Intel Core i9-9900KS Review Part One: The 5 GHz Powerhouse Fights for the Desktop Crown

Manufacturer: Intel Intel Core i9-9900KS Review Part One: The 5 GHz Powerhouse Fights for the Desktop Crown

With the launch of the Core i9-9900KS Intel is vying for the desktop performance lead, creating as they have a special edition, specially-binned version of their 8-core, 16-thread Core i9-9900K CPU that runs at an unprecedented 5.0 GHz all-core turbo frequency out of the box. To make this new part even more impressive, it also boasts a 400 MHz higher base frequency of 4.0 GHz (compared to 3.60 GHz with the 9900K, and all of this comes at a recommended price of $513-$524 (though early retail pricing has this in the $600 range).

As exciting as this new processor is it still isn’t the ultimate solution to Intel’s recent AMD problem. AMD has a 4-core/8-thread advantage at the $500 price level, which begs the question: can a frequency boost alone – however massive – take the i9-9900KS beyond AMD’s 12-core Ryzen 3900X? There’s only one way to find out. Fortunately we have both CPUs on hand, as well as the original Core i9-9900K.

Part one of this i9-9900KS review will be a three-way CPU performance face-off between powerhouse ~$500 CPUs, with all-new testing conducted using the latest motherboard firmware and platform drivers with both an Intel Z390 and AMD X570 system. GPU results will follow in part two.

Let’s get started!

Intel Core i9-9900KS Review Part One: The 5 GHz Powerhouse Fights for the Desktop Crown - Processors  1
Product Specifications
  • Vertical Segment: Desktop
  • Processor Number: i9-9900KS
  • Lithography: 14 nm
  • Performance
    • Cores: 8
    • Threads: 16
    • Processor Base Frequency: 4.00 GHz
    • Max Turbo Frequency: 5.00 GHz
    • Cache: 16 MB Intel Smart Cache
    • Bus Speed: 8 GT/s
    • TDP: 127 W
  • Memory Specifications
    • Max Memory Size: 128 GB
    • Memory Type: DDR4-2666
    • Memory Channels: 2
    • Max Memory Bandwidth: 41.6 GB/s
    • ECC Memory Supported: No
  • Expansion Options
    • PCI Express Revision: 3.0
    • PCI Express Lanes: 16
    • PCI Express Configurations: Up to 1×16, 2×8, 1×8+2×4
  • Package Specifications
    • Sockets Supported: FCLGA1151
    • Max CPU Configuration: 1
    • Thermal Solution Specification: PCG 2015D
    • TJUNCTION: 100°C
    • Package Size: 37.5mm x 37.5mm
Pricing
  • Recommended Customer Price: $513.00 – $524.00 (“…for direct Intel customers, typically represent 1,000-unit purchase quantities“).
  • Street price: currently $569.99 and up.
Manufacturer Description

“Intel has raised the bar for desktop gaming with the new 9th Gen Intel Core i9-9900KS Special Edition processor. Based on the 9th Gen Intel Core i9-9900K architecture, it’s the world’s best gaming desktop processor made even better and created specifically for extreme gamers who want the most performance possible. This processor demonstrates another innovation milestone for Intel, following last year’s limited edition 8th Gen Intel Core i7-8086K.”

CPU Performance

Before we begin I’ll point out that results from our last CPU launch review – the Ryzen 7 3700X and Ryzen 9 3900X – have been omitted. We are starting from scratch with this review. Why? The biggest issue was that all Ryzen processors must be re-tested with current AGESA firmware (right now that means 1.0.0.3ABBA, soon to be 1.0.0.4x). Some new benchmarks have been selected as the CPU test suite continues to evolve, and more CPUs will be benchmarked in the near future and added to the next review.

Reviewing processors is a never-ending process, and rather than use out-dated results we have chosen to present fresh benchmarks here, using only the latest motherboard firmware and Intel/AMD platform drivers as of this week. Sparse as these charts are with just three CPUs, they are at least current, and as accurate as possible with the usual three repetitions of each test, averaged on the charts to follow.

