8700k High Temperatures

[Mavflight09]

Hey guys, I was wondering if the temperatures I am getting with my H115i was normal.

 

I have an 8700k, delided, die lapped, applied liquid metal between die and ihs, and liquid metal between ihs and cold plate of the cooler. The cooler is an H115i from Corsair, with 4 sp140, 140mm static pressure fans in push-pull, the fans and pump are all running at max speed.

Theoretically I should get some ballin' temperatures, but at 5.1 ghz, 1.392 vcore my temps still get up to 83 degrees C. This is with Prime95 small FFT's with AVX disabled in Prime. 

I got these temperatures in a NZXT H440 with the front panel on, but with the dust filter removed for better airflow.

 

 

What do you guys think? Is this normal for a 5.1 Ghz overclock? Or is this too hot?

[Jurrunio]

RIP THAT GHASTLY FRONT PANEL OFF

[Mavflight09]

1 minute ago, Jurrunio said:

RIP THAT GHASTLY FRONT PANEL OFF

I thought of that too, and with the front panel off, along with the dust filter off, I got marginal improvements, about 5 degrees C, resulting in about 77 degrees max.

 

However it looks very unsightly and you could easily stick your fingers into the fans by accident.

 

It would definitely work, but its not the greatest solution.

 

I thought about doing a front panel mod, where I could cut a hole into the front panel and make a custom mesh grill, but it doesn't seem worth it for the marginal temperatures improvements that I would get.

 

Here is a test I did with all of the case panels removed:

 

[Jurrunio]

6 minutes ago, Mavflight09 said:

I thought of that too, and with the front panel off, along with the dust filter off, I got marginal improvements, about 5 degrees C, resulting in about 77 degrees max.

 

However it looks very unsightly and you could easily stick your fingers into the fans by accident.

 

It would definitely work, but its not the greatest solution.

 

I thought about doing a front panel mod, where I could cut a hole into the front panel and make a custom mesh grill, but it doesn't seem worth it for the marginal temperatures improvements that I would get.

 

Here is a test I did with all of the case panels removed:

Then that's about it. I mean, 1.4V is rather high. People get 60-65C temperature at some 1.3V 5GHz overclock after delidding.

[Mavflight09]

2 minutes ago, Jurrunio said:

Then that's about it. I mean, 1.4V is rather high. People get 60-65C temperature at some 1.3V 5GHz overclock after delidding.

So you think these temperatures are reasonable?

 

I guess they will work, it's just, I expected better temperatures from a 280mm rad, in push pull, pulling nice clean, fresh air from outside of the case.

 

Do you think there will be any improvement if I were to move to a custom loop, outside of the case?

 

As in, a radiator mounted outside of the case with tubes bringing the liquid in and out of the case.

[Jurrunio]

3 minutes ago, Mavflight09 said:

So you think these temperatures are reasonable?

 

I guess they will work, it's just, I expected better temperatures from a 280mm rad, in push pull, pulling nice clean, fresh air from outside of the case.

 

Do you think there will be any improvement if I were to move to a custom loop, outside of the case?

 

I think a top performing 360mm aio should help get temps a bit lower (or same temps and less noise), like the Arctic Freezer 360.

 

Custom loop dont increase performance unless you use more radiators. It's definitely more trouble for little performance gain compared to a good 360mm aio

5 minutes ago, Mavflight09 said:

As in, a radiator mounted outside of the case with tubes bringing the liquid in and out of the case.

If your case has good airflow (enough intake fans and exhaust fans), then the difference between radiator hanging outside and inside (without air filters and front air restrictor plate) will be small.

[Mavflight09]

2 minutes ago, Jurrunio said:

I think a top performing 360mm aio should help get temps a bit lower (or same temps and less noise), like the Arctic Freezer 360.

 

Custom loop dont increase performance unless you use more radiators. It's definitely more trouble for little performance gain compared to a good 360mm aio

If your case has good airflow (enough intake fans and exhaust fans), then the difference between radiator hanging outside and inside (without air filters and front air restrictor plate) will be small.

Gamer's Nexus did a review of the NZXT Kraken x72 a while back, and I personally think it's the best looking 360mm aio, but my H115i still out performs it, albeit at a much higher noise level, but I don't really mind noise.

 

I would however mind spending $200 + fans for a worse or equally performing aio.

 

[wildgg]

here's a few thoughts from what i can tell you

 

i feel like you should get lower temps with that setup.

 

that case is not helping you what so ever for airflow. (you could mod the front panel for extra airflow)

i personally would use normal thermal paste between the IHS and the cpu cooler, imo it's not worth using LM between the IHS and cpu cooler. (more on this later)

not sure if you did this but when you use LM on the DIE also make sure you apply LM on the IHS, this way you get better contact with LM between the DIE and IHS.

