Reader Note

This will be a rolling update page so check back frequently for new additions! Click here for the latest version of the raw data!

The Intel Core i9-9900K

First off, I need to give a big thank you to Intel for providing a majority of the hardware used in this review. Without them this wouldn’t be possible until the 9900K would have faded even further into irrelevance. Don’t think for a second that I’m going to go easy on Intel just because they sent me some hardware, I mean, this is a costly, beastly and power hungry 8 core processor in a new land of efficient Zen 2 hardware.

Well that is what I’m going to find out. I’ve had nothing but good times with my personally purchased 8700K. Before that CPU I had the 5820K but at around the same time I had that CPU I also purchased the 1800X at launch… That hardware relationship didn’t last long at all. I found nothing but problem after problem with that CPU and it was a major downgrade for me. From what I’m hearing about Zen 2 and the desktop Ryzen 3000 series it seems to be a mixed bag of growing pains but tons of potential. Obviously if AMD would have sent me any of the current line up to test, I’d have been all over it. With the way things are right now I just have to make do with what desperate ol’ Intel decided to throw at me.

Ok, enough throwing sarcastic shade. Let’s break down what I’ve got on the testing block!

Setup

The Hardware

  • Hardware Provided by Intel
    • (What we’re testing) Intel Core i9-9900K
      • SRELS - L907C559 E4 - P0 Revision
    • Noctua NH-D15
    • ASRock Z390 Taichi
      • Using P4.10 BIOS
    • G.Skill Trident Z RGB F4-3466C16D-16GTZR
    • Samsung 970 EVO Plus 250GB NVMe
    • Fractal Design Define S2
    • Seasonic 850w Focus Gold
  • My Personal Hardware
    • EVGA RTX 2080Ti XC Black
    • PNY CS1311 960GB SSD
    • SanDisk Ultra II 480GB SSD

The Tests

First, I’m going to start off with my usual voltage scaling tests to find the minimum voltage required to run all the benchmarks. Those benchmarks are:

  • Asus RealBench – 6 passes
  • Cinebench R15 - 6 passes MP 1 pass SP
  • Cinebench R20 - 6 passes MP 1 pass SP
  • AIDA64 - 3 passes
  • GeekBench 4.3.4 - 1 pass
  • Passmark PerformanceTest - 1 pass

The Testing Results

Intel Core i9-9900K Voltage Scaling

Every test was able to be run with these voltages on this 9900K. For stabilities sake I’d add 0.020v to each one if you want to be totally sure you’ll be stable.

NOTE: These voltages are only what were required for running the benchmarks below and are not indicative of voltages you might require for power virus stability tests.

9900kvoltagescaleplot

Asus RealBench 2.43

9900krealbenchplot

Cinebench R15 & R20

9900kcinebenchplot

AIDA64 Extreme Edition 6.00.5100

AIDA64 CPU Tests

9900kaida64benchplot

AIDA64 Memory Tests

9900kaida64memoryplot

GeekBench 4.3.4

GeekBench 4.3.4 Single Core

GeekBench 4.3.4 Single Core CPU
9900ksinglecoregeekbenchplot
GeekBench 4.3.4 Single Core Memory
9900kgeekbenchsinglecorememoryplot

GeekBench 4.3.4 Multi Core

GeekBench 4.3.4 Multi Core CPU
9900kmulticoregeekbenchplot
GeekBench 4.3.4 Multi Core Memory
9900kgeekbenchhmulticorememoryplot

Passmark PerformanceTest 9.0 Build 1031

Passmark PerformanceTest CPU Tests

9900kpassmarkcpupolt

Passmark PerformanceTest Memory Tests

9900kpassmarkmemoryplot

Comparison Tests

AIDA64 Comparison

AIDA64 Memory

aida64memorycomparisonplot

AIDA64 CPU

aida64cpucomparisonplot

GeekBench Comparison

GeekBench Memory

GeekBench CPU

Sorry about the long names for the Ryzen results. I’ll think of more space friendly names in a later article update.

geekbenchsinglecorecomparisonplot

Efficiency Testing

This is a testing technique that SkyJuice in our discord came up with. Basically, it goes like this:

  • Reset BIOS to default
  • Open HWINFO & Log
  • Run CB R20 2x
  • Stop Log
  • Note the CPU Package Power for the run
  • Throttlestop set PL1 & 2 to noted W
  • Log
  • Run CB again 2x
  • Stop log
  • Throttlestop lower PL 1 & 2 by 5W
  • Repeat until you hit 800mhz all core

When you get down to your base clock you will need to enable clamp mode on both PL values for the clock to go lower than the base.