PC Perspective CPU Test Platforms
Motherboard AMD: ASUS Crosshair VIII HERO WiFi
Intel: ASRock Z390 Phantom Gaming X
Memory AMD: Crucial Ballistix Elite 16GB DDR4-3200 @ 3200 MT/s (16-18-18-38)
Intel: Crucial Ballistix Sport LT 32GB DDR4-3200 @ 3200 MT/s (16-18-18-38)
Storage Corsair Neutron Series XTi 480GB
Power Supply CORSAIR RM1000x 1000W
Operating System Windows 10 64-bit (Version 1903)
GPU Drivers NVIDIA GeForce Game Ready 441.08
Cinebench R20
Intel Core i9-9900KS Review Part One: The 5 GHz Powerhouse Fights for the Desktop Crown - Processors  2

It goes without saying that the overall performance of the 9900KS is going to be higher than the 9900K, so the only thing worth pointing out here with Cinebench is the improvement in single-threaded performance with the new CPU. Single-core boost should be 5.0 GHz with the 9900K, but the 9900KS did outperform its predecessor consistently in single-core CBr20, though just by 2 – 3 points. The multi-core results obviously benefit from the all-core 5.0 GHz boost with this CPU, though AMD still has a huge lead from its 12-core, 24-thread Ryzen 9 3900X.

Blender Benchmark

A handy utility released last year, the Blender benchmark tool offers both a quick and complete CPU or GPU rendering test. For our purposes I ran just the quick CPU test, which provides results from both BMW and Classroom.

Intel Core i9-9900KS Review Part One: The 5 GHz Powerhouse Fights for the Desktop Crown - Processors  3

While shaving valuable seconds off of both the BMW27 and Classroom rendering times, the i9-9900KS is no match for the R9 3900X in either test – though this is an ideal scenario for a processor with more cores and threads available, even if the core clocks with the Ryzen part are lower. Intel’s Core X Series processors might make Blender very interesting – but of course AMD is coming out with a 16-core desktop part soon (and Zen 2 Threadripper looms on the horizon).

X264 Benchmark
Intel Core i9-9900KS Review Part One: The 5 GHz Powerhouse Fights for the Desktop Crown - Processors  4

While Intel enjoys a healthy lead in the first pass of the x264 HD 5.0.1 benchmark, which compresses a raw HD file using the x264 codec, the Ryzen 9 3900X is significantly faster with pass 2. The second pass re-compresses the file created in the first pass, and is always slower than pass 1. Those extra cores and threads from AMD definitely seem to help with the more challenging half of the test. So how does this translate into total encode times?

Intel Core i9-9900KS Review Part One: The 5 GHz Powerhouse Fights for the Desktop Crown - Processors  5

Calculating duration is simple as each pass is encoding the same 11812 frames, so with some quick math the total average time (pass 1 + 2) is 255.3 seconds with the Ryzen 9 3900X, 313.6 seconds with the Core i9-9900KS, and 329.2 seconds from the Core i9-9900K. Due to the huge advantage in the second pass AMD enjoys the overall lead by a large 58.3 second margin. The 9900KS was 15.6 seconds faster overall than the original 9900K in this test, but core counts are king in multi-threaded CPU benchmarks like this so Ryzen is the clear winner here.

7-Zip Benchmark
Intel Core i9-9900KS Review Part One: The 5 GHz Powerhouse Fights for the Desktop Crown - Processors  6

Another small but measurable improvement with the 9900KS over the 9900K, but 7-Zip takes full advantage of the core count provided by the Ryzen part. No surprises here.

3DMark Time Spy CPU Test

While not an obvious pick for a CPU benchmark, the physics test from Time Spy 1.1 (as well as Fire Strike) recently made news as leaked – and subsequently pulled – results teased upcoming Intel and AMD CPU performance. No leaks here, but it’s at least mildly interesting to see exactly how these parts compare.

Intel Core i9-9900KS Review Part One: The 5 GHz Powerhouse Fights for the Desktop Crown - Processors  7

The Core i9-9900KS manages to leapfrog the Ryzen 9 3900X in this test, which seems to favor clock speeds over core count.

