Jump to content
Corsair Community

H115i Core Load temps vs Coolant Temp


Random02

Recommended Posts

Hello , i have a h115i cooler with a core i7 9700k. it is currently at stock with a vcore of 1.15. Load under the cores get to about 82 degrees with prime 95 v26.6. The coolant tempreture goes to about 32 max. The pump is on performance mode 3000rpm .

I have re seated the pump a couple of times using arctic silver 5. the paste spreads and covers the hole cpu. Is it normal for the core temperature to be this high vs the coolant temperature ?

Link to comment
Share on other sites

Run a different stress test (AIDA, Intel XTU, OCCT, etc) before you start tearing things down. A bad mount or poor contact will be terrible on everything. On the other hand, if you only need to tweak some settings in the BIOS (maybe more than a few), the tests should scale with their relative load --- XTU and AIDA on the lower end, OCCT in the middle, and Prime at the top.
Link to comment
Share on other sites

What is the vCore when it's under load? It's not going to be 1.15 ... that much is certain.

 

Asus is notorious for dumping on the voltage under load, especially with the defaults.

 

Hi , Thanks for the reply. The voltage under load goes to 1.05. But how would that increase the core temp vs 1.15vcore ?

Link to comment
Share on other sites

Run a different stress test (AIDA, Intel XTU, OCCT, etc) before you start tearing things down. A bad mount or poor contact will be terrible on everything. On the other hand, if you only need to tweak some settings in the BIOS (maybe more than a few), the tests should scale with their relative load --- XTU and AIDA on the lower end, OCCT in the middle, and Prime at the top.

 

Ok so i did Adia cpu + fpu + cache. cpu core temp was 71C on the hottest core . Coolent was at about 31C. Ambiant is about 25 C.

Link to comment
Share on other sites

I think you most likely need only to tweak the settings for running synthetic benchmarks. Auto voltage plus your standard stress test will certainly come out hot as it loads everything up. The board's goal is to make the worst CPU ever pass, so it rather heavily pushes voltages. I have not looked inside the the new Z390 BIOS versions, but assuming it is similar to Z370, you will want to set your IA/DC load line voltages to 0.01 (in the Advanced BIOS, AI Tweaker column, Power management sub menu at the bottom. Also way back at the top of the AI Tweaker column will be something that describes the default voltage boost or "behavior". Set it to best case scenario, if they are still using that term on on the Strix boards. You likely want to google around for some board specific advice.
Link to comment
Share on other sites

My setting in the bios.

 

xmp enabled

Asus MCE disabled

LLC level 6 .

vcore at 1.15v ( thats the lowest i could go stock freq 4.6 Ghz sync all cores)

set the max cpu power to 255.5

CPU Cache 43

Is There anything else i need to change ?

 

Im scared i might just have a bad IHS sTIM connection. if that is possible :/

Link to comment
Share on other sites

I think you most likely need only to tweak the settings for running synthetic benchmarks. Auto voltage plus your standard stress test will certainly come out hot as it loads everything up. The board's goal is to make the worst CPU ever pass, so it rather heavily pushes voltages. I have not looked inside the the new Z390 BIOS versions, but assuming it is similar to Z370, you will want to set your IA/DC load line voltages to 0.01 (in the Advanced BIOS, AI Tweaker column, Power management sub menu at the bottom. Also way back at the top of the AI Tweaker column will be something that describes the default voltage boost or "behavior". Set it to best case scenario, if they are still using that term on on the Strix boards. You likely want to google around for some board specific advice.

 

 

I did set the vcore manual . I thought that the cpu svid setting was ignored if the vcore was set manually. I change the IA/DC to 0.01. Thanks

Link to comment
Share on other sites

Hi , Thanks for the reply. The voltage under load goes to 1.05. But how would that increase the core temp vs 1.15vcore ?

 

That does not sound right at all. Not in the least. Load voltage should be higher than idle voltage. That's especially true if you are overclocking on all cores.

Link to comment
Share on other sites

I did set the vcore manual . I thought that the cpu svid setting was ignored if the vcore was set manually. I change the IA/DC to 0.01. Thanks

 

Adaptive mode allows you to set the vCore manually but will adjust voltage. Full manual has a single, steady, unchanging vCore and you typically don't want that as your idle temps will be warmer than they need to be as the CPU won't down-volt.

Link to comment
Share on other sites

I did set the vcore manual . I thought that the cpu svid setting was ignored if the vcore was set manually. I change the IA/DC to 0.01. Thanks

 

Yes, the IA/DC is important for Auto and Adaptive voltages. It should not matter for manual/fixed.

 

MCE off is good. Keep it off. You can always set your cores to the frequency you want.

 

LLC6 is likely fine and also what I use on the Z370 Code X.

 

 

What about your actual usage temps doing whatever it is you like to do? Presumably you didn't put this together to hunt Mersenne Primes.

Link to comment
Share on other sites

I don't have as much real world data as I would like for the 8c/8t model, but I will base this off the comparative wattage for Hand Brake published about the internet. My coolant delta on a H115i PRO for that load is about +5 to 5.5C. I'/m going to give you a target of +6C. If you have not gone over 32C coolant temp so far, things are running quite well. Regardless of coolant delta, the lower the coolant temperature, the lower the overall CPU temp baseline. A CPU doing job X at a 25C coolant temp will have a -10C lower average CPU temp than the same chip doing it at 35C coolant temp. In a CPU load situation, this is fairly straightforward and you can see this in program with line graphs of CPU temp. Watch the knock-on effect during the first ten minutes as the CPU temp will go +1C ever 1-2 minutes, right in step with the rise in coolant temperature.

 

Where this becomes more difficult to assess is during gaming or any other CPU/GPU multi-component workload. Just as an increase in room temperature will boost everything in the room by +1C, a rise in case temperature will do the same to the CPU , radiator, and coolant. All that GPU waste heat has to go somewhere. It's not unusual for people with 300W GPUs to see higher coolant temps when gaming compared to a maximum CPU load. This is a function of the increase in case temp, rather than the cooler's inability to function. You have to put these number into some context by trying assess and then deduct any change in case temp. Usually a combination of motherboard temp senors, thermistors, or even drive temps can give you some idea of what's going on.

Link to comment
Share on other sites

Archived

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

×
×
  • Create New...