I9 9900k

Colonel_Klinck · 9 Apr 2019 at 17:58

Distracted said:
What block are you putting on it?

EK Supremacy. I believe it works ok.

Distracted · 9 Apr 2019 at 19:31

If it has the same mount as the velocity blocks then it won't make enough contact with the die even with the spacers removed. You may need to get some screws or a different bracket to get it to sit right.
That is my experience with 3 separate velocity kits, 2 different mobos and 2 delidded cpus with the die frame from der8auer.

moorhen2 · 9 Apr 2019 at 19:40

Well my 9900k died the other day, was delided and direct die cooled, I think something was not right with it after I delided it, damaged some traces I believe as the IMC was never the same. So got another 9900k today, retail this time, and its looking very good at the moment, straight in at 5.1ghz 1.230v in the bios, 1.190v load voltage. Max core temp 64c with Aida64, non FPU/AVX, as I don't use it. Using my XI Apex for the first time, see how it compares with my XI Extreme, my Z270 IX Apex is a beast, so the XI should be just as good, here's hoping.lol

Wont be bothering to delid and direct die cool this one, shouldn't need it. Batch L903E740, so produced third week in January this year, looks like they are getting better with the production process maturing.

Distracted · 9 Apr 2019 at 20:45

oof sorry to hear that Moorhen. I hope the new CPU is just as good or better than the last.

moorhen2 · 9 Apr 2019 at 21:09

Distracted said:
oof sorry to hear that Moorhen. I hope the new CPU is just as good or better than the last.

Thanks. This chip looks to be on a par with my dead one so far, at least this Apex can run my 3200 memory kit at 4133 like I know it can do, so that's a bonus straight away.

Colonel_Klinck · 9 Apr 2019 at 22:36

Distracted said:
If it has the same mount as the velocity blocks then it won't make enough contact with the die even with the spacers removed. You may need to get some screws or a different bracket to get it to sit right.
That is my experience with 3 separate velocity kits, 2 different mobos and 2 delidded cpus with the die frame from der8auer.

I'm sure Ber8auer said in his vid that he'd tested with the Supremacy and it was fine but wasn't sure on the Velocity. I'll be sure to check though and will set the bios to a low clock and volts until I'm sure.

Distracted · 10 Apr 2019 at 00:05

Colonel_Klinck said:
I'm sure Ber8auer said in his vid that he'd tested with the Supremacy and it was fine but wasn't sure on the Velocity. I'll be sure to check though and will set the bios to a low clock and volts until I'm sure.

He says he tried both the supremacy evo and the velocity and had no problems. However if you watch the video at the part where he says that, you'll see he is installing the mounting screws without the washers which the instructions say are mandatory. To me this means he is aware that the washers raise the height of the block too much so he is aware that there is an issue there.
I tried copying his approach but even with the washers removed the temps and contact were poor so I used an alternative mount. Now I am using m3 screws as well as the ek springs and nuts with the velocity kit (and a kraken backplate i think - works really well). Temps are much better (easy 10 degree drop, maybe more) and mounting pressure is also much improved.

Initially i was pleased with the velocity since direct die dropped temps by 10+ degrees. But after mounting it properly for an additional 10 degree reduction at least, i have to recommend against using their mounting kit unless you use longer screws.

Colonel_Klinck · 10 Apr 2019 at 17:58

Distracted said:
He says he tried both the supremacy evo and the velocity and had no problems. However if you watch the video at the part where he says that, you'll see he is installing the mounting screws without the washers which the instructions say are mandatory. To me this means he is aware that the washers raise the height of the block too much so he is aware that there is an issue there.
I tried copying his approach but even with the washers removed the temps and contact were poor so I used an alternative mount. Now I am using m3 screws as well as the ek springs and nuts with the velocity kit (and a kraken backplate i think - works really well). Temps are much better (easy 10 degree drop, maybe more) and mounting pressure is also much improved.

Initially i was pleased with the velocity since direct die dropped temps by 10+ degrees. But after mounting it properly for an additional 10 degree reduction at least, i have to recommend against using their mounting kit unless you use longer screws.

