pastrychef's Asus ROG Strix Z370-G Gaming (WI-FI AC) build w/ i9-9900K + AMD 6600 XT

[pastrychef]

Nice! I've never seen a soldered die this clean! Did you use any polishing cream/pad, or just conductonaut?

Yes, with LM+IHS, you should get a 10-15℃ drop, compared to stock. I assume it's the Rockit Copper IHS. If yes, then note that since it's not nickel plated (and its surface, although very flat, not mirror polished), the gallium in the LM (conductonaut is a type of galinstan alloy) will slowly diffuse into the copper to form a copper-gallium alloy (resulting in increased temps), so you might need to replenish/reapply LM for a second time after a few weeks. After that, the diffusion becomes slower.

In the meantime, (if you still have your 8700K) you could try testing with Prime95 (if you haven't already) to make sure that your M/B can handle the extra load. Using Prime95 v26.6 (i.e. non AVX), run a torture test, for at least 30 minutes, with the settings in the attachment. It would draw ~180W of package power, so about the same (if not a bit less) with what an overclocked 9900K would draw under real-world heavy workloads (e.g. handbrake, blender).

If, during the test, the clocks remain stable at 5Ghz (you could check with Intel's power gadget), then you should be good to go, otherwise, you might also need to look into improving your VRM cooling while you are at it.

Good luck and have fun!

Yes, I used some Flitz with paper towels to clean it afterwards.

Yes, I have a Rockit Cool Copper IHS coming. I noticed that the stock IHS is a bit convex on the top and after my delidding, it teter-toters on the die far more than I ever remember on any of my other delidding jobs. I'm sure the copper IHS will be much flatter.

Yes, I know that Gallium reacts to copper, but I've never had to reapply liquid metal because of it. As far as I understand, it's does not affect thermal performance. In the past, I've used liquid metal between GPU die and copper plate without ever having to redo it a second time. Temperatures remained fantastic for many months.

I did run Prime95 when I first put my build together and there weren't any problems. I'll try again a bit later. I already received my Fujipoly and I'll be applying at the same time I pull my i7-8700K.

 

[elfcake]

I already received my Fujipoly and I'll be applying at the same time I pull my i7-8700K.

Do you find that the fuji is better than premium thermal grease (Kryonaut)?

 

[pastrychef]

Do you find that the fuji is better than premium thermal grease (Kryonaut)?

No. I will be using the Fujipoly to replace the thermal pads between the VRM mosfets and the heatsink. Supposedly, Fujipoly makes the best thermal pads...

 

[elfcake]

No. I will be using the Fujipoly to replace the thermal pads between the VRM mosfets and the heatsink. Supposedly, Fujipoly makes the best thermal pads...

Do you apply that under the existing heat sync?

 

[pastrychef]

Do you apply that under the existing heat sync?

Yes. On top of the VRM mosfets and below the heatsink.

 

[pastrychef]

Nice! I've never seen a soldered die this clean! Did you use any polishing cream/pad, or just conductonaut?

Yes, with LM+IHS, you should get a 10-15℃ drop, compared to stock. I assume it's the Rockit Copper IHS. If yes, then note that since it's not nickel plated (and its surface, although very flat, not mirror polished), the gallium in the LM (conductonaut is a type of galinstan alloy) will slowly diffuse into the copper to form a copper-gallium alloy (resulting in increased temps), so you might need to replenish/reapply LM for a second time after a few weeks. After that, the diffusion becomes slower.

In the meantime, (if you still have your 8700K) you could try testing with Prime95 (if you haven't already) to make sure that your M/B can handle the extra load. Using Prime95 v26.6 (i.e. non AVX), run a torture test, for at least 30 minutes, with the settings in the attachment. It would draw ~180W of package power, so about the same (if not a bit less) with what an overclocked 9900K would draw under real-world heavy workloads (e.g. handbrake, blender).

If, during the test, the clocks remain stable at 5Ghz (you could check with Intel's power gadget), then you should be good to go, otherwise, you might also need to look into improving your VRM cooling while you are at it.

