i4 4770k Custom Watercooling Temps

[Theo]

My setup is: a 360mm and 280mm, 60mm thick rads and the fans are running at 600rpm

The 4770k has a small OC to 4.2GHz at 1.123v and the gpu is a 980ti but it doesnt really matter here.

While running Aida64/OCCT the CPU gets to around 70C  (no gpu load) and for all that rad space that looks a bit high. Even if i reset my OC, the cpu still gets to 60c. 

This is normal or do i need to reapply thermal paste or something else?

[SeanAngelo]

1 minute ago, Theo said:

My setup is: a 360mm and 280mm, 60mm thick rads and the fans are running at 600rpm

The 4770k has a small OC to 4.2GHz at 1.123v and the gpu is a 980ti but it doesnt really matter here.

While running Aida64/OCCT the CPU gets to around 70C  (no gpu load) and for all that rad space that looks a bit high. Even if i reset my OC, the cpu still gets to 60c. 

This is normal or do i need to reapply thermal paste or something else?

increase rpm speeds?

 

what fans are they?

[W-L]

4 minutes ago, Theo said:

My setup is: a 360mm and 280mm, 60mm thick rads and the fans are running at 600rpm
The 4770k has a small OC to 4.2GHz at 1.123v and the gpu is a 980ti but it doesnt really matter here.
While running Aida64/OCCT the CPU gets to around 70C  (no gpu load) and for all that rad space that looks a bit high. Even if i reset my OC, the cpu still gets to 60c. 
This is normal or do i need to reapply thermal paste or something else?

Try reseating your waterblock to ensure it's not a bad mount as for rads which ones are you using at 60mm thick if they are also high FPI your fan may not be effectively pushing air through the rads. 

[Theo]

On 10/31/2016 at 3:28 PM, SeanAngelo said:

increase rpm speeds?

 

what fans are they?

Im comparing this to my old H100i which got around 80c with same OC. They are silverstone ap121's and they are good for rads 

 

On 10/31/2016 at 3:33 PM, W-L said:

Try reseating your waterblock to ensure it's not a bad mount as for rads which ones are you using at 60mm thick if they are also high FPI your fan may not be effectively pushing air through the rads. 

Rads are blackice SR2's that have low FPI (7 if i remember correctly) so they are optimised for low rpm fans.

Reseating block is a possibility but i cant imagine how someone i can do something different ( its an EK block and did exactly what it said)

[done12many2]

Is 600 RPM your max speed for those fans?  

 

I'm going to assume that you aren't monitoring water temps.  You may want to throw a water temp sensor in your loop as it's the best way to ensure that your loop is running efficiently.  

[W-L]

6 minutes ago, Theo said:

Rads are blackice SR2's that have low FPI (7 if i remember correctly) so they are optimised for low rpm fans.
Reseating block is a possibility but i cant imagine how someone i can do something different ( its an EK block and did exactly what it said)

Sometimes remounting can help if it's not applying enough pressure or uneven pressure, do you have a photo of the loop that usually can help a lot to diagnose issues.

[SeanAngelo]

19 minutes ago, Theo said:

Im comparing this to my old H100i which got around 80c with same OC. They are silverstone ap121's and they are good for rads 

increase fan speeds to 1200rpm and see if it makes a difference in temps.

[done12many2]

3 minutes ago, SeanAngelo said:

increase fan speeds to 1200rpm and see if it makes a difference in temps.

 

That seems like the best thing to do before tearing things apart. 

[Theo]

 

44 minutes ago, done12many2 said:

Is 600 RPM your max speed for those fans?  

 

I'm going to assume that you aren't monitoring water temps.  You may want to throw a water temp sensor in your loop as it's the best way to ensure that your loop is running efficiently.  

They go higher but i have them set to that just for them being quiet. and you are right I will buy one for the next time that I need to empty the loop

42 minutes ago, W-L said:

Sometimes remounting can help if it's not applying enough pressure or uneven pressure, do you have a photo of the loop that usually can help a lot to diagnose issues.

http://imgur.com/a/mq4CD

I screwed them on diagonally and all

31 minutes ago, SeanAngelo said:

increase fan speeds to 1200rpm and see if it makes a difference in temps.

Ill try that again.

It may be the case that the software causes more volts to go to cpu, HWmonitor showed 1.18v to the cpu whereas it should be 1.12

[Theo]

49 minutes ago, SeanAngelo said:

increase fan speeds to 1200rpm and see if it makes a difference in temps.

Well setting everything to 100% (fans and pump) made no difference to temps 

[W-L]

27 minutes ago, Theo said:

They go higher but i have them set to that just for them being quiet. and you are right I will buy one for the next time that I need to empty the loop

http://imgur.com/a/mq4CD
I screwed them on diagonally and all

Ill try that again.
It may be the case that the software causes more volts to go to cpu, HWmonitor showed 1.18v to the cpu whereas it should be 1.12

Hmm that looks good, odd why you would be getting such high temps, did varying the fan speed not lower the temps at all?

