chispy:

Watch this video again ( 6.44 mark ) , nowhere does he states 12-18c difference from stock to direct die with liquid metal , in fact he says around 4-10c expected depending on the cpu ( every cpu has a different vid ). Delid the cpu buy this frame use liquid metal for an average of 5~6c degrees on an average cpu with a vid of 1.30v it is not worth it. Worst case scenario 4c and best been 10c better temps only and that's not great and not a massive difference for a 9900k.

Maybe listen to what he says one more time at 6:45 mark. You got the time right but maybe try again when you listen. He didn't spell out the full potential but he actually says 4-10c compared to a CPU that already has liquid metal. The bigger difference coming from the 9900K and lower coming from the 9600K. So where did I come up with 12-18c? Well there is already temperature savings at delidding and applying liquid metal so saving 4c-10c beyond that is roughly 12-18c from a stock 9900K CPU.

SweenJM:

Seems like something you would only really want to do if you were running LN2 and every degree counted.

On 9900K overclocks every degree does count even on much more practical cooling solutions. IMO the 9900K is a very niche CPU anyways so applying a more niche direct die cooling solution is only fitting.

Fox2232:

In good old days, there were no IHSes. But there were thicker PCBs under chips.

This is what keeps coming to mind as well. Back in the day we didn't have IHS on our CPUs maybe we need to go back to those days. lol

Jayp:

This is what keeps coming to mind as well. Back in the day we didn't have IHS on our CPUs maybe we need to go back to those days. lol

But one has to wonder how many chips ended up damaged due to system building person had no moderation. Just last week colleague told me story that some of his family members attempted to build system on their own to save $20 for assembly done by parts selling shop. Ended up bending pins on CPU, paying $50 to guy in clock repair shop who straightened them as good as he could. But USB ports are still not working and therefore OS installation can't be done. He is going there to attempt to get it into working order. I wonder what story he'll tell after Christmas.

Fox2232:

But one has to wonder how many chips ended up damaged due to system building person had no moderation. Just last week colleague told me story that some of his family members attempted to build system on their own to save $20 for assembly done by parts selling shop. Ended up bending pins on CPU, paying $50 to guy in clock repair shop who straightened them as good as he could. But USB ports are still not working and therefore OS installation can't be done. He is going there to attempt to get it into working order. I wonder what story he'll tell after Christmas.

Would be interesting to know. Back in the day I never thought of it as an issue but I am sure it was for someone. Much like this computer building story you just shared. People just need to slow down and learn how to do things. Building computers is super easy but I always hear stories of it going terribly wrong. I was 14 when I built my first brand new PC (Athlon XP 2500) and had no issues. I did mess with older used systems before that but still I was a kid. This was before YouTube was there to provide a helping hand.

BLEH!:

Intel switches back to solder - OCers still delid... xD

Liquid metal conducts better than any solder.

oxidized:

Liquid metal conducts better than any solder.

der8auer says you lie,solder conducts better but any liquid metal, but requires a certain thickness to be used so it does not fracture. Thats what he says in the delid of the 2600X [youtube=VEUG9d4Mjug]

So better but a cryo-tec cooler instead of delidding

the temperature is about 4 to 10 Kelvin lower.... Kelvin ???? 😏really who the F@%^$ uses Kelvin to measure CPU temps

Somebody does like bigger figures - others perhaps bigger boobs...

Dazz:

der8auer says you lie,solder conducts better but any liquid metal, but requires a certain thickness to be used so it does not fracture. Thats what he says in the delid of the 2600X [youtube=VEUG9d4Mjug]

I might be wrong but i'm not sure how he tested what he's saying. Liquid metal probably has a worse ELECTRICAL conductivity, but the same cannot be said for HEAT conductivity, simply because they're not the same thing, and they don't necessarily depends on the same things. In theory liquids/fluids or solutions if we want to be accurate, conduct heat better than solids, simply because they're liquid and not solid. The liquid metal which is an alloy of multiple metals, has its melting/freezing point at room temps, therefore it could be considered a solution, hence conducting heat better. This is what i learnt so far, i might be totally wrong, but i'd like to see some proof first.

actually some people are getting great results with a delid, thicker die or not. if you just use the ihs as normal its only about 6c difference but if you sand the ihs down its much much greater. the reason being, i suppose, because the solder is so much thicker that without it the gap between the ihs and the die are too great. liquid metal will fill the gap but wont do much better than the solder (liquid metal and solder are both made with similar metals). this guy delidded and sand down the die and got aroun 16c difference. direct die will get about the same without sanding or anything. derbaurer said that the direct die frame would only get about 6-10c difference but he was comparing to a standard delid with liquid metal. so thats 6-10c in addition to what the 6c from a "normal" (non sanded) delid. and the 4 or 6c difference (whichever it was) was for a 9600k or maybe a 9700k. those cpus dont even need delidding because the power\heat generated is so small. a 9900k at 5ghz delidded and with the die sanded will get nearly what a direct die will get. but a direct die will get at least around 15 and if you are pushing temps into the 100c range because you are running hihg stress apps and\or have the voltage higher and\or the clocks higher than 5ghz you will see even a greater difference. but if you are not going to take the time to sand the die down and make sure the gap is as small as possible than do the direct die. in fact better to just do the direct die either way. with direct die and a decent quality chip and even just a dual tower air cooler you can tame this chip fairly easily if you watch your llc and voltage settings. dont use more than about medium on gigabyte or level 3 on asrock. find the one that keeps voltage flat and then go down about 2 steps. yes i said that right. DOWN about 2 steps. not up. a few boards might be able to scale up to the "flat" llc level but most wont. my gigabyte pro z390 clocks no better with high or turbo than it does at medium but this saves a lot of heat and power. with a direct die kit im fairly confident i could run 5.1ghz and put the short power limit to 190w and still keep the cpu under 78c or so at all times.