Jump to content
Corsair Community

H115i Pro is not for Performance Cooling, unless you buy additional fans


RayzTheRoof
 Share

Recommended Posts

Fair warning to those looking into this. I thought this would be an updated version of the H115i, which the pump and radiator pretty much are. But the fans are half the RPM of those included with the H115i.

 

I used both in the same system in pull configuration at front intake. I would get 40s for my temps in Overwatch, not even at max pump speed or high fan speed. With the Pro I am running at max fans and pump speed and get high 50s and low 60s. THis is at 15-25% load. Silent, but horrible performance.

Link to comment
Share on other sites

Hmmm ... interesting.

 

I'm actually seeing quite different performance, at least when compared to my previous H100i V2. Max CPU temps are about the same ... perhaps 1C lower, but that's within the margin of error. Average temp over an extended load test (1 hr) using RealBench is about 2-3C lower. The coolant temperatures are also lower by 4-5C. It's the coolant temp, btw, not the CPU temp, that is directly affected by the fan performance. And my temps in Overwatch were also about the same ... 40-50C and that's at 5.1Ghz.

 

Now ... even before I got the 115i PRO, I reduced the speed on my ML-120 fans that I had on the H100i V2 to see if the ML-RGB fans would be up to the job. Note that the ML-120 fans (non RGB, not with the PRO coolers) have a higher max speed and perform at least as well as the stock fans ... if not better. I saw absolutely no difference in CPU temps (expected) and about 1C difference in coolant temperature.

 

From what you are describing, it seems to me that you either have a mounting issue or perhaps a wonky cooler. Of course, it's difficult to even begin to analyze it because you've not filled out your system specs. ;-)

Link to comment
Share on other sites

Quick note: checking HW monitor shows there are even spikes to high 60s, with the max being 71. Though that happens for only a millisecond.

 

That actually sounds pretty normal for a stock Intel chip that hasn't been delidded. Regardless of cooler.

What is your utilization and what are your voltages?

Link to comment
Share on other sites

I thought multicore enhancement might be the problem but disabled it, noticed no significant differences. Stock speeds meaning I have not changed anything in my system, only BIOS adjustment ever was disabling multicore enhancement and enabling XMP. Here is a look at my temperatures linked to clock speeds and voltages:

https://i.imgur.com/g19p2mZ.png

https://i.imgur.com/NDe7v5p.png

 

71 seems a bit high for an AIO with the CPU not even hitting max turbo.

 

Edit: the example that I gave in the original post was getting into high to low 60s with 30% lutilization.

Edited by RayzTheRoof
Link to comment
Share on other sites

Actually, need vCore, not VID.

But ... if you're at auto settings, that's likely the problem. You have a VID of 1.46, which is quite high. If you're at auto and the system is actually giving the CPU that kind of voltage, I'm surprised your temps aren't higher.

Link to comment
Share on other sites

That is not really Vcore either. Minimum on the 8700k is around 0.60v. ASUS must have done something dreadful to their Z370s because all of the monitoring programs are struggling. Only HWInfo64 can read my Code X. That is also key to your issue.

 

You have a BIOS settings problem. Fans, fan speed, and pump speed make single digit differences. When you see temps go up +20C, something isn’t right and the voltage is always the first thing to check. These new boards are very prickly about some settings and auto has gone from safe to unpredictable. This is where we need to look.

 

If you are using auto/adaptive voltage, go into the advanced Bios F7, Extreme tweaker, go half way down to internal CPU power management. Set the IA and DC voltage to 0.01 for both boxes. You can also see a visual depiction of this in the ASUS Kaby Lake overclock guide. Even with a specific set adaptive voltage , leaving those two also be will allow my vcore to go from 1.30 to 1.52v. You have to clamp down on those.

Edited by c-attack
Link to comment
Share on other sites

That is not really Vcore either. Minimum on the 8700k is around 0.60v. ASUS must have done something dreadful to their Z370s because all of the monitoring programs are struggling. Only HWInfo64 can read my Code X. That is also key to your issue.

