plate within the pump unit. Due to the design of the cooler, the temp probe is going to be inside that pump housing somewhere. I think the Corsair one is better placed, but it is still going to be on the outflow side after the heat pickup and so I don't think you should consider it an average coolant temperature or use it in that way. I still understand and it annoys me a little as I try to become accustomed to its actual range and behavior versus my expected range.

 

 

As for part 2, the quick way to check would be to press down with two fingers on the pump while under load and see if the CPU temps drop a peg or two. They probably will not and the issue is likely related to something else. You are never going to get a cooler that has perfect thermal transfer efficiency. On any modern CPU, you can heat it up faster than you can transfer heat out, through the IHS, TIM, and cold plate. This is particularly true of the recent 4 core models. The CPU itself will always be the limitation. You know how the cooling system works. Some amount of heat transferred into the water stream, heat dispersed as it passes through radiator channels, water returns to pump head for another trip. So it would make sense if the coolant heats up when the CPU puts more heat into the stream than the radiator and fans can remove. However, we again have the issue of where the coolant temp probe is located. It is going to give you a temperature of coolant leaving the pump chamber. Obviously the temperature of the coolant right after the radiator is going to be lower. How much lower? No idea. Does the temp probe measure actual exit water temp or is it assessing a mixing of the incoming colder water with the exiting warmer water? Regardless, you need to use the coolant temperature as a comparative value and not an absolute one. If my starting coolant temp was 30C and the load value is 35C, that gives me a rough idea of how much heat is in the pump chamber. What this is really useful for is deciding how much fan speed you need. When there is a little change in coolant temp, you don't need more fan speed and that really is the advantage of liquid cooling.

 

This same process is going on in an air tower, but with no real ability to hold heat in the system, you need to get rid of of it quickly and fan speeds need to be responsive. With water cooling, you can just let it run. The water is doing most of the work and you generally do not need those higher fan speeds. The ability to remove heat from the coolant stream is not the limiting factor. My 5930K will draw about 200-210W in its current state. Even with that much wattage, I do not see a temperature reduction when using fan speeds above 1200 rpm on a 280mm cooler. A further reduction of 1-2C for another 400 rpm is a bad trade. The cooler is removing as much heat as is added to the system, but nevertheless the it cannot transfer heat away from the CPU faster than the voltage adds to it. Since your coolant temp hits the same plateau each time, that also suggests the cooler is working properly and is removing the maximum amount of heat. However, it must pass by that temp probe before it can dump its heat and I think you should consider that value a measure of the pump chamber thermals, not an average coolant temperature.

 

I am not sure what to make of the disparity between your H100i v2 temps and the U14. There isn't a 15C difference between any respectable type of cooler. So either there is some sort of problem with the H100i v2 or your BIOS settings are not exactly the same. Perhaps pressing down on the block will shed some light. You can try the washers. It was mostly older Asus boards (Z87/97) that had this issue, but not having your board I won't make any blanket statements about its thickness. Also, make sure you are not running these tests on Auto voltage. There is no guarantee you will be get the same results back each time and the voltage table on Skylake is a little too accommodating. Even if you intend to run the Intel default frequency, setting a specific adaptive voltage (or fixed for synthetic stress testing) is recommended. The auto voltage scales too high on these tests.

Link to comment
Share on other sites

Thank you for replying, @c-attack.

 

I'm guessing that the first issue is just that the pump is using a thermistor that isn't calibrated. I remember back when PC CPU temps were measured from a thermistor mounted on the motherboard, sticking up to make contact with the underside of the CPU (as there were no pins in the center of the CPU package in those days.) The things were usually wildly wrong and slow responding. However, they were at least fairly consistent within their limits. If it claimed 20C when an infrared showed 32C, you could be sure that the thermistor would claim ~40C when the infrared showed 52C.