To set up and clean up the logs I just go and remove all the rows that have less than 100% CPU usage along with that extra row at the bottom that has the headings in it.

The CPU Package Power is taken from the HWINFO64 logs by just selecting all the values and taking the average. Yes, it’s an average of an average, sue me! Same thing with the CPU Package Temperature and the clocks.

For the math parts of this whole thing it’s a simple score divided by GHz and a score divided by power.

816 Testing

9900K 816 On The ASRock Z390 Taichi

This was run using the stock turbo ratios.

9900kpowerefficencyplot

612 Testing

8700K 612 On The Gigabyte Z370 Aorus Gaming 7

This was run using the stock turbo ratios.

8700kpowerefficencyplot

9900K 612 On The ASRock Z390 Taichi

For this run I used the same turbo ratios as the 8700K.

9900kusing8700kratioplot

Thermals

For thermal and power testing the room was kept to 73F/23C using a portable AC unit. The Fractal Design Define S2 had all its panels removed. There are two Fractal Design fans on the front for intake with one on the back for exhaust. The Noctua NH-D15 has both of its fans installed. All fans are at 100% speed.

FIRESTARTER

ASRock Z390 Taichi - CPU Package Power

9900kfirestarterpowerplot

ASRock Z390 Taichi - CPU Package Temperature

9900kfirestartertempplot

ASRock Z390 Taichi - Motherboard Vcore

9900kfirestartervcoreplot

Prime95

ASRock Z390 Taichi - SmallestFFT

9900kp95smallestplot

ASRock Z390 Taichi - SmallFFT

9900kp95smallplot

ASRock Z390 Taichi - LargeFFT

9900kp95largeplot

ASRock Z390 Taichi - Blend

9900kp95blendplot

AIDA64

ASRock Z390 Taichi - AIDA64

9900kaida64thermalspowerplot

Opinion and Conclusion

Time has passed since I first posted up this article and I think I can give a fair assessment of how I feel about it. I can come at this from multiple angles and I will do so under different headings so you can get a full organized picture.

Overclocker’s Perspective

This is basically a 8700K with two more cores and with the heat that comes with that. There is more cache but that is just what happens when you add two more cores, that cache comes with the cores. There have been minimal changes architecturally, but you still get the same tried and true performance that can be pushed to the same limits as an 8700K. You do get solder which I personally dislike but it seems like I’m in the minority with this.

Gamer Perspective

The 9900K is a great buy if you absolutely need the best gaming performance. If you are going to pair this with a 2080Ti there isn’t anything better. High refresh gaming loves this chip and you don’t even need to do anything special to squeeze out that performance. It very much is plug and play.

General User Perspective

Why are you even considering the 9900K? Save your money and get anything else. Any other Skylake will be just as fast in general purpose dicking around in Windows. In fact, GET A LAPTOP OR A TABLET!

Unfiltered Opinion (I’m not saying my opinions above were filtered per se)

If I had to give the 9900K a score out of 10 I’d give it a 6 or a 7. Zen 2 is out and the 9900KS will be coming in a couple weeks. If you want to torture yourself with attempting to cool the KS, go for it! If you are okay with being a glorified beta tester for dynamic turbo clocks, then go ahead and get a Ryzen 3000 series. If Intel wanted this CPU back, I would have no shame in moving back to my delidded 8700K with its copper IHS and liquid metal. Coffee Lake is just Skylake stretched to its absolute limits. The problem isn’t process node at this point, I think its just that Skylake isn’t meant for this kind of core count and frequency when overclocked. It’s not a bad processor! This is just a fast-moving transitional period for processors and Intel needs to reshuffle everything just for sanity’s sake.