 

You have a BIOS settings problem. Fans, fan speed, and pump speed make single digit differences. When you see temps go up +20C, something isn’t right and the voltage is always the first thing to check. These new boards are very prickly about some settings and auto has gone from safe to unpredictable. This is where we need to look.

 

If you are using auto/adaptive voltage, go into the advanced Bios F7, Extreme tweaker, go half way down to internal CPU power management. Set the IA and DC voltage to 0.01 for both boxes. You can also see a visual depiction of this in the ASUS Kaby Lake overclock guide. Even with a specific set adaptive voltage , leaving those two also be will allow my vcore to go from 1.30 to 1.52v. You have to clamp down on those.

 

Here is the vcore from HWInfo: https://i.imgur.com/IWzs2ER.png

 

Seems high, right? I will make those adjustments soon, thanks. What exactly do they do? But also, is it a bad sign that I got a faulty or unlucky board if the vcore is that high and I am not even hitting 4.7GHz boost?

Link to comment
Share on other sites

Here is the vcore from HWInfo: https://i.imgur.com/IWzs2ER.png

 

Seems high, right? I will make those adjustments soon, thanks. What exactly do they do? But also, is it a bad sign that I got a faulty or unlucky board if the vcore is that high and I am not even hitting 4.7GHz boost?

 

Oh, it's high all right. And high voltage directly leads to high temps. And no, it's not a sign of a bad board ... it's one of those things that "Auto" voltage settings do. Asus, with the multi core enhancement enabled by default, is particularly bad about it.

 

Try setting your voltage to (IIRC) 1.2V ... IIRC, that's the stock max voltage for the i7. Also, from what I've heard, Coffee Lake overclocks pretty well with low voltages; I've seen reports of 5Ghz with around 1.3V.

 

c-attack may have more specific details and advice.

Link to comment
Share on other sites

Oh, it's high all right. And high voltage directly leads to high temps. And no, it's not a sign of a bad board ... it's one of those things that "Auto" voltage settings do. Asus, with the multi core enhancement enabled by default, is particularly bad about it.

 

Try setting your voltage to (IIRC) 1.2V ... IIRC, that's the stock max voltage for the i7. Also, from what I've heard, Coffee Lake overclocks pretty well with low voltages; I've seen reports of 5Ghz with around 1.3V.

 

c-attack may have more specific details and advice.

 

new and improved vcore with temps barely going over 50 except for spikes, using the same game to test: https://i.imgur.com/YW7ncs5.png

 

Is there a specific way to lower the voltage even further because I feel that is still unnecessary? I used what the poster above you said and adjusted IA AC/DC Load Line to be 0.01, which I guess adjusts the voltage parameters for Turbo.

 

Should I change that number to be something else for lower voltages as well?

 

Side note: isn't it kind of odd that the motherboard lets the voltage go that high without warning in the first place? Makes me wonder if my board was used and had different settings

Edited by RayzTheRoof
Link to comment
Share on other sites

OK, that's better!

 

IA and DC Line Load Calibration determine how much additional voltage will be applied in certain situations. The problem seems to be that it is getting padded on both ends (IA/DC and Vcore) and the combined cushion is way too much. If you use Adaptive or Auto voltage on Kaby/Coffee Asus boards, you must set these values effectively to zero (0.01) to prevent pilling on. Even with very tight settings, my 5.0/1.30v OC will hit more than 1.50v if I put those two back on Auto. Yes, I think it is a bit counter-intuitive and most people think they are safe leaving these values on auto and the board will keep things restrained. Instead, it does the opposite and the primary focus is on making sure the board posts with whatever crazy settings you enter. The overall goal is to make it work with the worst 8700K every made. Whether this is how it should be is a much longer discussion, but you can't use auto settings for much on the Z170-, 270, 370 platforms.

 

I am not sure why your VID and Vcore were so high on the straight default settings. I have seen some peakier numbers, but those were odd. Part of the issue seems to be overlapping tuning tools. Some are broad and others pointed, but it takes some practice with finding an effective combination. It is possible the combination of the "default" (best case/worst case/typical case slider) in combination with a very flexible IA/DC line was letting the VID get way too high. I have not seen a clear explanation of the thae best case/worst case thing from Asus yet, but presumably it shifts the VID table along a slider with Best Case giving the lowest scale or boost. The default is worst case and thus a large boost.