 

this is the reason i don't use LM between the IHS and cpu cooler, normal thermal paste like mx-4 fills the gap better between the IHS and cpu cooler.

 

for you to compare:

i have a delidded 8700k 5ghz 1.376V cooled by a S36 360mm AIO cooler and my max load temps i see on p95 are between 65/70C with a room temp of 22C

Case: Define S with 3x 140mm intake fans and a modded front panel for extra airflow.

 

so i think your case is somewhat of a problem but not the biggest issue here, i think you should consider redoing the delid, using LM between the DIE en IHS (apply LM on the DIE and IHS for proper contact). and use normal thermal paste between the IHS and cpu cooler and see what temps you get.

 

 

 

[Mavflight09]

1 hour ago, wildthing said:

i think your case is somewhat of a problem but not the biggest issue here

Right now I'm just trying to figure out the problem, if the case truly is the problem a front panel mod is no big deal. But if removing the front panel entirely and the dust filter only nets a 5C decrease in temperatures, I feel like something else is wrong.

 

1 hour ago, wildthing said:

i personally would use normal thermal paste between the IHS and the cpu cooler,

I was using Arctic Silver 5 in between the ihs and the H115i, but changing to liquid metal dropped temperatures by about 3-5C, so I stuck with it. However I haven't tried something like Thermal Grizzly Kryonaut, so I may try that later.

 

1 hour ago, wildthing said:

i think you should consider redoing the delid

I have already redone the delid, multiple times, with varying amounts of liquid metal, from not enough, to what I would consider "just enough" ( thin layer on ihs and cpu die ), to dumping liquid metal everywhere in frustration. I went through 2 entire tubes of Thermal Grizzly Conductonaut.

 

1 hour ago, wildthing said:

not sure if you did this but when you use LM on the DIE also make sure you apply LM on the IHS, this way you get better contact with LM between the DIE and IHS.

Yes I put liquid metal on both the cpu die and ihs.

 

1 hour ago, wildthing said:

for you to compare:

i have a delidded 8700k 5ghz 1.376V cooled by a S36 360mm AIO cooler and my max load temps i see on p95 are between 65/70C with a room temp of 22C

Case: Define S with 3x 140mm intake fans and a modded front panel for extra airflow.

Is that with the S36 in push only at the top of the case?

 

As a final note I also tried swapping the cooler with another H115i that I bought off of Amazon to see if my current H115i was bad, but alas, no they gave the exact same results temperature wise.

[aezakmi]

Quote

liquid metal between ihs and cold plate of the cooler

 

use paste

[Mavflight09]

3 minutes ago, aezakmi said:

 

use paste

Is there any particular reason for this?

 

I used Arctic Silver 5 to begin with, and then swapped to Thermal Grizzly Conductonaut as paste, and got a 3-5 degree reduction in temperatures.

[aezakmi]

7 minutes ago, Mavflight09 said:

Is there any particular reason for this?

 

I used Arctic Silver 5 to begin with, and then swapped to Thermal Grizzly Conductonaut as paste, and got a 3-5 degree reduction in temperatures.

paste will cover more surface as is thicker, also the metal could just move to the sides and leave the center of the IHS uncovered

[wildgg]

30 minutes ago, Mavflight09 said:

Right now I'm just trying to figure out the problem, if the case truly is the problem a front panel mod is no big deal. But if removing the front panel entirely and the dust filter only nets a 5C decrease in temperatures, I feel like something else is wrong.

 

I was using Arctic Silver 5 in between the ihs and the H115i, but changing to liquid metal dropped temperatures by about 3-5C, so I stuck with it. However I haven't tried something like Thermal Grizzly Kryonaut, so I may try that later.

 

I have already redone the delid, multiple times, with varying amounts of liquid metal, from not enough, to what I would consider "just enough" ( thin layer on ihs and cpu die ), to dumping liquid metal everywhere in frustration. I went through 2 entire tubes of Thermal Grizzly Conductonaut.

 

Yes I put liquid metal on both the cpu die and ihs.

 

Is that with the S36 in push only at the top of the case?

 

As a final note I also tried swapping the cooler with another H115i that I bought off of Amazon to see if my current H115i was bad, but alas, no they gave the exact same results temperature wise.

seems like you did everything fine, weird

 

my S36 is in the top of the case with 3x noctua NF-F12's pushing at 900rpm.

 

 

[Mavflight09]

21 hours ago, wildthing said:

seems like you did everything fine, weird

 

my S36 is in the top of the case with 3x noctua NF-F12's pushing at 900rpm.

This shouldn't be the difference between a 360mm aio and a 280mm aio right?

 

As far as I know the performance difference between a 360mm and a 280mm is very little.

