[SOLVED] 4790K Delid - Same temperatures as before

[carbide]

Hi there, after deliberating about delidding for some time, I did it.

I watched all the gamers nexus videos, and the buildzoid 'how to apply liquid metal to things' video over and over to avoid any potential mistakes.

I used the Der8auer delid tool, cleaned all silicone off the cpu and IHS to get the closest spacing, used a nitrocellulose nail polish to protect the surface mount components etc.

Initially I saw no improvement, so did it all again, this time added slightly more liquid metal as I was concerned I hadn't added enough.

Still, no change. I think in XTU benchmark I see in HWinfo64 the max recorded package TDP went up from about 110w to 130w which leads me to believe there has been improvement, just not in temps.

I'm using conductonaut LM and kryonaut TIM. Cooler is H100i GTX.

I'm starting to wonder if the 3-4 year old H100i is past its best? With the fan profile on performance, and pump at max, coolant temp never gets to 40c, however max core temperatures reach 85-90

4790k is 4.6 all cores with 1.26v. 1.296v seen with AVX loads.
PL1/PL2 set to unlimited as it current limit.

Any ideas?!

To add: max core variance (after both applications of LM) is only 7c

 

[sla70r]

The 4790k is just a hot chip. Mine has been running at full load around 85c for the last 3-4 years on water cooling. I remember when i purchased the chip countless threads about the heat generated from the 4790k.

 

[cookiemania66]

In normal gaming usage you wont encounter any AVX Workloads so i would suggest you use Prime V26.6 (latest version without AVX) since AVX instructions produce a lot more heat.

I dont think you did something wrong if you followed the Videos from the persons you mentioned.

And make sure that your radiator does have acess to fresh air and is not blocked off by fancy tempered glass or somehting like that.

 

[carbide]

forgot to add, yes, was using prime 26.6 to stress test as I know the haswell AVX issues since I bought this chip back in Sep 2014.

case is a prodigy, with H100i in the roof with 2 maglev fans, 230mm from fan running flat out on 12v molex - it moves a ton of air

idle temps are 4c above ambient.

I just expected to see some difference, iven if minor, or even, worse, or even less/greater core variance, I'm wondering if this means the Intel TIM wasn't the temp issue?!

 

[carbide]

So, just as an update, the main issue was that H100i number two went bad. Gurgling, pump grinding, high temps no matter how much elevating and shaking. I can only assume corrosion filled the micro fins and the coolant loss (I later found dripped all over my front fan) didn't help the pump either. Maybe uninstalling/reinstalling etc. moved a load of silt and that clogged it up.

A quick cinebench run at stock settings took the temps to ~90 and a bsod. limiting frequency down to ~1500mhz still got the COOLANT temp to spike into the 50s suggesting, it wasn't moving.

People always say it, and I've always wanted to defend AIOs due to their simplicity (compared to custom loop) but, I can't be bothered to try a third. Also, the clamp bracket and socket support plate is rubbish, after buying an NH- D15s to replace it, I was so impressed with the build quality in comparison. With the H100i I'd actually bent my motherboard with the force I was using (obviously I shouldn't have been clamping it down so hard, but it improved temps) and the noctua bracket pulled it back flat. Also, the noctua cooler has two screws, and circlips on the shank, so you can't over tighten - every uninstall/reinstall results in the same clamp force wheras the H100i has 4 screws and is a pig to screw down, balance the force, and ensure the head sits flat on the IHS.
I'm not a hater, been using the H100i product for 4 years but, it's loud (even with ML120s) takes up space, usb header, should be using an additional power splitter for the fans... Wheras the D15s provides the same cooling with just a single 140 fan to plug into the header.