 

Fortunately, the slow response time of a thermistor is perfectly acceptable for measuring liquid which is also slow to change. (Back when I did old school water cooling, I'd keep a thermistor in my water tank inlet to measure the water temps coming from the radiator...) I just wish there was a way to calibrate the software reading it (only to satisfy my OCD.)

 

Its the second issue that I find more concerning...

 

As you mentioned, there shouldn't be a 15C difference between the U14 and the H100iV2 on the cores. The processor voltage is exactly the same in both situations (as reported by HWInfo.) Ambient temps might be +/-1C. At full load, the processor vcore (not VID) is at 1.175 volts. (I won't leave any modern Asus board to default the voltage, as they seem to like to overvolt the processors in seemingly random ways.) This is the VCore as shown in HWinfo.

 

So, I'll move outside of the exact temps measured by the water pump, and instead I'll view them as an offset. Then I see this:

 

(all numbers are in Celsius and are offsets of the "x" and "y" representations of the idle temps.)

 

idle: core @ "x" degrees with the water at "y" degrees after >1 hour idle

 

(not system S3, but inactive/idle with all C states enabled. Some background crud going on, but HWInfo reports an average core temp over the span as within 1 degree of 'x')

 

load: core @ (x + ~50) with the water at (y + ~6) after > 30 min full load (and the water seems to stay around y+6 after that.)

 

The cores, as I'm sure you know, shoot up very quickly in temp. They immediately shoot to x+45 (which is normal) and then creep up to x+50 as the water heats up.

 

Comparing core to core and water to water: When the core temp jumps 50, the water temp (after given ample time to heat up) only goes up 6.

 

I know that temp increases aren't going to be linear, and in fact should be a curve due to the thermal properties of water, the anti-freeze, the amount of water, radiator, etc, etc. Regardless, I really expected a bigger increase in the water temp when viewed in relation to the increase in core temp. If the lack of water temp increase was due to cooling, then the processor would be cooler, so the only other explanation I can come up with is poor thermal transfer between the core and water block. (Normally, I might also consider poor contact between the physical CPU core and the IHS, but U14 results showed more reasonable heat transfer on this same processor.)

 

What might be useful is for you (and others) to give idle/load temps for the cores and water after a steady load of more than just a few minutes. I might be experiencing "normal" for this water cooler (in which case I'll return it, as it's performing as badly as a stock intel heatsink/fan with just as much noise.)

 

Of course, your 200 watt processor is going to generate a lot more heat than my 95 watts (and your 280mm radiator should dispel it quicker), but I'd be surprised if your water temps only went up 6C after your cores were up 50C for 30min.

 

Tonight, I'll remount the water block with some washers (using NH-1 TIM) and see if that makes a difference. I don't like the idea of pressing on the waterblock with my hand, as my arm/hand isn't a good consistent pressure source, and it might cause the mb to flex.

 

Take care

Gary

Link to comment
Share on other sites

(Back when I did old school water cooling, I'd keep a thermistor in my water tank inlet to measure the water temps coming from the radiator...) I just wish there was a way to calibrate the software reading it (only to satisfy my OCD.)

 

Agreed. In a custom loop, I would want one in the tank and one in the pump head so I can actually make some meaningful observations. I am probably heading back to that sooner, rather than later.

 

 

 

load: core @ (x + ~50) with the water at (y + ~6) after > 30 min full load (and the water seems to stay around y+6 after that.)

 

The cores, as I'm sure you know, shoot up very quickly in temp. They immediately shoot to x+45 (which is normal) and then creep up to x+50 as the water heats up.

 

Comparing core to core and water to water: When the core temp jumps 50, the water temp (after given ample time to heat up) only goes up 6.

 

The second part checks out for me. Increase in coolant temp is additive to the end core temps. However, I am wondering about the initial CPU core temp rise as the voltage is applied. +45C seems like a big hit for 1.175v. This of course depends on the program as well. What are you using for this? OCCT Linpack? Something else? I would expect less for that voltage and that might be further evidence toward a contact issue. Assuming everything else is the same, a +45C core temp instant rise from Linpack would mean the cores would have to be in the mid teens for U14 to be at 60C. It seems more likely the U14 has a much smaller instant core temp rise and that again points the finger at a contact problem.