 

I still can't believe you have 3 fans slowly pushing at 900rpm while I have 4 fans in push pull screaming along at 2000 rpm, and I get worse temperatures.

[wildgg]

2 hours ago, Mavflight09 said:

This shouldn't be the difference between a 360mm aio and a 280mm aio right?

 

As far as I know the performance difference between a 360mm and a 280mm is very little.

 

I still can't believe you have 3 fans slowly pushing at 900rpm while I have 4 fans in push pull screaming along at 2000 rpm, and I get worse temperatures.

360 vs 280 should be very close but the difference is prob the pump design with the cold plate inside

[nick name]

Was there a sticker on the cold plate that you forgot to remove? I highly doubt it since you've remounted several times and had to clean off paste, but I figured I'd ask a silly question. 

[Mavflight09]

1 hour ago, nick name said:

Was there a sticker on the cold plate that you forgot to remove? I highly doubt it since you've remounted several times and had to clean off paste, but I figured I'd ask a silly question. 

If that was the problem I would kill myself....

 

But alas no, H115i's, and subsequently most Corsair made aio's come with a hard plastic cover, so it's literally impossible to mount the cooler with the cover still on there, because the cover is massive.

 

Even if there was a plastic film on the cold plate I have scoured the cold plate so many times to remove liquid metal residue it would have been torn off ages ago.

[Mavflight09]

2 hours ago, wildthing said:

360 vs 280 should be very close but the difference is prob the pump design with the cold plate inside

I'm sure I could try a different aio, but according to Gamer's Nexus, and other reviews the Corsair H115i performs very well, even better than some 360mm aios, albeit with a very loud decibel output to accompany it.

 

I have already tried another H115i I bought off of Amazon, and the two performed nearly identically, so my aio isn't like half borked.

Unless both happened to be broken? But I highly doubt that.

 

At this point it's either the IHS, motherboard, or CPU it's self.

 

I doubt it's the IHS though, because previously I tried a copper IHS made by RockitCool:

 

While they are beautiful, my copper IHS performed about 1-2 degrees C worse then the stock IHS, probably because of the liquid metal staining the underside of the IHS:

 

[wildgg]

2 hours ago, Mavflight09 said:

I'm sure I could try a different aio, but according to Gamer's Nexus, and other reviews the Corsair H115i performs very well, even better than some 360mm aios, albeit with a very loud decibel output to accompany it.

 

I have already tried another H115i I bought off of Amazon, and the two performed nearly identically, so my aio isn't like half borked.

Unless both happened to be broken? But I highly doubt that.

 

At this point it's either the IHS, motherboard, or CPU it's self.

 

I doubt it's the IHS though, because previously I tried a copper IHS made by RockitCool:

 

 

While they are beautiful, my copper IHS performed about 1-2 degrees C worse then the stock IHS, probably because of the liquid metal staining the underside of the IHS:

 

 

are you gluing back the IHS with silicon or are you using the motherboard to clamp down the IHS on the CPU?

if you use silicon to re glue the IHS you create a gap between the CPU DIE and IHS.

 

with my 8700k i did re glue the IHS back but with very little silicon on the corners just enough to hold it in place.

i'm running out ideas for you here haha

[Mavflight09]

5 minutes ago, wildthing said:

i'm running out ideas for you here haha

Don't worry, I ran out of idea's months ago.

6 minutes ago, wildthing said:

are you gluing back the IHS with silicon or are you using the motherboard to clamp down the IHS on the CPU?

No, I leave the IHS free floating, held on only by the socket clamp.

 

6 minutes ago, wildthing said:

if you use silicon to re glue the IHS you create a gap between the CPU DIE and IHS.

Yea, Gamer's Nexus tested this a little while back and found that gluing the IHS back on could result in about 50% loss in temperature improvements.

IE: If you got a 20 degree drop by deliding and leaving the IHS free floating, gluing it back on would result in only a 10 degree drop compared to stock.

 

Maybe I'm crazy and chasing the impossible; I may have just gotten a bad sample 8700k that runs particularly hot.

 

But I was thinking of more extreme solutions for my temperature issues.

 

I have a very early revision of the Asrock z370 Gaming K6, so maybe the socket on the motherboard is slightly protruding upwards, preventing my cooler from getting full contact to the IHS.

 

But I mean, this thermal paste spread looks pretty good, and no, I didn't leave the thumbscrew there when putting the cooler on.

 

The only other thing I can think to do is either remove the socket entirely and use the cooler to hold the IHS to the CPU. Or I could get some form of direct die cooling kit, but that also requires cutting the corners off of your socket, and I don't really feel like doing that to my motherboard.

[wildgg]

7 minutes ago, Mavflight09 said:

Don't worry, I ran out of idea's months ago.