Well job done. No issues with contact. I removed the washers and when I placed the block on I could feel it rocking on the die. In fact I was careful not to tighten down the block too much as I didn't want to risk too much pressure on the die. Temps wise it knocked 10c off. At 5.0Ghz 1.27v it went from 78c maximum to 68c after 30 mins load. Trying to get it stable with zero offset at 5.1Ghz means jumping to 1.33v and then it is 83c maximum after 30 mins of Realbench. It has closed the core temp delta though. Now it is 80-83c across the cores. LLC is set to turbo.

Distracted · 10 Apr 2019 at 21:16

Colonel_Klinck said:
Well job done. No issues with contact. I removed the washers and when I placed the block on I could feel it rocking on the die. In fact I was careful not to tighten down the block too much as I didn't want to risk too much pressure on the die. Temps wise it knocked 10c off. At 5.0Ghz 1.27v it went from 78c maximum to 68c after 30 mins load. Trying to get it stable with zero offset at 5.1Ghz means jumping to 1.33v and then it is 83c maximum after 30 mins of Realbench. It has closed the core temp delta though. Now it is 80-83c across the cores. LLC is set to turbo.

Glad to hear it worked.

Colonel_Klinck · 10 Apr 2019 at 21:26

Distracted said:
Glad to hear it worked.

Thanks. Was hoping I’d be able to get to 5.2 but I think it would just take too many volts to get it stable.

moorhen2 · 14 Apr 2019 at 17:14

Well my new 9900k is a good sample, needs less voltage than my dead one did for the same overclocks, 5.0ghz with 1.175v set in the bios, 1.146v under load using Aida64 non FPU/AVX as I always do, max core temp 61c, bearing in mind that its straight out of the box, its far better thermals wise than my previous one. Not going to bother Deliding this one for direct die cooling, no point.

moorhen2 · 14 Apr 2019 at 17:46

Finished an hours run, no change in max temp, still 61c. Load voltage 1.146v, 1.175v set in bios, LLC7.

Colonel_Klinck · 14 Apr 2019 at 17:50

Wow that is great volts. I’m stable at 1.335v at 5.1. Temps are fine as cpu is on it’s own custom loop but wish it was down nearer your 9900k.

moorhen2 · 14 Apr 2019 at 18:20

Colonel_Klinck said:
Wow that is great volts. I’m stable at 1.335v at 5.1. Temps are fine as cpu is on it’s own custom loop but wish it was down nearer your 9900k.

Thanks, looking back to when I had my first 9900K, this chip needs slightly less voltage for 5.0ghz, and max core temp is about 2c lower with my new chip.

5.1ghz up next.

moorhen2 · 14 Apr 2019 at 19:28

5.1ghz 1.225v in bios, 1.181v load voltage. Max core temp 65c.

Distracted · 15 Apr 2019 at 06:37

@moorhen2 are you now using the apex?

moorhen2 · 15 Apr 2019 at 09:18

Distracted said:
@moorhen2 are you now using the apex?

I am indeed using my XI Apex.

moorhen2 · 15 Apr 2019 at 13:01

5.2ghz, 30 minutes in to an hours run of Aida64, 1.270v set in bios, 1.243v under load, max core temp 71c.

oxf77 · 15 Apr 2019 at 14:23

I assume you guys are overclocking all 8 cores. Can anyone give me an idea how higher you can overclock if only enabling 2 cores?

LePhuronn · 15 Apr 2019 at 15:06

MaXxBoulton said:
Is that even a thing anymore. I thought it was shown to give no real world benefit years ago ?

My Google Fu is weak today, but I recall there being worse performance on Devil's Canyon direct-die. Die area wasn't sufficient to dissipate heat effectively enough through the base plate of the cooler so it was hot spotting. I recall minimal gains with Skylake too. Hence the use of an Integrated Heat Spreader.

However, direct die chops a few degrees off cooling the 7900X according to Tom's from June last year
https://www.tomshardware.com/reviews/core-i9-7900x-overclock-ln2,5618-4.html

I'd presume it's die area again - this time it's a lot bigger so there's greater surface area in contact with the cooler plate, and therefore is more efficient.

By that logic then I don't see there being much to gain if you direct die a 9900K. Not that I can find any numbers...