 

 

What might be useful is for you (and others) to give idle/load temps for the cores and water after a steady load of more than just a few minutes. I might be experiencing "normal" for this water cooler (in which case I'll return it, as it's performing as badly as a stock intel heatsink/fan with just as much noise.)

 

Of course, your 200 watt processor is going to generate a lot more heat than my 95 watts (and your 280mm radiator should dispel it quicker), but I'd be surprised if your water temps only went up 6C after your cores were up 50C for 30min.

 

Actually, that is what happens. I don't normally run Linpack, but and tend to favor AIDA64 blend tests, mostly for the ease of data tracking. After putting in this X62, my core delta starts +30C and over the 60 minutes and the coolant temp only has a delta of +7C (5930K 4.5@1.285v). Now, these are different coolers, different CPUs, and different loads, but that is a fairly consistent value across multiple different tests.

 

What what be useful is if we can get some other Skylake and H100i v2 owners to comment on the coolant delta under load. There are a lot of them out there.

Link to comment
Share on other sites

observations. I am probably heading back to that sooner, rather than later.
I honestly don't want to go back there. I still remember the nightmare of sourcing silicon tubing (from medical supply houses), the "radiator" (an automotive heater core), the pump (a submersible pond pump), making my own reservoir, and ordering a water block (just to find that the barbs on it didn't match the ID of the tubing I already had.) Then cutting a square "tupperware" plastic thing to use as a fan duct to try and increase the efficiency over putting the fan directly on the "radiator." Oh, and don't forget the wires and tubes all over the place.

 

(Of course, these days a custom loop uses all purpose built parts, so it's a bit cleaner. ;)) However, I really just want a "set it up and forget about it" type thing. I'm willing to lose 200Mhz on an overclock if I don't have to actively maintain the cooling system.

 

 

However, I am wondering about the initial CPU core temp rise as the voltage is applied. +45C seems like a big hit for 1.175v. This of course depends on the program as well. What are you using for this? OCCT Linpack? Something else?

 

linpack via linx.exe 0.6.8 (using 24GB of 32GB RAM) The idea is to drive the core temps as high as possible to observe the best possible heat transfer. After changing a heatsink or processor, I'll do this with stock settings to get an idea of what the heatsink is capable of before I start messing with multipliers and voltages. In this case, seeing 75C core temps with a stock multiplier and stock voltage (which comes out to only 91-92 watts) was disappointing. In fact, if memory serves me correctly, it's slightly higher than the temps I'd seen with a default multiplier/voltage i7-4790K using the intel stock heatsink/fan (and that processor is 88 watts.)

 

Assuming everything else is the same, a +45C core temp instant rise from Linpack would mean the cores would have to be in the mid teens for U14 to be at 60C. It seems more likely the U14 has a much smaller instant core temp rise and that again points the finger at a contact problem.
I don't understand. With the U14, the cores went from ~25 idle to ~60 load (+35C) after a few minutes (using the single fan provided with the heatsink.)

 

I don't normally run Linpack, but and tend to favor AIDA64 blend tests, mostly for the ease of data tracking. After putting in this X62, my core delta starts +30C and over the 60 minutes and the coolant temp only has a delta of +7C (5930K 4.5@1.285v). Now, these are different coolers, different CPUs, and different loads, but that is a fairly consistent value across multiple different tests.
That's interesting. Obviously, your setup must be sending MUCH more heat out through the radiator simply by virtue of your cores generating more heat to begin with, but showing lower delta temps. (While linpack is more stressful than AIDA64, I still think your processor on AIDA64 is pushing more than 100 watts of heat.) The question becomes how is that heat getting from the cores out to the surrounding air. One aspect is the IHS contact (I believe yours is soldered.) Another is a larger radiator area. Lots of variables, and my forte is software development, not thermal physics.