No, I leave the IHS free floating, held on only by the socket clamp.

 

Yea, Gamer's Nexus tested this a little while back and found that gluing the IHS back on could result in about 50% loss in temperature improvements.

IE: If you got a 20 degree drop by deliding and leaving the IHS free floating, gluing it back on would result in only a 10 degree drop compared to stock.

 

Maybe I'm crazy and chasing the impossible; I may have just gotten a bad sample 8700k that runs particularly hot.

 

But I was thinking of more extreme solutions for my temperature issues.

 

I have a very early revision of the Asrock z370 Gaming K6, so maybe the socket on the motherboard is slightly protruding upwards, preventing my cooler from getting full contact to the IHS.

 

But I mean, this thermal paste spread looks pretty good, and no, I didn't leave the thumbscrew there when putting the cooler on. 

 

The only other thing I can think to do is either remove the socket entirely and use the cooler to hold the IHS to the CPU. Or I could get some form of direct die cooling kit, but that also requires cutting the corners off of your socket, and I don't really feel like doing that to my motherboard.

i did re glue my IHS back but very little silicone dots on the corners so it's easier to get the CPU out the socket etc but ye if you use allot of silicone it will create a pretty big gap.

 

in that picture, your cooler seems to have good contact.

 

you could remove the socket just to test how temps are but i would not use that permanently.

stuff like direct cooling shouldn't be needed with your setup tho.

 

maybe that cooler is just not good enough when you almost push 1.4V 

i really don't know anymore

 

[Mavflight09]

2 minutes ago, wildthing said:

i did re glue my IHS back but very little silicone dots on the corners so it's easier to get the CPU out the socket etc but ye if you use allot of silicone it will create a pretty big gap.

 

in that picture, your cooler seems to have good contact.

 

you could remove the socket just to test how temps are but i would not use that permanently.

stuff like direct cooling shouldn't be needed with your setup tho.

 

maybe that cooler is just not good enough when you almost push 1.4V 

i really don't know anymore

 

Unless someone else has another idea I guess that's it.

I don't feel comfortable having my CPU reach 85 degrees so I'll hang up my enthusiast boots and back my overclock down.

 

I can compensate with my pretty good ram overclock though.

 

I have a stock 2133mhz 1.2v 14-14-14-34 kit of HyperX Fury Ram

I overclocked it to 3400mhz 1.35v 19-19-19-39

[wildgg]

7 minutes ago, Mavflight09 said:

Unless someone else has another idea I guess that's it.

I don't feel comfortable having my CPU reach 85 degrees so I'll hang up my enthusiast boots and back my overclock down.

 

I can compensate with my pretty good ram overclock though.

 

I have a stock 2133mhz 1.2v 14-14-14-34 kit of HyperX Fury Ram

I overclocked it to 3400mhz 1.35v 19-19-19-39

i hope there is someone that has that one little thing we didn't think of!

yeah 85C is for sure safe but i personally wouldn't like it to run that high, settle with 5ghz at lower Voltage i guess?

 

i never really got into ram overclocking

my kit is running at 2400mhz 1.2V 16-16-16-39

 

from what i understand 1.35V is the recommended voltage but 1.5V is safe?

i'm definitely interested and see if it's worth doing it

 

[Mavflight09]

9 minutes ago, wildthing said:

i hope there is someone that has that one little thing we didn't think of!

yeah 85C is for sure safe but i personally wouldn't like it to run that high, settle with 5ghz at lower Voltage i guess?

 

i never really got into ram overclocking

my kit is running at 2400mhz 1.2V 16-16-16-39

 

from what i understand 1.35V is the recommended voltage but 1.5V is safe?

i'm definitely interested and see if it's worth doing it

 

I wouldn't go to 1.5v, I'd call 1.4v the max I would go personally.

There isn't much performance to be gained, especially in real world scenarios, but if your interested my process is:

 

1. Set dram voltage to 1.35v

2. Set System Agent Voltage to 1.15, 1.2, or 1.25 (start at 1.15 and go up if unstable)

3. Increase memory speed

4. Increase timings until you get something that's stable.

 

Honestly it's a lot of trial and error to see what works.

On my set of ram 3466 fails to post at all, but 3400 posts fine, so that was my hard limit.

From there I increased timings and voltages (dram and System Agent Voltage) until it was stable.

 

Personally I don't mess with secondary and tertiary timings, mostly because I don't know what they mean.

 

For stability I use IntelBurnTest set to 8192mb.

 

If you want more information I followed this guide: https://www.overclockers.com/forums/showthread.php/785102-DDR4-RAM-overclocking-101-guide

[SolarNova]

The only thing I can thnk of to improve temps with the gear you have is going bare die mounting. Though i cant say i know if your motherboard would allow this without requiring you to mutilate the socket.