Good block contact, low idle temps. Extreamely high load temps. Advivce ?

KillBoY_UK · 18 Apr 2015 at 14:09

Carrying on from the 5820k overlocking thread where I was getting very high load temps (http://forums.overclockers.co.uk/showpost.php?p=27913655&postcount=146)

So today I reseated the CPU, contact was good and just the right amount of compound being used. So I reapplied and booted her up. Yet I still have extremely high temps.

Mid to late 30's at idle at 3.8ghz and high 60's low 70'sunder prime and about 10c less on real bench :-/. Still one bit of odd behaviour. If i use the 4ghz profile on my borad (which applys a Vcore of 1.21v) idle temps are LOWER in the low 20s but still shockingly high under load (80c).

Trying the 4.4ghz profile (1.3vcore) i still get the low normal temps but as soon as you fire up prime it quickly hits the high 90's then low 100's and throttles. :confused:

So I have good contact with the CPU and waterblock and the water gets warm. So I'm relay confused in what to try next. Go to the original inserts for the block and not the ones called out in the manual for LGA2011. Could it be a faulty temp sensor somewhere, as the system is rock stable.

Ayahuasca · 18 Apr 2015 at 14:39

Doesn't prime cause crazy temps on the newer chips?

I've seen reports that custom water isn't really better than AIO/top end air coolers on the newer Intel CPU's as they still can't dissipate the heat from the poor TIM quick enough.

This could just be a complete guess and 100c does sound way too high for water.

MrMD · 18 Apr 2015 at 14:46

What are your temperatures like in realbench when the system is set to stock in the bios?

KillBoY_UK · 18 Apr 2015 at 16:12

MrMD said:
What are your temperatures like in realbench when the system is set to stock in the bios?

Temps only got a couple of degrees higher after a hour

Mick-1965 · 19 Apr 2015 at 23:05

I take it the 780's are not in the loop ?

You say the water is getting warm, is it roughly the same temp around the whole loop. ie before and after your CPU block.
The fact the water is getting warm indicates you have good cpu/block contact, but if you are significantly hotter on the out side of your cpu block would point to a flow rate issue.

What is you ambient temp ? and assuming the temp is equal around the loop, what temp is your water actually getting to ?

andybird123 · 19 Apr 2015 at 23:43

Have you got the in and out connected up the right way on the cpu block?

Freddie1980 · 20 Apr 2015 at 08:10

If the water is getting warm you may have a trapped air the loop still. It's worth checking out as the hotter the water gets the softer the tubes become which can lead to them coming off the compression fittings.

ubersonic · 20 Apr 2015 at 08:58

Is the air coming out of the rad warm or hot?
Are you running your fans at full speed? (if not does that have any effect on temps?)
Are you running your D5 at full speed? (if not does that have any effect on temps?)
As another guy said is the tubing as warm around the whole loop or noticeably hotter/colder in parts? (means not enough flow rate).

Undesirable · 20 Apr 2015 at 17:41

I reckon if you delid and apply liquid metal then you will not have this issue. I believe the processor's voltage increases by 0.1 when you run programs that use AVX instructions unless you force a set voltage. My processor has this issue and I'm on custom water and not delidded.

AndoriAvalonia · 20 Apr 2015 at 23:58

Undesirable said:
I reckon if you delid and apply liquid metal then you will not have this issue. I believe the processor's voltage increases by 0.1 when you run programs that use AVX instructions unless you force a set voltage. My processor has this issue and I'm on custom water and not delidded.

The OP is using a 5820k, which I thought had a soldered IHS rather than using a paste so delidding would have a minimal effect.

ErikN · 21 Apr 2015 at 08:21

AndoriAvalonia said:
The OP is using a 5820k, which I thought had a soldered IHS rather than using a paste so delidding would have a minimal effect.

Delidding it would have the effect of (most likely) destroying the CPU, which some could consider major..

On topic (for comparison):
I have a [email protected] 1.25v, and I applied thermal paste sort of sloppy last time I set the waterblock, and my cores max out at between 62 and 67 degrees when running OCCT.

Could you check temps running that?

Also, there is not the old issue of adaptive voltages? So the vcore is increased to unhealthy levels during prime95 loads?

stujim · 18 Jun 2015 at 17:02

Ayahuasca said:
Doesn't prime cause crazy temps on the newer chips?

I've seen reports that custom water isn't really better than AIO/top end air coolers on the newer Intel CPU's as they still can't dissipate the heat from the poor TIM quick enough.

This could just be a complete guess and 100c does sound way too high for water.

the latest version of Prime95 isn't recommended for Haswell chips (according to a post on the intel haswell forums). i personally use version 26.6 build 3. this gives it a good test without going mental. i've only ever ran it on the blend test