Likely won’t see the same gains as with water but yeah I’d say it’s worth it.
These CPU’s run hot enough as it is.

I’d suggest using Rockit Cool kit to delid and direct die.

 

Maybe because on custom blocks you get more finecontrol
Big Air-Coolers have a focus on double mounting mechanisms and quite some pressure
You'd crack the silicon, just alone from the required pressure to hold this bulky tower vertically

Sure mounting pressure is spread across the backplate of the mounting system and across the board
But every tiny bump - just case movement -> will for sure move the huge tower a tiny bit up and down
Resulting in higher pressure on one side, following by high pressure on the bottom side = fully cracking the sillicon

Very bad idea, don't even try it
It would potentially only work on flat test benches ~ only IF this testbench rly stays flat
But you again have to keep in mind the pendeling force which will 100% happen even on by hand lift-up or hitting with the fist on the desk for example (unlikely to happen as event, but can happen - which will result in also a crack in sillicon)

On the other hand flat mounts have more structure to them
Taking as example Near Sub-Zero cooling, while pots on it's own are never horizontally standing
A normal waterblock requires too quite some pressure, but it's side focused ~ while they are quite flat and so not that heavy
Tho if you take for example the Alphacool Eisbär, or Deepcools bigger Pump-Block combos
Those would have a higher chance of wingle on movement ~ be it simple nature earthquake movement
Which 100% will result in sillicon cracking

Flat stuff like Alphacool Eisbaer LT Solo, or any other tiny sharp edges block which focuses pressure on the sides may work
But even then, it's very recommendable to use fixed standoffs for vibration dampening - else you'll crack the Die on any kind of wiggle pressure

 

It will keep your max temp from 2c to 6c lower.

 

Hi,
If you have a copper air cooler i suppose it is

But most I've seen direct die are at least using a clc cooler not air and majority are custom water loops.

Main question is are you willing to flush a 3 year intel warranty good bye ?
Chips die.

 

Depends what CPU you're talking about and what kind of air cooler you're talking about and of course if you're okay with potentially destroying your CPU

 

If no one has done it before I'm not going to be the first. I was thinking of doing it with a 9700k and a DH15s.

 

Hi,
I personally don't think it's worth it
Wait until the warranty expires then do it

 

Yup, definitely not the first and it does give better numbers....as long as you're okay with the potential risk I'd go for it.

 

Just did it with a 9900K and a Noctua D15. Works fine! Just be precise and carefull: I designed and 3D printed this bracket: https://www.thingiverse.com/thing:4118891

I used this de-lid tool to remove the heathspreader: https://www.thingiverse.com/thing:2684510 (printed in PLA 90% infill) PLA is very hard, harder then ABS or PETG, so I definitely recommend PLA

Used old credit card to remove the remaining glue and the indium solder. Finally very carefully use some very fine grind sandpaper (1000+) on top of a little mirror (flat surface) and sand off the remaining indium solder. Indium is very soft, you can push a dent in it with your fingernail so the sanding should not take more then a minute.

If you got the balls, you can sand off 0.15 mm from the CPU die for even better temps. Do be careful because its VERY easy to make the surface of the CPU die not flat anymore because of uneven pressure while sanding. I do not recommend that!

 

We can't tell you what you value (if it is worth it to you or not) but my two cents is I wouldn't go to that extent for lower temps while still on air cooling.

To me, going direct die (especially on a soldered 9900k/ks) means you are going all-out for either lowest possible temps or max possible daily overclocking. Going direct die but then slapping an air cooler on it seems to me like building a really lightweight drag racing chassis and then putting in a factory engine. Sure, it will still be capable of accelerating and be faster than the fully stock unmodified version of the car... but why? In PC terms, yes you will still see lower temps with direct die than with the stock soldered IHS even with an air cooler but you're risking destroying the CPU for a less-than-best gain. I just don't see the point.

 

In my case it shaved off about 13 degrees. Or from another point of view I can set the voltage about 0.07 volts higher and still have slightly lower temps.

I did notice that before going direct die, temps would ramp up anyway way into 90's territory while the CPU cooler would not be more then lukewarm. I've seen posts from other people with even better (water)cooling experiencing the same problem. Apparently when overclocking the limit isn't really set by whether you use high end air coolers like the Noctua D15 or high end watercooling, but the limit is how much heat the chip can dissipate through that layer of indium into the heatspreader and then into the cooler.

 

Deliding and wet sanding the ihs and cooler base is always worth it if they're not perfectly straight. The less TIM is between both surfaces the better it works, don't forget copper transfers heat 50x better then any TIM.

 

I read this thread today, doing my preparations for delidding my 6700K using aircooling. Reading the comments with recommendations to not do this when using air coolers, I thought I need to reply.

