4770k, z87x-oc, high temps on water??

[Ewok]

Ok slightly worried about my temps now after seeing someone else, on air, with higher voltage, seemingly getting better temps than me!

I have just built a 4770k, gigabyte z87x-oc with an ek supremacy block, a gtx 780 with xspc block all in a single loop with 320 and 240 rads and a D5 vario. My first attempt at overclocking it was simply to up the multiplier to 43 for 4.3ghz. The second I started prime95 small fft's the temp shot up to 100c. I noticed it had shoved the voltage up to 1.39v as well. I quickly shut it down and since then have restricted the voltage to 1.16v as I want this thing to last 3-4 years not 3-4 days! Now still usnig x43 multiplier and restricting voltage to 1.16v I am still hitting temps of just under 80c when under a small fft's prime95 test. I havent really looked at pc stuff for 4 years since I build my last watercooled setup which lasted quite nicely, but I am starting to get the impression something is seriously wrong if a guy on air using 1.272v is getting 74c with the same cpu and same mb on AIR.

Does something seem wrong here? What kind of temps should I be looking at with this setup at say 1.16v (or any other voltage). The graphics card in the same loop doesnt seem to have a problem with temps at all, I have oc'd that to 1228mhz core and 3348mem and using msi kombustor to stress it with various tests its barely breaking 40c!.

[GliderHR]

Can you report the water temperature and the fan speed?
Your radiatorsurface should be sufficent.
Did you check the contact of the cpu block, what thermal paste did you use?
Maybe a picture of your system could help.

[m_jones_]

Sounds like it's time to delid your chip, you will get much better temps.

[Jakusonfire]

Likely just the overclock settings need work. Modern motherboards, even if in the BIOS it seems like you have 'set' it to 1.16V, often up the voltage much higher if using one of many different auto settings.

Use CPUz to look at the volts it is reporting at load. Lower the PLL? A lower LLC? There are lots of things to look at.

As you point out, the GPU would be getting hot too if it had to do with the loop, unless the CPU mount is bad.

[Garrett]

Sounds like it's time to delid your chip, you will get much better temps.

This!
Lots of videos around of how to do it, plus in some cases people gained 25+ degrees.
Should be a lot of fun, make sure to make a video of your delidding

[Ewok]

ok pic attached. my swiftech mcres v2 is sitting in the 5.25 drive bay on the right so you cant see it



fan speeds I have set to 60%. Varying the D5 pump speed makes very little difference to the temps. I used Gelid Extreme paste that came with the block (although I also bought some for the gpu). The voltage being reported by occt is the voltage that I set in the bios, I took it off auto to set this voltage, without doing that it decided it was a good idea to run 1.39v through it when i set it to 4.5ghz by changing the multiplier to 45. I know nothing about PLL or LLC, I just set vcore to 1.16 as this seemed to at least keep the temp below 80c. cpu-z reports vcore at 1.157, occt reports 1.18 and thats running a small fft prime95. temps go straight to 75-79c within a few seconds and hover around there. Idle temps are around 33-34c.

Either its normal, theres some other voltage that is causing these temps rather than vcore? or theres some issue perhaps with the mounting, although its pretty straight forward to mount it, 4 screws with springs and you just tighten it until it doesnt go anymore, perhaps it isnt quite on right but the temps seem extremely high for the voltage i am using considering people on air are apparently getting better temps with higher voltage!

[Ewok]

Right I spent the last 3 hours reading about delidding and liquid pro/ultra. I have just ordered some liquid ultra which I plan to use between ihs and the die, as well as between the ihs and the EK block. When it arrives on Tuesday I shall go for a delid. I figure I am going to have to unmount it anyway to try to re-seat it and I may as well delid it if its coming out anyway. I just need to do it without draining the loop cos I am not going through all that hassle again. hopefully I can just take the block off, whip the chip out, do my slicing and stick it back in to be presented with a nice 40c drop

At this point I figure these silly temps are either due to bad contact somehow with the block, or due to a very bad ihs contact with the die, both of which should be solved when I delid.

Right now I can get 4.2ghz stable with 1.16v (showing as maxing 1.18v in occt). I am not sure what constitutes a safe long term voltage on these, I want 4 years out of it so I certainly wont be whacking 1.4v through it. I am guessing 1.25 might be ok?

[andressergio]

ok pic attached. my swiftech mcres v2 is sitting in the 5.25 drive bay on the right so you cant see it

thats remembers me ma old rig when the tygons got blured due to plasticizer, that is very bad as all dust from plastic gets stucked inside the tubes, another suggestion is to put your fans pushing fresh air not pulling the hot trapped in that little space

u have to use LIQUID PRO, not ULTRA but it works fine either way

try to remove those tubes cauz inside the blocks you will find hell

Cheers !
Sergio

[Conumdrum]

What fans? Hopefully good ones. That front fan is really not helping much. No fan on the bottom either.

What did you have before?

