Why is my 2500k running so hot with a H100i?

Pocah · 31 May 2018 at 17:37

There is still something wrong. Idle temps should be in the thirties not the fifties.
The fact that it is variable does tend to indicate an issue with the pump or an excessive amount of air in the system.
Incidentally, have you checked Task Manager to make sure there isn't some rogue piece of junk software running wildly in the background? I mean I am not saying that is the problem because even so it shouldn't get the CPU to 80+ degrees, but it may explain the variability.

LuckyBenski · 31 May 2018 at 18:08

Have you checked the temperature in some other software like HWinfo or Real Temp?

azibux1 · 31 May 2018 at 23:29

Could look at new paste I guess if it would help - not sure if this stuff has a shelf life?

Regarding task manager I can see Microsoft Office Click to Run SxS with joint highest CPU usage (alongside chrome). Nothing else untoward.

I've been debating a fresh re-install - might just do it!

Temp was checked in Corsair link and NZXT CAM to confirm it is high in both. Just odd how this time last night, doing the same stuff, it was impossible to get it under 79c, now with fans at half the speed I'm at 58c only.

Didn't get round to doing thermal paste today unfortunately

DarkBahamut · 1 Jun 2018 at 13:06

Could be the system was busy during those times, did you note the CPU load? Quite a few background tasks occur when you idle so the CPU isn't always as idle as you think it might be. You pointed out the MS office SxS (side by side updater) which is a prime example of this. Idles of nearly 60c will easily turn into loads of >80c.

Still as other have said the temps are still pretty high. Could be the paste but my paste is >7 years old and has no issues (IC diamond). I'd suspect mounting or pump issues myself. For reference my 2700k on air cooling (NH-D14) is idling at 31c right now, ambient room temps of 24c. Fans running at 750rpm (2 intake, 2 exhaust, 2 CPU - PSU/GPU fans not running).

EsaT · 1 Jun 2018 at 14:15

Thermal paste is either good or bad, not something varying from hour to hour.
Pretty much same for cooling block mounting...
Or maybe there's some tiny gremlin living in your PC screwing around with it...

Akagi · 1 Jun 2018 at 14:26

Don't bother. Re-seat with new compound first. If your CPU usage looks fine I doubt an OS re-install will achieve anything.

Threepwood · 1 Jun 2018 at 14:48

My 2500k @ 4.7 is ideling at 31c with a dark rock pro and.. ic diamond paste. For extra reference

azibux1 · 2 Jun 2018 at 12:31

Threepwood said:
My 2500k @ 4.7 is ideling at 31c with a dark rock pro and.. ic diamond paste. For extra reference

Might tackle this today. Don't think mine has ever been that cool so will really annoy me if I had a dud cooler from day 1 and just put up with it lol

EsaT · 3 Jun 2018 at 09:33

azibux1 said:
Might tackle this today. Don't think mine has ever been that cool so will really annoy me if I had a dud cooler from day 1 and just put up with it lol

Did you disable power saving modes when overclocking?
Those being disabled would rise idle temperatures.

azibux1 · 4 Jun 2018 at 00:15

EsaT said:
Did you disable power saving modes when overclocking?
Those being disabled would rise idle temperatures.

I did notice the other day that it was running at 3.7GHz constantly and not dropping down, tweaked the settings accordingly and will see how it goes.

PaulCa · 4 Jun 2018 at 08:09

NZXT CAM reports my cores at 80*C when idle and 110*C+ when loaded. Yet the AMD Master software reports the package at 41*C idle and 80*C loaded. With a H100x. So I ignore the CAM core temps completely. MSI Afterburner and Argos monitor both coorelate fairly close to the AMD Master package temp, so I trust it more.

I have an x470 and a 2700x and I have read the core temps are offset by 30*C or so and thus read incorrectly.

However I am still concerned about the 80*C and I can get it higher with Prime95 FFT to the point of crashing the overclock due to getting it too hot.

The radiator is barely above ambient.

I intend to check the thermal paste and see if I can get the persistent gurgle that comes up from time to time in the block. Either I didn't use enough paste, or squish it around enough or I have air in my block.