[Theo]

3 minutes ago, W-L said:

Hmm that looks good, odd why you would be getting such high temps, did varying the fan speed not lower the temps at all?

it made no difference, even set the pump to 100%. Flow rate is good because when i filled it it was like opening a tap.

Also GPU temps are fine, 10 minutes in Heaven temps went slowly from 30c to 40c and fan speed seems to have an effect on it.

[0ld_Chicken]

I'd reinstall your block and make sure to check both the block and the IHS to see how the current paste is spread.  That can tell you a lot about what kind of contact the block is making with the IHS of the cpu.   It should be spread even and thin all around after being removed. 

 

If you've tried everything else and you really want a solution, think about delidding that chip. 4770k's were notoriously bad for having temp issues due to the Thermal Interface Material between the die and IHS.  Most 4770k delidders saw a 15-25°c drop in temps (higher core voltages usually yield higher temp drops) 

 

Obviously it is somewhat dangerous and you could damage your cpu if you aren't careful.  If you take your time and do plenty of research, you shouldn't have an issue.  I delidded my 4790k and a G3258 with a vice only and it was very simple.  If you want to know more or need help with anything let me know, I'd be glad to help.  

 

I left a link below to OCN's delid club, which is a great source of info, but it's also very overwhelming.  Using the "search this thread" tool at the top makes things a bit easier. 

http://www.overclock.net/t/1313179/official-delidded-club-guide/0_60

[W-L]

3 hours ago, Theo said:

it made no difference, even set the pump to 100%. Flow rate is good because when i filled it it was like opening a tap.

Also GPU temps are fine, 10 minutes in Heaven temps went slowly from 30c to 40c and fan speed seems to have an effect on it.

Try doing a remount, that is probably your best bet as a right now to see if it makes a big difference in temps, all that sounds normal so that would really be the only thing left that would really cause an issue. 

[Theo]

10 hours ago, 0ld_Chicken said:

I'd reinstall your block and make sure to check both the block and the IHS to see how the current paste is spread.  That can tell you a lot about what kind of contact the block is making with the IHS of the cpu.   It should be spread even and thin all around after being removed. 

 

If you've tried everything else and you really want a solution, think about delidding that chip. 4770k's were notoriously bad for having temp issues due to the Thermal Interface Material between the die and IHS.  Most 4770k delidders saw a 15-25°c drop in temps (higher core voltages usually yield higher temp drops) 

 

Obviously it is somewhat dangerous and you could damage your cpu if you aren't careful.  If you take your time and do plenty of research, you shouldn't have an issue.  I delidded my 4790k and a G3258 with a vice only and it was very simple.  If you want to know more or need help with anything let me know, I'd be glad to help.  

 

I left a link below to OCN's delid club, which is a great source of info, but it's also very overwhelming.  Using the "search this thread" tool at the top makes things a bit easier. 

http://www.overclock.net/t/1313179/official-delidded-club-guide/0_60

 

9 hours ago, W-L said:

Try doing a remount, that is probably your best bet as a right now to see if it makes a big difference in temps, all that sounds normal so that would really be the only thing left that would really cause an issue. 

I might remount it but i am seriously considering delidding it as well, what tim did u replace it with?

The one thing that i already changed so far is the voltage and forced it to 1.12v(manual voltage) instead of it going to 1.18(adaptive) and temps in aida64 after 1h were around 54c at 600rpm.

From what I've gathered so far it seems like Haswell is being haswell with shit tim.

Thanks for the replies so far btw

EDIT: forgot to also say that i tried 1.3v for 4.7ghz and temps got to around 80c under aida64 in 10 minutes ( in case this helps with anything)

[done12many2]

1 hour ago, Theo said:

The one thing that i already changed so far is the voltage and forced it to 1.12v(manual voltage) instead of it going to 1.18(adaptive) and temps in aida64 after 1h were around 54c at 600rpm.

EDIT: forgot to also say that i tried 1.3v for 4.7ghz and temps got to around 80c under aida64 in 10 minutes ( in case this helps with anything)

 

The .06v difference will not impact temps much at all.  

 

Just to clarify your temps when running AIDA64, what tests are you running?  If your 4.7 at 1.3v hit 80c with CPU, FPU and cache only being tested, I'd say your temps are a little on the warm side.  If those temps were with FPU only, you're doing really good. 

 

In addition to running AIDA64 (CPU, FPU, and cache), try running RealBench 2.44 for at least an hour and OCCT (CPU:OCCT) for another hour.  RealBench will give you a good idea of how well your loop is displacing the heat as RealBench heats up the CPU and GPU at the same time.  OCCT is a much harder CPU stress test than AIDA64.  Don't be surprised if you AIDA64 stable overclock crashes in OCCT.