It looks like you have enuff rad for sure, plenty, fans matter somewhat. But just loading the CPU, temps shouldn't be that high. Maybe you need to work the overclock settings. Remount the block?

The guy who is getting 74C and your getting under 80C now isn't that bad. His room could be 5C cooler. Every C in room temps is 1C in CPU temps. And are you sure the guy's temp checks are done using the same exact program as you? Running the test for the same amount of time? How much rad does he have, what block, what paste, his fan setup? It's hard to compare exactly.

And that CPU is HOT HOT, and you could of got a lemon. Good luck. Sorry not much help really, you'll work the temp down I'm sure.

Lastly, could you make a sig of what you have in your setup please? Focus on the hot parts and the cooling parts. Having to find what you have within your posts is tedious and hard for me to follow, meaning less support

[Ewok]

What fans? Hopefully good ones. That front fan is really not helping much. No fan on the bottom either.

What did you have before?

The guy who is getting 74C and your getting under 80C now isn't that bad. His room could be 5C cooler.

The left fan is an exhaust, blowing onto the 2 fans that are behind that rad which are blowing through the rad and away from the pc. There is a 230mm 110cfm fan blowing in from the front at the bottom which came with the case (coolermaster HAF932) and another of the same fan which sites inside the side panel blowing in over the board.

What I had before was pretty much the exact same loop with different blocks on an I7 920 and a GTX295 which I suspect was putting out more heat than this new setup is with only 1 gpu.

Hes getting 74c.....on AIR with a higher voltage as reported by cpu-z and occt and being in the UK as well, chances are the room is about the same temp. Theres a huge delta between what I should be getting with this voltage on water and what I am actually getting. with 1.16v I dont expect the thing to be going over about 60c, which fits right in with the ~20c drop I expect from a delid.

Anyway I am certain no issue with the cooling, its either the cpu, or the block mounting or both. Both of which should be fixed with a delid and some liquid ultra (which my research tells me is better than pro for temps, and much better than pro for spreading) which wil replace the Gelid Extreme that I have on it now, although the Gelid has been tested as second only to the liquid pro/ultra so its pretty good paste anyway.

Slightly confused as to why people dont seem to be using the liquid pro/ultra between the ihs and their block/heatsinks as well as between the die and the ihs. If its the best between die and ihs surely its the best between any 2 things. I assume its because it is conductive? Which still makes no sense to me, you arent going to put enough on that it actually RUNS off the thing and onto something else, being conductive really is irrelevant.

[Makymaco]

...There is a 230mm 110cfm fan blowing in from the front ... another of the same fan which sites inside the side panel blowing in ...

Seems good to me.

Perhaps redo your thermal paste. How much do you use ? What method do you use to apply it ?

[Makymaco]

Can you test your flow ? Perhaps the blocks are not clean and the flow is poor or there is something obtructing the flow in the cpu bloc.
When you have time, open your cpu block and check if anything is obstructing. Those blocks have jet plates whith tiny holes so, it is easy
to obstruct the flow (it happened to me). But even with bad flow, the temps can be good.

[Ewok]

flow is fine or the gpu in the same loop would be getting hot and thats topping out at just over 60c even at 1228 gpu, which is around what I would expect the cpu to be doing. The flow meter is reporting 80lph via the lamptron with the pump set pretty low, but the koolance flow meter converter reports lpm not lph so I'm not really sure what the 80 actually is, it cant be lph as that would give 1.3lpm and its nowhere near that slow, it cant be lpm as that would give 4800lph so i dont really know what the number means. either way the temp difference between low speed and high speed on the pump is very small.

[Makymaco]

Then : Perhaps redo your thermal paste. How much do you use ? What method do you use to apply it ?

[Ewok]

the Gelid comes with a little plastic spreading device so I used that. it was a thin layer and no way would that have caused temps this high. its either not quite mounted right somehow, or its just a really poor heatspreader, which will soon be sorted!

[Makymaco]

A very thin layer ?
Try only a thin short line like a grain of rice at the center

[Jakusonfire]

flow is fine or the gpu in the same loop would be getting hot and thats topping out at just over 60c even at 1228 gpu, which is around what I would expect the cpu to be doing. The flow meter is reporting 80lph via the lamptron with the pump set pretty low, but the koolance flow meter converter reports lpm not lph so I'm not really sure what the 80 actually is, it cant be lph as that would give 1.3lpm and its nowhere near that slow, it cant be lpm as that would give 4800lph so i dont really know what the number means. either way the temp difference between low speed and high speed on the pump is very small.

Is that 60C figure for the GPU correct? earlier it was a more reasonable 40C

What is the ambient temp of the room? That loop should easily be able to keep the water temp at or under 10C above ambient. The GPU core temps should be only 10 - 12C at most above the water temp.

[Ewok]

oops, I meant 40 ambient is around 21c, I do not know the water temp but I can probably stick a probe into it if I take the cap out of the fill hole of the res.