First of all I'd like to say: you definitely do not require a water cooling configuration in order to justify delidding. From a somewhat technical point of view, it's completely irrelevant what kind of cooling you use in this discussion. The cooling comes after the internal heat spreader (IHS). The domain right now is everything before that. The IHS and how it's set up with silicon glue for the PCB of the CPU and inferior TIM is the thing that you want to get rid of. Regardless of the cooling system that comes afterwards, if you do this properly, then a significant decrease in cooling requirement is realized.

I understand Kalm_Traveler's point that this is for enthusiasts. In fact, I agree with that. But I consider myself an enthusiast (who generally tries to make financially wise decisions, hence no water cooling config yet), just with an air cooler. The best air coolers are still very good options for a bit of overclocking. If you read the better CPU cooler reviews where they establish the efficiency of a cooler (where both thermal performance and noise are factored in (e.g. by tuning all fans and pumps so that 40 DBA are produced, and then comparing the delta temperatures)), you'll find that a Noctua D15 performs almost as well or even better than the best 280MM AIO's. And, as an enthusiast who just wanted a 10 degrees drop under load, I wanted to try this delid first, before moving to a custom loop.

Even though I agree that delidding is more for enthusiasts, my view is that delidding makes even more sense if you have a weaker cooling. Check Youtube for videos from people who actually delidded their laptop CPU's and used liquid metal, and were able to get great results. You'll find some results where people did this to a few (very expensive) previous Macbook Pro 15" models, for example this one from Snazzy Labs: https://youtu.be/iw4gqfrBN4c?t=374 Via this link it starts at his conclusion which sums it up pretty well: no more throttling (he's mentioning a 15% improvement in performance (sorry, didn't check the whole video again) and less (in your face) fan noise, something that many laptop (especially ultrabook) users will understand.

So, does it make sense to do it with a desktop with an air cooler? Sure, as long as you're aware of the risks (just take it slow and do the IHS removal very carefully) and think it's worth it.

So, my delidding results? Honestly, I am a bit surprised about how well it works. I have a 6700K that is about 3.5 to 4 years old now and therefore (I assume) had a bit older and less performant TIM under the IHS and even under my air cooler. My air cooler is an old but still fine 870 gram (just the aluminum+copper+nickel) Scythe Mugen 2 Rev. B, with 2 Corsair ML120 fans attached to it. These PWM fans can range from 400 to 2400 RPM (this huge range is the reason I bought them). Stock, this cooler comes with 1 Scythe 120 PWM fan that have a range from about 300 up to 1300 RPM, which makes a lot of sense I realize now (read on for the reason).

I currently have it running at 4.6 Ghz with a voltage of 1.4 (which is about 15-20 degrees hotter than running it stock). In my Prime95 Small FFT torture tests I was getting 78 to 82 celcius temperatures when running the 2 fans on my air cooler at full speed (a ridiculous and very annoying sounding 2500 RPM). With the delid, using those same fan speeds, I am getting 52 to 59 celcius. And the nice thing is, when I lower the fan speeds enormously, to a pleasant level (around 1000 RPM), the temperature barely rises. I am running this Small FTT torture test while writing this with 50% fan speed (so about 1250 RPM) and the temperature still hovers between 57 and 60. It seems that the CPU loses the heat just that much more easily. And I have to draw the conclusion that using this air cooler in combination with running these fans at a high RPM does not make sense. There is just not much airflow resistance.

Using Cinebench Release 20 as a benchmark, before I was getting temperatures of around 60 to 65 I believe at full fan speed. Now, I am getting temperatures of between 46 and 50. Same situation when halving the fan speeds to 1250 RPM.

Overall, naturally I am very happy with this result. I did not expect this, especially considering the fact that I ended up not using this die guard https://www.aliexpress.com/i/32846548144.html in a naked / delidded / no IHS config. The reason is that I first need different screws in order to mount the air cooler properly, because now it's not tightened closely enough to the CPU die (because the CPU is about 3mm less high I guess because of the missing IHS). From what I understand, most CPU water blocks have better (out of the box) adjustability for this, which is an argument for doing this with liquid cooling.

Anyway, I just did a delid with liquid metal, and no reglueing the IHS back onto the CPU PCB. I just dropped the naked CPU (chip with PCB) in the socket, dropped the IHS with liquid metal on the CPU, held it together, and closed the motherboard bracket to secure all parts as if they were still glued together. This is the recommendation of Gamers Nexus who created some very good articles and videos about this topic. This 'config' is definitely something to remember when taking the CPU out though .

Apologies for the long text, hope it helps though.

 

9900k and D15s, max temp doesnt even hit 80C under AVX load test, 40s in games, for me it wouldnt seem the risk, but I dont put crazy voltage through my cpu.

 

훌렁따 direct die 공랭 air cooling