H115i with Intel I9-9900X

Hey everyone, I have seen a few threads on this issue, but wanted to see what people thought of this. Please note, this is my first ever liquid cooling solution I have ever used, so I'm a bit of a noob when it comes to what to expect temperature differences between the package temp and the H115 temp to be.

I have the H115i with an Intel I9-9900X processor (Intel stock speeds with turbo boost). I'm getting a rather high thermal difference between the CPU Package temperature and the H115i liquid(?) temperature (whatever "temp" means in the iCUE software).

If my PC is mostly idle, the H115i temps are about 25 to 30C (depending on ambient air temp in my room), and the Package temp is between 30 and 40C. So, the difference there is only 5 to 15 degrees C, which is pretty good.

However, when under load (stress tests or some heavy simulation games), the H115i temps will only get to about 36C (never ever seen it get above 38C), but the package temp will be between 80 and 90C (usually high 80s). This concerns me as my CPU rated max temp is 92C per Intel design, and hitting high 80s is a bit uncomfortable for me.

When the system is being stressed, there is a temp difference between the CPU Package and the H115i temps of around (and sometimes over) 50C. That just seems way too much IMO.

I have tried reseating the pump (using new paste), and various combinations of fan speed and pump speed options (including running everything 100%), but nothing seems to really reduce the CPU temp, or the difference between the CPU and the H115i temperatures.

My question is, is it normal to have a 50C difference between the temperatures of the CPU and the H115? Do I need to look at getting a bigger radiator setup (H150i maybe)?

First, there are way to many 9900's, I had to look it up. The TJ Max on that is lower than I thought, though I have never played with 2011.

What are you using to stress test it? I wouldn't use Prime95 w/ AVX instructions, it gets really hot.

I would look at what other people have done, though I bet a lot of people who go for the HEDT might go with a custom loop. At some point Corsair will release HydroX, that might be an option. Though a H150i does have a larger area.Your liquid seems warm and I would watch your temps.

Maybe you should look at overclocking, I bet the boost is being a little aggressive on the voltage.

Thank you for the reply.

For an actual stress test, I am using the AIDA built in "System Stability Test" with only the "Stress CPU" option checked. It 100% maxes out each core for as long as I leave it running.

For a "game", one I have recently started messing around with is "Universe Sandbox", which basically allows you to run simulations of different universe layouts (Black holes, large planets, messing with Earth's gravity, etc). Believe it or not, it is a very threaded application, and some of the simulations can actually hit 80 to 90% of all 10 cores (20 threads). It gets to about the same temperature as the AIDA stress test using this also.

When you say my liquid seems warm, what do you mean? I have seen other posts in here saying that liquid temps should be able to get to the mid 40C to 50C without issue. If anything I feel my liquid temps are a bit on the low side, like the heat is not efficiently transferring from the CPU to the H115i.

And I have to ask how overclocking a CPU that is already getting pretty hot is supposed to help. I do want to overclock it, but I want to make sure my cooling solution is up to the task before I really push it.

Remember the CPU is not submersed in a tube of liquid. All CPU cooling, whether water or air, is conductive contact heat transfer. The voltage is applied on the motherboard pin side and heat is created at that junction. The heat must pass through the CPU to contact plate of the cooler to be then be transferred away. Heat transfer is never 100% efficient. The stove top heating element, the frying pan, and the egg are not the same temperature.

The liquid really serves as a heat transport system. Water is much better at this than air and you can handle more added heat for longer time periods before the effects become noticeable negative. The relevance of water temp in liquid cooling is it's a helpful measure of the amount of heat added to the system. The heat transfer across the cold plate goes both ways, so this makes the coolant temperature the effective minimum CPU temperature. Coolant can't be below local ambient temp and the CPU can't be below the coolant temperature. Those values help you establish the minimum and the cold you keep the coolant, the lower the ultimate CPU temp will be.

It also gives you an idea of how hard to run the fans. Each -1C in coolant temp is -1C in CPU temp. Powering the fans up to 1500 rpm to reduce your coolant by -2C may not always be a good trade for noise and a measly +2C in CPU temp. If you coolant is getting into the 40C+ range, you need to take a hard look at your set-up and environmental factors. Rather than a positive "look how much heat this cooler is sucking up", it's really the opposite. That would be "why is my cooler not able to get rid of it's heat" and you need to look for reasons.

CPU voltage is the limiting factor for everyone. I can give you a 10ft tall radiator and fans to match and you still won't be able to push past the voltage that takes to the CPU instantly to 90C - whatever that is on a 9900K. The cooler keeps it from getting worse, but that instant CPU temp rise when going to 100% load is unavoidable and the hard limit for your specific settings.

Thank you for the reply.

 

For an actual stress test, I am using the AIDA built in "System Stability Test" with only the "Stress CPU" option checked. It 100% maxes out each core for as long as I leave it running.

 

For a "game", one I have recently started messing around with is "Universe Sandbox", which basically allows you to run simulations of different universe layouts (Black holes, large planets, messing with Earth's gravity, etc). Believe it or not, it is a very threaded application, and some of the simulations can actually hit 80 to 90% of all 10 cores (20 threads). It gets to about the same temperature as the AIDA stress test using this also.

 

When you say my liquid seems warm, what do you mean? I have seen other posts in here saying that liquid temps should be able to get to the mid 40C to 50C without issue. If anything I feel my liquid temps are a bit on the low side, like the heat is not efficiently transferring from the CPU to the H115i.

 

And I have to ask how overclocking a CPU that is already getting pretty hot is supposed to help. I do want to overclock it, but I want to make sure my cooling solution is up to the task before I really push it.

Here's how I built my 5.1GHz / PC3400 i9-9900K rig, in a Fractal Define R6 case...

H115i radiator at the front of the case, with the supplied ML (slower) fans at the rear of the rad, pulling (Pump sensor). An additional pair of ML (faster) fans at the front of the rad, pushing (CPU sensor). Another pair of ML (faster) fans pulling air out of the top vent and yet another pair pushing air up from the base vent. These 140mm ML fan twin-packs are a bargain on Amazon at the mo!

Everything in this setup runs cool, even when benchmarking with AIDA. In spite of the numerous fans, nine in total, it is also quiet, because they are rarely driven above 40%, due to carefully crafted fan profiles. I've not seen my liquid temp rise above 34C running AIDA.

The trick is, IMHO, loads of fans running relatively slowly, delivering a steady cool air stream and also ensuring that your radiator is receiving cool external air. Oh, and don't worry about the different speed ratings of the radiator fans (front vs rear) as you can use this to your advantage with fan profiles.

I have done mostly OC'ing of i5, though my current CPU is a 9900K and I OC'ed that as well. IMHO the i5 were the most fun to OC as I think they really underrated for their boost and stock speeds. I think this is partially to help sell the i7 and i9.

When I first wanted to do some stress testing, I left my 9900K on stock settings, I was surprised at how hot it got in stock setting. My system runs cooler at 5Ghz static voltage OC than 4.8 stock boost. I think the boost can leverage a lot more voltage than really need for the CPU, you can look at running the system at the boost speed and undervolt the system to get it to run cooler. I would look at some under volting guides and see what others have done with your CPU. I have never done it, I just try to get the best OC w/o out going crazy on voltage.

Here is the video of my experience: [ame]