Are these temps too high? h100i v2

So I recently upgraded to an 8700k and was hoping to get anywhere between 4.7 - 5.0ghz based on what other people seem to be able to achieve. I can't even maintain 4.3ghz including undervolting to 1.24v without hitting 90c almost instantly. I get similar results no matter what stress test program I use, aida64, realbench, prime95 etc.

 

I've tried new thermal paste, upgrading the stock fans on the h100i to the magnetic levitation ones (ml120). I have 3 intake fans on the front of my case and 1 exhaust at the rear, with the h100i mounted at the top with the fans in a push setup. So I should have pretty good airflow all around with slight positive pressure as the ml120s won't exhaust as much since their mounted on the radiator.

 

I also seem to get significantly different readings on temperatures depending on the software. aida64 seems to report too low of a temperature. realtemp and corsair link seem to be closer together, however corsair shows the first core as core #2 all the way #7?? It almost looks as if each piece of software is reading each core in a different order.

 

I'm really bummed to not even be able to hit 4.3ghz with water cooling. Do these readings seem normal to others with a similar setup, did I lose the silicone lottery on my cpu? The h100i is only a year old but is it starting to crap out?

 

Screenshot attached. Any input is greatly appreciated.

Deckard

What kind of temps are you getting at relaxed idle on the desktop? I am especially interested in your H100i v2 Temp (coolant temperature) between idle and the load image (~34C).

 

Is this the Hero X motherboard? From what I gather, that one is more up to date with sensor applications, but my Code X for instance is completely unreadable by AIDA and you do need to be careful about putting too much faith in sensor values right now. That said, I do believe you probably are looking 80C+ when fully loaded, so there are some things you can do.

 

1) How are your temps when gaming or doing anything else that is not a synthetic stress test? If the block and CPU are not making good contact, you will get terrible temps at everything from opening browsers to Overwatch. If it is only in the synthetics, it is likely you need to set-up the BIOS to handle those kinds of loads. LLC values are important and so are the IA/DC LLC settings. Even though it is for Kaby Lake, you can get the same information here in the Asus OC guide.

 

2) Your synthetic load delta is way off (+50C core temp over coolant temp). Mine is about +30-35C at 1.30v, but that is delidded. What is your peak Vcore registered in AIDA on those loads? Good chance it is pushing 1.40v, which is certainly not needed. That would be good news, since it would also suggest there is nothing wrong with the cooler and you only need to tune the BIOS.

Thanks for the response! Gaming temps again vary between sensor apps but tend to sit around ~50c but can raise close to 70c for brief periods. Idle temperatures are generally around 35c.

 

That is the Hero X, with the most recent bios version. According to aida64 my cpu core peaks at 1.28v and usually sits around 1.26v under full load. Which again is why I think there's something going on considering the temps I'm getting.

 

I'm going to have a read through the asus guide and see if I can fine tune my bios as you mentioned. Will report back after some tinkering.

Hmmm.... shouldn't be hitting 85-90C at 1.28v. Make sure that is Vcore and not VID. I know AIDA misreports one as other on my board. HWInfo64 appears to be on the mark.

 

Then of course if it is not the auto voltage settings, there may actually be a cooling issue. Given the coolant/core delta is 50C, this would most probably be a contact issue between the CPU and cold plate. Try pressing down on the block with two fingers and see if temps drop. Also check the back to make sure the backplate didn't slip out of one of the holes. Mine was a bit pesky that way.

 

It also could be neither. Gaming load temps of in in the 50's is pretty much standard and those 70 peaks are likely on weird app shader loading moments, something characteristic of many recent games. With a contact issue, those would also hit 90C and I would avg temps to be higher.

 

You really have to watch the auto voltage behavior on the newer CPUs. There is a bit of press about using Asus multi-core enhancement with stock settings and it may cause the CPU to demand even more voltage. That is something that can be turned off in the short run to verify temps and becomes irrelevant when you overclock.

So I went through the asus oc guide in earnest and did a lot of testing, also I've verified that the block is making good contact with the cpu. There's no change really.

 

The only thing I can do to keep temps from going in the red (<100c) is to undervolt to 1.26v which will end up peaking around 1.28v under full load and off setting the avx by a factor of four or five. And that's just to maintain turbo speeds on all cores which is 4.7ghz. So running a synthetic test with the avx offset I'm really only running each core around 4.3ghz and I can just barely keep temps under 100c.

 

I think at this point it may be time to contact corsair directly and see what they have to say about the issue.

That certainly doesn't sound right.

What are the cooler temperatures?

Idle temps on the rad itself are around ~28c. After stress testing or gaming for periods of time it can climb into the 40s. This leads me to think that maybe there's a partial blockage? However I've tried rattling the tubes and radiator to no avail.

A +12C coolant delta is a bit high for a 95 TDP CPU. On a H100i v2, I would expect 6-8C for a 15-30 min run. There are other factors that can contribute to this, including case design, fan layout, etc. Your specs are not detailed, so you might to let us know how the case model and how it is set-up.

 

A partial blockage is possible, but still would not explain the 100C temps at a 40C coolant temp. If it was a blockage, pump failure, or any other condition where the cooler can't move and release heat, the coolant temp would climb in the 40-60C range plus your normal +30-40C CPU to coolant delta. That gets you too 100C. However, your coolant is only at 30-40C which puts the CPU to coolant difference at 60C. That is way too much for any CPU at less than ludicrous voltages. Since we believe that not to be the case and the peak Vcore is in check, then that means there is some kind of contact issue. The only other possible explanation for a discrepancy of that sort would a contact issue under the lid, something that would be difficult to prove. Any stock 8700K should not have a CPU temp to coolant delta of 60C. Even an overclocked one should be several rungs below that.