Motherboard heatsinks, removeable? (8700k build)

[wockawocka]

Hi all,
I've built a tiny pc based around a Z370 chipset and 8700k cpu.

Laughably it does 5Ghz (boots, but quickly falls over when pressured) so I'm happy sitting at the standard turbo speeds but I'm a bit concerned about the heatsinks on the board (Asus Z370-I Gaming Mini ITX) in that one of the heatsinks (almost) touches the cpu cooler and I think it's restricting some of the airflow from the cooler (Noctua LH9i).

Is it a silly idea to remove the heatsink and increase the airflow which would then flow over that area anyway?

Case is a Chopin BQ696.

Thanks!

[Korth]

This ROG article says:
"The data we obtain from the binning process is also of interest for ambient cooling. Our Coffee Lake testing revealed that 50% of early 6-core CPUs can achieve 5GHz with water cooling. If retail CPUs achieve the same, it shows Intel's mastery of the 14-nm process."

This is based on an "if" statement, your retail i7-8700K could be better (or worse) than the pre-release samples ASUS used for testing and binning. But, assuming the parts are equivalent and the ASUS results do apply, we can expect 50% of i7-8700K parts will achieve 5GHz on ambient/water cooling - and we can expect 50% of i7-8700K parts will not achieve 5GHz on ambient/water cooling.

Your Noctua NH-L9i CPU cooler is just not beefy enough for hard overclocking, not even beefy enough for stock full-rated operation of a (95W) i7-8700K.
"The NH-L9i is a highly-compact low-profile quiet cooler designed for use in small form factor cases and HTPC environments. While it provides first rate performance in its class, it is not suitable for overclocking and should be used with care on CPUs with more than 65W TDP (Thermal Design Power)."

If you want a real shot at stable 5GHz then you have to upgrade your CPU cooler (unless you're phenomenally lucky and got a i7-8700K part which can somehow run stable 5GHz overclock while undervolted, lol). Noctua's TDP Guidelines basically recommends only huge dual-tower or large single-tower air coolers (no low-profile or flat air coolers) for a 95W CPU, although these items are obviously incompatible with your Chopin. An AIO cooler would be preferred but the Chopin has nowhere to (internally or externally) mount any fans or rads. You probably already have pretty much the best (and quietest) cooler "in its class" that you could possibly mount in that chassis.

If you're using the chassis-integrated PSU then you're also limited to 150W total system power. No GPU card (no room for one), but that's still not a lot of power available for running a 95W CPU (with active iGPU) at full spec or at any sort of overspec.

It appears the reality is that you cannot realistically expect to achieve fast CPU overclock within the cooling and power constraints imposed by that SFF chassis. And even with adequate cooling and power you can realistically expect only 50/50 odds of achieving 5GHz on your CPU.

[Korth]

(It is a beautiful little case, though. I always loved simple, elegant designs, clean lines, and charcoal-grey brushed metal surfaces. It would make a fine home for a mighty HTPC or, with some modding, a happy little cluster of busy SBC boards, lol.)

[JustinThyme]

Are you talking about removing the heatsinks on the VRMS? Please say it isnt so!!!! If it is so then unless you are going to use a monoblock to cool the VRMs as well then leave it be or count on letting the magic smoke out of your MOBO.

If you are running a micro board in a micro case then you should also be running micro clocks. While these chips are getting easier to OC its all highly dependent on cooling. Unless you have a massive 1 foot tall Heat sink don't expect much from air cooling if anything, maybe a 100 mhz bump.

[Korth]

I didn't catch that detail in the OP, lol.

But I also disagree with the notion of removing these heatsinks to increase airflow. The whole reason they're "obstructing" airflow is that ... well, air is flowing across them, pulling heat off the components they're cooling, lol.

That being said ... it is possible to replace garish/ornate gaming motherboard heatsinks with more efficient (and much less pretty) industrial designs. But that chassis relies entirely on only CPU fan and PSU fan and top convection to remove heat from chassis. I doubt upgraded VRM heatsinks would substantially improve temps.

[wockawocka]

I have to say, I love this chip (thanks Korth for turning me on to it), this won't be in the build for too long as I'll be putting either a non K version in there or an i5 8400, whatever it is it'll be a 65w TDP chip as this rig has a 150w psu and the temps are high. I don't want to cook the build. I even looked at the 6700k as it seems to have really good starting temps with things going up from the 7700k onwards. But whether it's worth it to lose the two cores....

I wasn't sure of the safety of removing heatsinks, I know that usually if it's there it's usually for a reason which is why it's always a good idea to check in with the internet before I do something silly. Prime 95 torture test does AOK at standard chip settings. Hits 72 degrees.

I'd love, love, love to get some sort of watercooling in there. A bigger case would answer this I know but it's not the reason for the build.

[Korth]

The biggest components in a DIY cooling loop will be fans and rads. There might actually be versions small enough for your case, but their cooling power and efficiency is largely based on physical bulk and surface area so unless they have dimensions bigger than your existing cooler they won't be better coolers. An externally-mounted fan/rad setup or a completely separate external fan/rad chassis are possible but would wreck much of the Chopin's appeal.

The fastest 65W CFL/DDR4-2666 processors are i7-8700 (12MB, 1200MHz UHD630) and i5-8400 (9MB, 1050MHz UHD630).
The fastest 65W KBL/DDR4-2400 processors are i7-7700 (8MB, 1150MHz HD630) and i5-7600 (6MB, 1150MHz HD630).
The fastest 65W SKL/DDR4-2133 processors are i7-6700 (8MB, 1150MHz HD530) and i5-6600 (6MB, 1150MHz HD530).

There's honestly not much "real" difference between these iGPUs, or even between high-end Z370/Z270/Z170 m-ITX mobos.
But while on-die caches and memory speeds are usually secondary to raw clock speeds in "real" performance terms, they're seriously worth considering in a machine which relies entirely on iGPU.

I notice that Intel has cleverly designed their "non-K" Coffee Lakes to make anything but the very best (most expensive) "-K" parts less desirable for iGPU-based systems. But CFL is still very new, I think it's likely that Intel will release i5-8600 and i5-8500 (and a variety of low-power "-T" parts) in upcoming weeks/months, and if these mirror their SKL/KBL predecessors then they'd be better choices for your build than the i5-8400.

[wockawocka]

After a bit of tweaking I've been getting 5ghz (dropping down to 4.7 occassionally) Temps not going over 80 degrees.

Not torture tested mind you, but running my work software (Adobe Lightroom) it doesn't crash out under load.

[Korth]

I'm surprised - and a little impressed - that DIP5 can pull off 5GHz!

4998.8MHz (99.976MHz x 50) is not 5000MHz (100.000MHz x 50) but close enough!

[Korth]

I'm surprised - and a little impressed - that DIP5 can pull off 5GHz.

4998.8MHz (99.976MHz x 50) is not 5000MHz (100.000MHz x 50) but close enough!

Why is GPU frequency 0.0MHz? You're running this processor on a different mobo with a GPU card, outside the Chopin chassis?