• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Core i9 9900k 5GHz on one of the cheapest Z370 motherboards - Possible? Yes!

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

yoadknux

Member
Joined
May 6, 2016
Hi everyone,

Just want to share my own personal experience with the i9 9900k on the Gigabyte Z370 HD3.

Some background - I originally had this board with 8700k - I did not like the board one bit, it was a pain to get to 5GHz. It took -2 AVX offset, voltage was all over the place, temps were nuts (90c+ cpu). It's in general not known to be a good board. How I ended up with a 9900k on it is a long story.

I plugged it into my system and it booted without problems. I immediately went to the bios and started overclocking: 5GHz, No AVX offset, Uncore 43, 1.35V. I disabled C states and changed the loadline calibration settings (AC/DC: Performance, Vcore Loadline: High). I did not touch power limits in bios.
XMP settings from Corsair 16G 3200MHz enabled. Cooled By Arctic Liquid Freezer II 280mm inside a Cooler Master H500P Mesh.

It booted at 5GHz, went into Windows. I started benching it. The results:

Cinebench R15
lB824Jq.png

Cinebench R23
2KPiTEt.png

The hottest core reached 91c, which is actually not that bad. R23 is a 10 minute continuous loop run, it gives a better indication of throttling and stability than R15.
These results are pretty good and match the results for i9 9900k 5GHz with top Z390 boards (as well as i7 10700k 5GHz with Z490 boards).
There is an important point here that I have to emphasize. The VRMs on this board need active cooling for OC. Even though I have decent airflow in my case - VRM area wide open (see picture), with unblocked 140mm exhausts - it wasn't enough. What do I mean by not enough: VRM temps reached a maximum of 125c. At 120c-125c, core clock drops to 3.6. What I did to resolve this issue was to place a 92mm slim Noctua fan on the VRM, and as you can see in the R23 picture, max VRM temp was 89c, no throttle.

K3ldS5A.jpg
(Yes, I know that RAM stick looks lonely, I have another stick arriving on Sunday)

What still needs to be done - more stress tests - I only ran RealBench for 30 minutes + 30 minutes of R23 loops as indicators of stability. Probably good enough not to run into a BSOD while gaming or doing office work.
If it passes a few hours of RealBench I'll try pushing for higher clocks (5.1-5.3 not unheard of for 9900k), but the results thus far are pleasing.

In short - You can you reach 5GHz on 9900k with a Gigabyte Z370 HD3, benchmark scores show no decrease in performance, VRM needs a fan for sustained heavy load. Still need more tests for max overclock + everyday and absolute stability.

Update: 9 days in: Stability thus far has been reasonable. I passed 1.5h small FFT (no AVX), 1.5h blend, 1.5h RealBench. Typical 24/7 usage (gaming, web browsing, etc) - Perfect stability at 5GHz, not a single BSOD.
Full blown P95 AVX was not tested - I don't think it's needed

Another update: 5.1GHz is R15 and R23 stable.
uVl2PJr.png
 
Ecellent so far well done get the memory timings sorted out too .
Perhaps you could change cooling for vrms you can buy aftermarket with heat pipes £20 Nice start
 
Ecellent so far well done get the memory timings sorted out too .
Perhaps you could change cooling for vrms you can buy aftermarket with heat pipes £20 Nice start
Hi, thanks for your response. I added a 2nd stick and they are XMP enabled with dual channel. Stability of 5.0 is great and I've been running it since the post with no issues. 5.1 I was able to run R23 with same voltage, 5.2 crashed.
Yes, I consider upgrading the VRM cooling, I look back at my old abit IP35 Pro board, this thing had proper heatsinks with a copper heatpipe:
28d9fd59eab3e683879f41bf4827721f-970-80.jpg
I think VRM heatsinks of newer boards are more inspired by aesthetics rather than thermal performance
 
Thats some good looking cooling . With your memory i was more thinking looking into the clock timings youll be able to tighten them up
 
I think VRM heatsinks of newer boards are more inspired by aesthetics rather than thermal performance
Well, yes and no. Note that Abit wasn't a bottom-of-the-barrel budget option like the HD3 is. As the price/tier of motherboard goes up, so does the quality of the VRMs as well as the cooling solution. The 50A MOSFETs just aren't meant to be driven so hard, so they don't put beefy cooling on it. There are plenty of premium mid-range+ boards with large heatsinks and even heatpipes connecting both parts of the VRM (and sometimes the chipset). But this is more a function of being the cheapest board on the platform than anything.
 
Well, yes and no. Note that Abit wasn't a bottom-of-the-barrel budget option like the HD3 is. As the price/tier of motherboard goes up, so does the quality of the VRMs as well as the cooling solution. The 50A MOSFETs just aren't meant to be driven so hard, so they don't put beefy cooling on it. There are plenty of premium mid-range+ boards with large heatsinks and even heatpipes connecting both parts of the VRM (and sometimes the chipset). But this is more a function of being the cheapest board on the platform than anything.
I haven't seen a single Z370 / Z390 board with a heatpipe. Top Z390 boards by Gigabyte and Asus just use a big piece of metal as a heatsink (not a spread fins heatsink like abit), and actually it is now trendy to cover the heatsink with a piece of plastic or whatever that is that conceals the back IO ports
 
I was talking more about modern motherboards, not that generation. ;)

I wouldn't worry too much about the plastic covering part of the VRM heatsink and on top of the rear I/O area. I get ya, but it's really of no concern. :)
 
I feel the need to post more results because I'm very surprised by what I'm getting.

5.1GHz at 1.32V core passes R23 with no problems (see OP).

It also passed 1h P95 Small FFT no AVX:

3auyKC0.png

This time I ran the stress test on an open case... and the core temps are... up to 85c? how in the world is that possible? Both CoreTemp and AB reported cpu wattage of up to 198W during the test.
1) Damn, is the liquid freezer with the ML fans that good? :shock:
2) 9900k at 5.1GHz runs cooler than 8700k 4.9GHz on same board - solder does matter
3) Why is 5.1 stable on this crap board :screwy:
 
3) Why is 5.1 stable on this crap board
Because people put way too much into VRMs, is why. An overwhelming majority of boards, including cheapO budget boards, have enough in them to support overclocking. Long gone are the days of AMD FX where literally only a handful of could overclock without rigging fans on the front/back of the VRMs to keep them cool. Now that doesn't mean some boards are potatoes at that budget level and can't, but, I can't say I'm surprised, honestly. :)
 
I don't think people realize how much punishment VRMs can actually take. Most of the components are rated for 100C or higher operation with lifespans at that temp in the thousands or tens of thousands of hours. The nice thing about overkill VRM setups is that they spread enough load out that they run much cooler even with relatively poor cooling design. All these fin stacks are best utilized with fans because the fin pitch isn't conducive to passive cooling. The massive slabs of aluminum you see on other VRM sinks are just a 'sink'. They soak up the heat but they don't really have a lot of surface area to dissipate it. Sure, beefy VRMS add cost, and they look cool, but they're really just like super big fuel injectors for endurance racing when you're mainly cruising on the highway most days.

That said, I'll pay for a quality VRM that can get the job done as long as I'm not overpaying for a board. Because I like nice stuff.
 
Back