 

As you said:

What what be useful is if we can get some other Skylake and H100i v2 owners to comment on the coolant delta under load. There are a lot of them out there.

 

Agreed. It doesn't really even have to be skylake. According to intel's ark site, the i7-4790k is 88 watts, and the 6700k and 7700k are both 91 watts. (Using a 4790k might be confusing as it has an "upgraded" TIM for the IHS mounting.)

 

Sadly, I can't find any H100i v2 examples (with 85-95 watt processors) in these forums that are running the tests for more than a couple of minutes (which is required for proper water cooling tests) or that don't have seemingly broken systems (such as blockages, dead pumps, etc.)

 

Take care

Gary

Link to comment
Share on other sites

The +35C instant CPU temp rise is closer to what I would expect. You normally don't see +45C unless you are running a very high voltage or you have a contact issue. We know the voltage is not high.

 

Before you responded, I had some time and ran a 30 minute OCCT Linpack test at 90% memory capacity. I thought you might be running from Linx, but since our set-ups are apples and oranges anyway, it doesn't really matter if the version is the same and I was more interested in trying a steady state load versus the more variable tests I use for stability. Starting coolant temp was 31C. Over the first 5 minutes it warmed to 36C and there it stayed for the next 30 minutes, only just ticking 37C a second before I shut it down. My core temps were very steady and even, typical of the Linpack test with a range of 58C +-2C on the cool core up to a steady 66C on the HW-E hot #3 & 5 cores. I set the fan speed to a fixed 1280 rpm to try and minimize my surface area advantage to some degree. So the CPU delta ranged from +33-38C depending upon core and the coolant delta was at +5C. It cooled back down to 31C in 3 minutes and 15 seconds following the conclusion of the test.

 

It isn't necessary for other people to run Linpack. Any stress test will do for comparative data, since we are looking for coolant temperature delta (pre-test to peak value). There will be some differences in actual CPU core temps between the models, but the coolant deltas should correspond with wattage (and voltage). Last month when Kaby launched, there were a lot of people here with new gear and a common question was why their CPU temps are 90C when running Prime95, but the coolant temp only went up 3-5C. In most of those cases, it was because the voltage had run away up to heavy levels and Prime itself. However, coolant deltas were still low. I think most users will see larger coolant deltas when gaming compared to stress tests, although that is likely more related to the rise in internal case temp from GPU activity.

Link to comment
Share on other sites

I pulled off the block and re-attached it - this time with a metal washer and rubber o-ring behind the back plate. (The o-ring is just to provide insulation between the washer and mb.) After attaching the mounting studs, it was secure instead wobbly.

 

Applied NH-1 (it's what I have around) and secured the block. I then ran a single 24 GB run of linpack, and came up with very similar numbers to what I had previously (but this time with an open case.) The cores are, perhaps, 2 degrees cooler. The water temp seems to have the same behavior as before. In 5 minutes (starting when linpack is done allocating 24 GB of RAM) the cores did the instant jump to 70-75 (typical) and the water only went up perhaps 3 degrees.

 

I KNOW that there's good contact between the block and IHS this time. I know the pump is working. I know the radiator fans are blowing. I know the thermal paste is decent.

 

Granted, if the chip's IHS was soldered instead of the automated (and questionable) intel TIM application, I'd see better numbers. However, I can't figure out why the U14 would give me temps in the 60's (peak) when this cooler is giving me peak temps that are 15 degrees higher. That just doesn't make sense to me. I shouldn't see numbers that high so quickly even if I turned the radiator fans off!

 

Something isn't right... and I'm not sure what it is.

 

I've attached "capture.png" which is a screen snippet from HWinfo while linpack is running. On the left you can see the core temps, lower left shows the H100iv2 data. The right side is populated with data from the motherboard (including vcore, MB temp, etc.) The "CPU" and "CPU Fan Opt" are the radiator fans. The rest really don't matter, as the side of the case is off. The Asus reported "water pump" speed could be ignored... I currently have the pump plugged into the "water pump" header.