 

Add the water temp sensor as soon as you can.  It's better to deal with the pain of draining and refilling fluid now then to go months or longer without actually knowing your water temps.  You'll start to learn how fan speed impacts your cooling.  You may learn that too fast of a fan speed does nothing to improve water temps or that too low allows the water to heat soak. 

 

It's so much easier to diagnose temp issues when you know the temp of your water.  Your current issue being an example.  If your water was only getting a few degrees above ambient, you would know that your cooling issues are elsewhere and you can rule out everything in your loop to include fans.  That would leave you with checking block mounting and other possible issues.

[W-L]

2 hours ago, Theo said:

I might remount it but i am seriously considering delidding it as well, what tim did u replace it with?
The one thing that i already changed so far is the voltage and forced it to 1.12v(manual voltage) instead of it going to 1.18(adaptive) and temps in aida64 after 1h were around 54c at 600rpm.
From what I've gathered so far it seems like Haswell is being haswell with shit tim.
Thanks for the replies so far btw
EDIT: forgot to also say that i tried 1.3v for 4.7ghz and temps got to around 80c under aida64 in 10 minutes ( in case this helps with anything)

As 0ld_Chicken said delidding might be something to look into, if the remount makes no difference. For 1.3V that is on the hot side with the cooling solution you have, some good options for thermal paste are MX-4, thermal grizzly, and Gelid extreme to name a few. If you want to use the liquid metal stuff that is really good but just be careful when applying to not get it onto any contacts or capacitors that are soldered beside the die. 

[Theo]

14 hours ago, done12many2 said:

 

The .06v difference will not impact temps much at all.  

 

Just to clarify your temps when running AIDA64, what tests are you running?  If your 4.7 at 1.3v hit 80c with CPU, FPU and cache only being tested, I'd say your temps are a little on the warm side.  If those temps were with FPU only, you're doing really good. 

 

In addition to running AIDA64 (CPU, FPU, and cache), try running RealBench 2.44 for at least an hour and OCCT (CPU:OCCT) for another hour.  RealBench will give you a good idea of how well your loop is displacing the heat as RealBench heats up the CPU and GPU at the same time.  OCCT is a much harder CPU stress test than AIDA64.  Don't be surprised if you AIDA64 stable overclock crashes in OCCT.

 

Add the water temp sensor as soon as you can.  It's better to deal with the pain of draining and refilling fluid now then to go months or longer without actually knowing your water temps.  You'll start to learn how fan speed impacts your cooling.  You may learn that too fast of a fan speed does nothing to improve water temps or that too low allows the water to heat soak. 

 

It's so much easier to diagnose temp issues when you know the temp of your water.  Your current issue being an example.  If your water was only getting a few degrees above ambient, you would know that your cooling issues are elsewhere and you can rule out everything in your loop to include fans.  That would leave you with checking block mounting and other possible issues.

Yeah that would make sense but temps changed only by that small voltage increase. In aida i just leave the default tests, CPU,FPU, cache, system memory. But i see what what u mean with FPU only,ill test this tonight again. I also have OCCT with linpack running atm and temps are around 63c ( only running for 5 min for the moment) If its anything useful, before starting OCCT the gpu was at 23c and after 5 min it went to 26c with no load on it so the water should be around that temp i guess(?)

You are right for the water sensor, i will probably do it a lot sooner than i expected, draining and refilling loop isnt that much of a hassle.But i could probably ghetto mount a temp probe from my fan controller on the top of the res for the moment to diagnose.

I also saw some vids on vice grip method for delid and it looked surprisingly quick and easy.

 

13 hours ago, W-L said:

As 0ld_Chicken said delidding might be something to look into, if the remount makes no difference. For 1.3V that is on the hot side with the cooling solution you have, some good options for thermal paste are MX-4, thermal grizzly, and Gelid extreme to name a few. If you want to use the liquid metal stuff that is really good but just be careful when applying to not get it onto any contacts or capacitors that are soldered beside the die. 

I will check what others with custom wc and a 4770k have gotten in terms of temperature. But ill probably remount the block in the next few days, good thing i have flexible tubing for this and i dont need to drain anything.

Also if i delid, i saw that EK has screws for direct dye cooling which should be even better than replacing the tim between the IHS and cpu

[W-L]

Just now, Theo said:

I will check what others with custom wc and a 4770k have gotten in terms of temperature. But ill probably remount the block in the next few days, good thing i have flexible tubing for this and i dont need to drain anything.

Also if i delid, i saw that EK has screws for direct dye cooling which should be even better than replacing the tim between the IHS and cpu

Yes, ideally if you are able to get a delid guard of sorts like on MSI motherboards that would be best. Just one last thing though how long have you've been running the fluid and setup for, if it's been a really long period of time it's not uncommon to have gunk or potential build up in channels of the blocks. 