 

Did you set the Vcore to manual 1.20v? Or Adapative? Right now you look good and I would not be overly keen to lower that value until you have some time at this setting. Additionally, it becomes very difficult to use adaptive with a negative offset if the workable value is below the VID table for that frequency. Really the better thing to do is bump up the frequency by 100 MHz rather than try and lower the Vcore. -2 to 3C at that temp range is meaningless and your room temps swings up and down more than that.

 

If you haven't looked at it already, read the Asus Kaby Lake guide. You don't have to understand all of it and it is not comprehensive, but does detail some of the key terms.

https://edgeup.asus.com/2017/kaby-lake-overclocking-guide/5/

Edited by c-attack
Link to comment
Share on other sites

I didn't change anything regarding vcore; I only adjusted the IA AC/DC option. I never attempted messing with voltages and overclocking before so I'm not the best at doing this without guidance :P

 

Could this a sign I got unlucky with my CPU? I just hope it doesn't interfere or cause problems if I try to OC down the line as well.

Link to comment
Share on other sites

I didn't change anything regarding vcore; I only adjusted the IA AC/DC option. I never attempted messing with voltages and overclocking before so I'm not the best at doing this without guidance :P

 

Could this a sign I got unlucky with my CPU? I just hope it doesn't interfere or cause problems if I try to OC down the line as well.

 

No, actually I would say the opposite. If all you did was clamp the IA/DC, then the CPU is using the natural VID table for it at the stock frequency. That appears to be around 1.20v with some normal overshoot to 1.23v. If you had a bad CPU, I would expect to see higher voltages at your current settings. If you want to tighten it just a little, change the setting at the top of the Extreme Tweaker column from "Auto/Default" to Normal or Best Case scenario. I would do it one step at a time, starting with Normal and then let it ride for a few days. This is also optional since it appears to be keeping a leash on the voltage now.

Link to comment
Share on other sites

Yes, 1.20-1.23v is definitely safe and appropriate for the frequency. 4.3 is the max when all cores are active, which is pretty much all the time. Theoretically, it can run 4.7 single core when at stock settings, but you may never see it. My first impressions of the 8700K is the cores are always busy.
Link to comment
Share on other sites

  • 2 weeks later...
Yes, 1.20-1.23v is definitely safe and appropriate for the frequency. 4.3 is the max when all cores are active, which is pretty much all the time. Theoretically, it can run 4.7 single core when at stock settings, but you may never see it. My first impressions of the 8700K is the cores are always busy.

 

Bit of a followup. These were my voltages after running Assassin's Creed Origins for a bit, are these safe as well? I don't know much about VCCSA and VCCIO. Though my Vcore jumped to 1.26 for a bit so I changed whatever setting it was to run in best case scenario, but that didn't change anything.

 

https://i.imgur.com/WRTAkLq.png

 

Is a constant 1.232 worrisome? That's the most consistent voltage it stays at. Getting spikes once in a while to 1.248 and 1.26.

Edited by RayzTheRoof
Link to comment
Share on other sites

I am going to add as I am happy and surprised on the performance improvement from my H100i V2 to H150i PRO.

 

I was not expecting much for the H150i Pro, at least not from a 240 to 360mm radiator size increase. As the default fans only runs at max 1600 it is only quieter, I accept that, and that is only my expectation ~ aaand also want the AR-GEE-BEE. I would've went for the 115i Pro but I have clearance issues with my GPU (too long) and a set of ML RGB fans I currently have is a 3x 120mm.

 

My cooling gains from H100i v2 to H150i Pro with my....

 

...wait....

...updating system specs on profile...

 

My 8700k is at most ~ 10 degrees Celsius. Both 2 hrs of RealBench and Aida64.

I retest and confirmed that my H100i v2 is fully functional (still cooling very good on a 5930K @ 1.300v). Maybe my thermal paste or pump install was faulty before I replaced it. I do not know lol! But that is the thermal improvement that my H150i PRO is performing right now. Coolant temp is about 5 degrees cooler than the H100i V2. Quite surprised and puzzled, but I am glad :)

Edited by gabby131
Link to comment
Share on other sites

  • 7 months later...