Capture.thumb.PNG.f7fe8d600a040689036a1051603567b8.PNG

Link to comment
Share on other sites

...another thought... The copper plate on the water block (which makes contact with the IHS) is very rough on my sample. Cleaning the old TIM off was annoying because of that. The machining grooves created a situation where I had to rub in specific direction to get the old TIM off.

 

Being that I've never actually used a heatsink that wasn't at least smooth, I'm wondering if that might be the problem I'm seeing.

Link to comment
Share on other sites

Hmm... nothing wrong with NT-H1. I have used it for years and TIM brand should matter in this way. The "machining lines" on the cold plate are interesting. They seem to be air brushed out of the official photos, but I found a few naked snap shots on the web where I can see them. Straight lines, not concentric? I think this is probably consistent on all units, but I need someone else with a H100i v2 to confirm. I don't like the idea that you needed to rub a certain way to remove it. I haven't heard anyone complain of it before, but likely not everyone knows that is a bit abnormal. I am not sure this makes up for the extra 10C or so.

 

I am out of ideas for the moment. Only two other things come to mind. Even if you are powering the fans from CPU/OPT, make sure the header that has the H100i v2 pump lead (CHA_2?) is set to 100%/12v/"Full speed" in the BIOS Q-Fan or Advanced Monitor Tab (disabled = full speed). Your BIOS will not display the proper pump speed, although Link will. It generally shows half the actual speed. Based on the numbers in your fan column, I can't clearly identify which header it is on. The setting in Link is a simple high/low switch, with Performance running the pump near 3000 rpm and the Quiet closer to 2000.

 

Also, it looks like the highest fan speed you used was ~1200 rpm. We talked about not needing more fan speed when the coolant temp doesn't rise and you mentioned testing this already. However, I wonder if a short run at max pump speed (Performance Mode) and a fixed higher fan speed (something like 1600-1700) would make a difference. There is the possibility the coolant temp sensor is not registering properly or another variable we have not realized is affecting the results. No need to go 30 minutes. If it is the same in the first 5, it's going to be the same all the way through.

Link to comment
Share on other sites

Straight lines, not concentric?

correct

 

I am not sure this makes up for the extra 10C or so.
No idea. I'm almost ready to just bring this thing back and return it as defective. Perhaps replace it with something more effective like the cheap intel heatsink/fan that came with an i7-7700 (non-K)

 

Even if you are powering the fans from CPU/OPT, make sure the header that has the H100i v2 pump lead (CHA_2?) is set to 100%/12v/"Full speed" in the BIOS Q-Fan or Advanced Monitor Tab (disabled = full speed).

At the moment, it's plugged into W_PUMP on the board which is configured to be an uncontrolled header (100% power.) On the screen snippet, it can be seen on the lower right corner labeled "Water Pump" and does show ~ 1/2 the RPM's that CL shows

 

Also, it looks like the highest fan speed you used was ~1200 rpm. We talked about not needing more fan speed when the coolant temp doesn't rise and you mentioned testing this already. However, I wonder if a short run at max pump speed (Performance Mode) and a fixed higher fan speed (something like 1600-1700) would make a difference. There is the possibility the coolant temp sensor is not registering properly or another variable we have not realized is affecting the results. No need to go 30 minutes. If it is the same in the first 5, it's going to be the same all the way through.

I'll pull the radiator fans off the motherboard and plug them into the H100i headers (and then move the h100i to the CPU header after changing that to manual/100%)

 

Then I'll put CL into "performance mode" which should run everything at 100% and I'll see what happens...

 

Take care

Gary

Link to comment
Share on other sites

Wow. Can you imagine what two of those corsair fans running at 100% sound like with an open case? "Loud" would be an understatement. According to CL, they are at 2700 rpm. Noisy.