IvanDobskey · 4 Jun 2018 at 15:58

Something is definitely not right there.

My h100i V2 was running warm when I first installed it, but remedied it by switching the fans to intake so cool air from outside was blowing across the radiator. I also changed the supplied thermal pad for some cooler master paste.

My overclocked 6700k runs around 25c idle, 60c when gaming and 78c prime 95 testing.

bartledvd · 4 Jun 2018 at 19:23

My poor old 2600k has been sat at idle 30-35 gaming 55-70c for years with original thermal paste and dark rock pro air so something seems wrong here.

wam7 · 5 Jun 2018 at 11:43

Like others have intimated, I'd be inclined to get a decent air cooler like Noctua, Dark Rock etc. Those H100's were never great coolers without major rpm and noise from the fans.

If they are still running now then they are bound to give even less cooling performance. I'm glad I got rid of mine years ago.

IvanDobskey · 5 Jun 2018 at 14:54

I do think a top end air cooler is better than an AIO. Certainly quieter and more reliable.

They just look so ugly

As soon as I got my case with a tempered glass panel, the huge bulk of my noctua and it's ugly brown fans were a little off-putting.

PaulCa · 11 Jun 2018 at 07:58

I was looking into this over the weekend as I put an AIO on my GPU and it performs a LOT better than the CPU.

It is not a problem with the radiator, the heat does not get out of the CPU. There is simply not enough time to warm the coolant because the CPU goes from 40*C to 60*C in about 3 seconds and then climbs slower towards 70*C in about 20 seconds. The radiator at this stage is cold.

I would suspect there isn't even enough time for the CPU to warm it's IHS at that rise in temp. It has to be the die to IHS thermal transfer that is the issue. (However delidding it does not seem to work much either with only a 4-5*C change - according to YouTuber tests)

I watched half a dozen ryzen overclocking videos on YT and it seems that while ultimate peak temps vary, they all suffer from a dramatic and nearly instant 30-40*C rise in temperature in the die when loaded before the cooler can even come into play.

The weird thing is, why does this not occur with a GPU? Is it because a GPU has no IHS to speak off and the cooler mounts directly onto the die or die casing? My 1070Ti does not see an instant rise in temp, the temp climbs fairly slowly from 34*C to 40*C with the rad fans at max and then rises very slowly over minutes to about 45*C.

I added more thermal paste, even though the spread looked fine, if a little thin, and accounting for the 10*C over read on the temperature with a full overclock (Asus xfr2 tweak Level 3), 4.2Ghz it reaches 60*C pretty much instantly and rises slowly to 70*C and eventually 74*C.

PaulCa · 11 Jun 2018 at 08:14

Oh and to the OP. The NZXT CAM software reports the core temperatures wrong. When I had CAM installed it popped up to say my cores had exceeded 100*C. When running Prime95 I seen them up at 110*C.

Actually I have found about 50% of monitoring apps report the 2700x temperatures wrong, some times wildly. I believe this might be the 2700X or the 470 chipset compatibility.

Further the 2700X reports it's package temp with +10*C.

If you want an accurate temp use the Ryzen Master software.

martin363 · 11 Jun 2018 at 15:49

PaulCa said:
Oh and to the OP. The NZXT CAM software reports the core temperatures wrong. When I had CAM installed it popped up to say my cores had exceeded 100*C. When running Prime95 I seen them up at 110*C.

Actually I have found about 50% of monitoring apps report the 2700x temperatures wrong, some times wildly. I believe this might be the 2700X or the 470 chipset compatibility.

Further the 2700X reports it's package temp with +10*C.

If you want an accurate temp use the Ryzen Master software.

whilst i think helpful the op has an intel sandybridge chip.

Orcvader · 11 Jun 2018 at 16:26

My old H40 still works and kept my overclocked 2600K under 70c most of the time.

Another thing to check is where the pump is plugged in, if it's on the motherboard make sure the fan header it's connected to is set to full speed/power. Alternatively get a molex adapter so the pump is connected directly to the PSU.

MagicBoy · 11 Jun 2018 at 16:41

H100i takes power for the pump direct from a drive power connector. Fan header is only used to report RPM back.