I come from an old(ish) water extreme 3 with 3 noctua fans attached to it. (2 top one bottom)

 

The water extreme 3 has a gen4 asetek pump. Also the radiator is 240mm

 

My 115i Pro, which has a 280 rad and I'm using two extra corsair LL140 fans at the top under performs compared to the water extreme 3. (I'm also using 2 top one bottom)

 

I'm using the exact same paste (noctua NTH1) and the 115i Pro is 4c hotter. Seems little but it affects my OC and how much vcore can I push. (1800x at 4ghz / 1.412v) previously I could OC a bit higher and maintain the temps under 75c. Not anymore.

 

I suspect the Gen6 pump is garbage. Specifically the cold plate. What where they thinking?

 

It came as a surprise. Th 115i Pro pump can go much faster and the rad is bigger. Yet is underwhelming.

 

Any ideas why? Can it be that the cold plate needs lapping? I might have to install that other fan on the bottom then. (is very tricky with the VRM heatsink in the way)

Link to comment
Share on other sites

Pump speed is almost never relevant on these models. You can toggle between the "Balanced" and "Extreme" pump speeds and see for yourself. However, you probably will see a difference when using the low "Quiet" speed at 1100.

 

Make sure you are thorough in your data tracking. A 4C difference in core temp is a flicker in time that could come from program changes, testing methods, and of course your room and case temp have more bearing on day to day changes than anything else. Your prior cooler likely did not have coolant temp data, so that makes comparison a bit more difficult. See how much coolant temp change you see on load in iCUE/Link. You shouldn't need the 3rd fan and the 280mm should be able to dump its heat with just two 140s strolling along at moderate pace. I never need more than 700-800 for my 8700K and can still keep the coolant delta under 5C.

 

I haven't quite gotten around to doing metallurgic testing on the cold plate. However, since most every AIO unit available is Asetek, it seems unlikely there will be a difference from generation to generation. Not much progress to make at this level.

Link to comment
Share on other sites

I'm testing with Mandelbulb, a very controlled test actually. I spent a month rendering with it at 4.025Ghz / 1.425v (vdroop around 1.4v)

 

I'm writing you this as i test, i've added the extra maglev fan at the bottom or the rad. A total of 4 fans.

 

Before with my old AIO I got 73c (pump full at 2000RPM, fans full at 1500RPM)

 

Now I get 77c at 4Ghz / 1.412v (pump at full 2700RPM, fans full, two at 1200 and two at 1500RPM)

 

You shouldn't need the 3rd fan and the 280mm should be able to dump its heat with just two 140s strolling along at moderate pace.

 

That's what I thought when I decided to buy this product, but something in this AIO's design is just not right.

 

I just hope 77 for long periods is ok with an OC'd Ryzen 1800x...I would guess so, is just two degree higher than the absolute safe 75c

 

I tried comparing Balanced and Full pump speeds after the AIO found it's balance at 77c as soon as I set the pump to balanced the temp started climbing. It makes a difference, little but it does.

 

Switching momentarily to quiet pump made the temps to climb over 80c real fast. So yeah, the quiet pump mode is basically useless at 1100RPM, also because at balanced is virtually silent whereas I started to hear my Gen4 pump from 1500RPM upwards.

 

Anyway, it feels real silent compared to the old AIO, the 140mm fans help. That's a good point.

 

I am kind of debating if I like the way this AIO works from iCUE.

 

On one hand I like that if I set a speed, that speed is constant. i can live with that better than having fans ramping up and down. Also I like I can pre cool the water before starting to render. Also I can have a "quiet balanced" profile with the top fans in quiet mode and the bottom fans in balanced. That helps when I render with less threads.

 

For any other tasks, even gaming, the quiet profile (quiet fans and balanced pump) handles it all. And jeez! is very quiet.

 

One issue would be, if I ever forget to switch from my quiet to my full profile I will have a problem.

Edited by Sacco Belmonte
Link to comment
Share on other sites

 Share

×
×
  • Create New...