[Theo]

1 hour ago, W-L said:

Yes, ideally if you are able to get a delid guard of sorts like on MSI motherboards that would be best. Just one last thing though how long have you've been running the fluid and setup for, if it's been a really long period of time it's not uncommon to have gunk or potential build up in channels of the blocks. 

i set up the loop 1 week ago, everything new. It could be possible that the cpu loop has gunk inside. I did flush the rads very well before in case you were going to ask

[0ld_Chicken]

3 hours ago, Theo said:

I will check what others with custom wc and a 4770k have gotten in terms of temperature. But ill probably remount the block in the next few days, good thing i have flexible tubing for this and i dont need to drain anything.

Also if i delid, i saw that EK has screws for direct dye cooling which should be even better than replacing the tim between the IHS and cpu

When delidding you've got a few options. First of all, ALWAYS use a liquid metal TIM on the die (PHOBYA LM or clu/cpu).  Most regular pastes will seem fine at first and then the TIM will pump out and your temps will rise dramatically.  You can use any regular paste between the ihs and block if you choose to go that route.

 

When I first delidded, I reused the ihs since it was easier. Since then I got that ivy bridge naked mount kit you referenced and dropped another good chunk of degrees.  I did not use a delid guard or anything, just set the block straight on the die and I'm extremely happy with the results.  I can go up to 1.6v without throttling (only tried a couple times)

[W-L]

5 hours ago, Theo said:

i set up the loop 1 week ago, everything new. It could be possible that the cpu loop has gunk inside. I did flush the rads very well before in case you were going to ask

Yeah it's very unlikely to cause the problem then. 

Edited November 1, 2016 by W-L

[Theo]

 

1 hour ago, 0ld_Chicken said:

When delidding you've got a few options. First of all, ALWAYS use a liquid metal TIM on the die (PHOBYA LM or clu/cpu).  Most regular pastes will seem fine at first and then the TIM will pump out and your temps will rise dramatically.  You can use any regular paste between the ihs and block if you choose to go that route.

 

When I first delidded, I reused the ihs since it was easier. Since then I got that ivy bridge naked mount kit you referenced and dropped another good chunk of degrees.  I did not use a delid guard or anything, just set the block straight on the die and I'm extremely happy with the results.  I can go up to 1.6v without throttling (only tried a couple times)

If i decide to delid which im really considering since the vice method seems extremely easy ill probably go straight naked mount kit 

1 hour ago, W-L said:

Yeah it's very unlikely to not cause the problem then. 

I will remount the block tomorrow and ill comment here again. I am guessing i might not have put enough tim.

Thanks again for replies. 

[Thx And Bye]

I can only tell you the story of my i7-3770k. I have a EKWB Supremacy Full Nickel and two aquacomputer 360mm airplex modularity copper radiator.

Also I have two 680s in my loop.

 

I also always had excellent GPU temperatures and my liquid temperature also was fine all the way.

It also seemed really odd to me that the CPU would be on about 70°C even on stock voltage and clock.

I decided to delid the CPU right after the warranty was over since it wouldn't matter anymore anyways. (vice method)

 

Next thing was, that i tied the Naked Ivy mounting kit, since I delided anyways. But that was where the problems started: one RAM channel refused to work and Windows would only boot if that complete channel was empty.

Turned out that the mounting pressure wasn't high enough and some pins hadn't enough contact.

So I used some Noctua NT-H1 and put the IHS back on, giving me my lost RAM channel back. Even tough I lowered my CPU temps about 10°C overall.

 

That still wasn't really satisfying since the CPU still would run fairly warm but bare die didn't seem like an option for me.

 

The last thing that finaly solved the issue for me was using coollaboratory liquid ultra between die and IHS. I wouldn't recommend it between IHS and the liquid cooler since it will affect the cooler even if it might only be ascetics (who knows, I wouldn't risk) and i plan to use mine long in the distand future.

Well since I'm using coollaboratory liquid ultra I can OC to 4.4GHz without voltage increase and still run about 30°C cooler then bone stock.

 

I'm not sure if the liquid metal tim will attack the die over time but even if it will probably time to upgrade the then CPU anyways.

Also I can't guarantee that this will apply 1:1 to your CPU but the chance is rather high.

 

TL;DR: Same problem with 3770k -> delid -> coollaboratory liquid ultra between die and IHS & Noctua NT-H1 beween IHS and liquid cooler -> OC + 30°C lower temps -> ??? -> win

[Simon771]

4770k have bad TIM under IHS, just like my i7 4790k

 

So the only solution would be to delidd that CPU. 

I'm afraid of it, so I won't do it. But I hope you are braver than me, so you can share your results when you successfually delidd your CPU