 

Pump set to "performance" (reported as 2910 rpm) and fans set to fixed % of 100% (and plugged into the water block) (reported as 2700-2760 rpm.)

 

After 5 minutes, I see nothing different from previous tests (other than the spots in my eyes from the headache I'm getting.)

Link to comment
Share on other sites

Yes, I hate those fans. Test concluded. Don't do it again.

 

140mm fans are usually very linear in their performance curve. 120mm fans are not, often with funny little dips in the mid range. Small changes in fan speed have a noticeable affect on my GPU cooler that is asked to handle a lot of wattage. I often see 5C+ swings over relatively small changes in fan rpm in the 1000-1300 range. We needed to rule out basing all our assessments on the coolant temp behavior. We have. It did not help.

Link to comment
Share on other sites

I have a pair of NF-F12 iPPC 2000 fans still in the amazon shipping box which were destined for the h100iV2, but I'm really not sure if I'm going to keep this thing. It does serve the purpose of cosmetics when compared to something like the noctua U14, but an intel stock heatsink (which cools just as well, if you can ignore the whining fan) is also very compact and wouldn't distract from the cosmetics of the case.

 

My biggest problem is kind of amusing: I bought the cooler while 2 hours away from my home at a micro center. That's the closest micro center to my house. So, to return it, I have to make a 4+ hour road trip. (I called and they won't accept a return shipped to the store.)

 

So, if I DO make that trip, I wonder what I should buy instead... They have Kraken X52's for $30 more (but aren't they manufactured by the same company that makes the H100iV2?) They also have EKWB kits, but I really wanted something simple and easy... Not many good choices. :(

 

Perhaps I should run to my local best buy, buy another H100iV2 kit to see if perhaps there might be some defect in the one I have?

Link to comment
Share on other sites

That is disappointing about Microcenter. After years of ignoring them, I have started to work with them again as they have become competitive.

 

I don't know about the Best Buy run. If they are friendly on returns, it might be worth the time. If you end up stuck with the Microcenter model, you would still be able to RMA it through Corsair at a later date.

 

As for the fans, I would change them if you keep it. I just don't like that fan. The NF-F12's are plenty effective. They will start to pick up a little hum and whir above 1500 rrpm, although most fans do. They are not completely silent at lower speeds, but were not designed to be. There are some other choices for that if needed. They would also work well on any 240mm radiator.

Link to comment
Share on other sites

A minor update for those reading:

 

At this point, I returned the H100i V2 to microcenter. While there, I replaced my motherboard and CPU (don't ask why) with an i7700k and an "Asus z270 really long name that ends with "hero." I think it was something like "Asus Republicans of Golfers Mates Nine Heros" (or maybe not.) (It wasn't very long ago that mb's had short names like "P8P67." I considered spending extra for a board that ended with "Code", but all it seemed to offer over the "Hero" was wifi and a plastic layer to keep all the motherboard heat trapped against the motherboard.

 

I had also picked up a different H100iV2 (from a local Best Buy) and found it had very slightly better performance, but still not at the level of even a hyper212.

 

I'm pretty sure that the problem with the H100i isn't the radiator or pump performance, but the lack of thermal transfer between the processor and the water block. Of course, part of that is the thermal interface between the processor core and the IHS (which is why people delid.) However, that's only PART of the problem (known because I get better cooling with the same chip using a big air cooler.) That implies that the remainder of the issue is with the water block and/or it's mount. (I wonder if the "fins" inside the block are all clogged up as I've seen in many pictures... that would really impact heat transfer, wouldn't it?)

 

So, considering that I can get at least the same level of performance from a mid-sized tower air cooler, and at a lower level of noise, this second h100iV2 will also be getting returned. I'll run with air for a while, and perhaps eventually go with a another brand AiO (not asetek based) or perhaps even a semi-custom loop "kit." (EKWB seems to have some interesting kits.)

 

As for corsair... I'm keeping the Crystal 460X RGB case. Even with the space constraints, I really like the case.

 

Take care

Gary

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...