Good luck and have fun!

Okay. I did the Prime95 test.

Test time: 1:15pm-2:00pm EST
Ambient temps: ~73-75F/23-24C

CPU temps remained right around 78.888-80C throughout the testing.


In the beginning of the test, CPU frequencies dipped a few times but I think that was because I was doing other stuff on the system. Once I left it alone and let it run, it stayed at 5GHz with no dips.


System power draw from the wall went from about 94-96W (Prime95 not running) to about 296-307W.



Best of all, the system remained nearly silent during this testing.

 

[sahandii]

Hey guys, installing my friend's Hackintosh as I'm writing this. I've done a couple through the years, and I'm hoping this one will go well too. Wish me luck! This is the build:

• ASUS ROG STRIX Z370-G (Wi-Fi AC)
  
• Gigabyte GeForce GTX 1080 TI Gaming OC Black 11GB GDDR5X
  
• Samsung EVO 860 1TB
  
• EVGA 650 G3
• Intel i7 - 8700k Coffee Lake
  
• HyperX Fury DDR4 2 x 16GB 3466MHz
  
• Corsair Hydro Hi80 v2

BTW WTH is up with NVIDIA not supporting web drivers as from Mojave?? That's so bad, if I knew this I wouldn't have recommended a NVIDIA GPU for my friend lol. Anyways, I'm installing High Sierra on it for now, I guess.. It doesn't bother him that much, he's gonna use it for video editing and such.

If there's anything I should now, let me know. I'm gonna start with the EFI High Sierra pack first to see how much I can get to work and take it from there. And @pastrychef, well done with this guide man! Very detailed and easy to understand. Cheers guys!

 

[mm2margaret]

Well, NVIDIA claims the drivers have been written and are awaiting Apple's approval.

So, they're fighting - Apple and NVIDIA. But the people who are really hurt are not Apple or NVIDIA as they battle back and forth over this, and that, and every other petty little detail.

It's us, the end users, who are the ones who really get hurt in the ongoing ego battles between two large companies.......

 

[Sofronis]

Yes, I used some Flitz with paper towels to clean it afterwards.

Yes, I have a Rockit Cool Copper IHS coming. I noticed that the stock IHS is a bit convex on the top and after my delidding, it teter-toters on the die far more than I ever remember on any of my other delidding jobs. I'm sure the copper IHS will be much flatter.

Yes, I know that Gallium reacts to copper, but I've never had to reapply liquid metal because of it. As far as I understand, it's does not affect thermal performance. In the past, I've used liquid metal between GPU die and copper plate without ever having to redo it a second time. Temperatures remained fantastic for many months.

I did run Prime95 when I first put my build together and there weren't any problems. I'll try again a bit later. I already received my Fujipoly and I'll be applying at the same time I pull my i7-8700K.

I see you're fully prepared and battle ready . I predict an easy all-core 5.0Ghz, coming your way (unless you're very unlucky with your sample, which I really hope you're not).

After that, you shouldn't have much trouble boosting all the other per-core frequencies by 300Mhz (1-2 cores -> 5.3Ghz, 3-4 cores -> 5.2Ghz, etc.). That should give a nice increase in performance in single/lightly-threaded workloads.

I also saw your prime95 test and you should be fine with the 9900K (especially with the fujipoly pads). Your CPU temps during the test were exceptional, considering the low fan speeds.

 

[Sofronis]

Well, NVIDIA claims the drivers have been written and are awaiting Apple's approval.

So, they're fighting - Apple and NVIDIA. But the people who are really hurt are not Apple or NVIDIA as they battle back and forth over this, and that, and every other petty little detail.

It's us, the end users, who are the ones who really get hurt in the ongoing ego battles between two large companies.......

Well as far as they are concerned, this only affects the users of the vintage/discontinued Mac Pro 5,1 from 2010/2012, which is a diminishingly small demographic, so you can imagine how far down in their priority list this is.