Power Consumption

Total power draw was measured at the wall using a Watts Up Pro meter, and the systems were using the same EVGA GeForce GTX 1650 graphics card we reviewed back in May – the idea being to keep GPU power draw to a minimum.

Intel Core i9-9900KS Review Part One: The 5 GHz Powerhouse Fights for the Desktop Crown - Processors  8

The X570 platform with AMD’s 7nm Ryzen 9 3900X pulls a maximum of just 209 total watts under an all-core Cinebench load, while the Z390 system with the new i9-9900KS peaks at 289W. This is an increase of 54W over the i9-9900K, though less than I was expecting given the 5.0 GHz all-core boost clocks with the KS. We will spend more time with this new Intel processor and analyze its boost behavior and thermals as we move forward, but overall power draw is not bad.

Remember, this 289W total includes the GPU, memory, storage, and platform power, and is before the delivered power at an 80 Plus Gold efficiency level. It makes the Ryzen part look that much better in comparison, however.

Conclusion (For Now)

In the works since at least March when it was first teased during Computex, the wait for the special edition Core i9-9900KS (not to be confused with the existing Core i9-9900K), and its 5.0 GHz all-core boost clocks, can most easily be explained if one understands the nature of the binning process. We assume that over the past few months individual chips have been tested and validated to hit these high frequencies, and within a specified thermal design profile. The aggressive 5 GHz all-core target was achieved with a rated TDP 127W, which is actually quite impressive for these clocks on this 14nm process.

Just looking at the CPU-specific benchmarks above, the obvious has been validated: all things being equal, higher clock speeds result in higher performance. The fact that Intel is offering a product with clocks previously attained only by overclocking, and doing it within a reasonable power and voltage target, is impressive. That Intel is not charging a larger premium for this is welcome, and a clear sign of their current competition from AMD.

Intel Core i9-9900KS Review Part One: The 5 GHz Powerhouse Fights for the Desktop Crown - Processors  9

Intel wants to be the undisputed desktop performance leader, and AMD’s Zen 2-powered Ryzen 3000 processors have caused the chip giant to become more aggressive with not just clock speeds, but core counts and pricing, as well. It’s trite at this point but I’ll say it anyway: it’s a great time to be an enthusiast. Competition has driven workstation-class core counts and performance down to the consumer level, and it just keeps getting better. The upcoming HEDT parts from Intel will make this even more interesting, and AMD’s 16-core Ryzen 9 3950X is right around the corner, as well.

More to Come

The next chapter in the story is about gaming performance, as we will revisit the Core i9-9900KS very soon with some new game benchmarks – at lower resolutions to put the bottleneck on the CPU – to see just how much the increase in clock speeds can affect frame rates in games. Stay tuned!

Video News

About The Author

Sebastian Peak

Editor-in-Chief at PC Perspective. Writer of computer stuff, vintage PC nerd, and full-time dad. Still in search of the perfect smartphone. In his nonexistent spare time Sebastian's hobbies include hi-fi audio, guitars, and road bikes. Currently investigating time travel.

14 Comments

  1. JohnGR

    In a world where 12-16 cores are already here, or coming to the mainstream platform, the new 9900KS is nothing more than that old unlocked Pentium CPU. High frequencies, too few cores, compared to the competition. Many “WOW” for the frequencies, nothing interesting about performance.

    Just two questions.

    The 9900KS is 54W above 9900K. So, if 9900KS is 127W, that makes the 9900K a 73W TDP CPU? I think not. So, I wouldn’t get so much impressed about false TDP numbers from Intel or at least I would point out that this is a number about the “UP TO” base frequency. Not to mention that AMD’2 3900K, with 105W TDP is 80W less than the 9900KS. 127W-105W is NOT 80W.

    About the x264 test. Shouldn’t the charts based on overall time? If pass 2 is slower and AMD is 35% faster in that part, then it is MUCH faster overall. An 11% faster 9900KS will not end the job faster, but in fact, it will be much much slower.

    • JohnGR

      PS Where is the “1 year warranty”?

      • Jeremy Hellstrom

        I could have added that in, it is a bit of a dick move

        • JohnGR

          It’s also funny considering that Intel even created a couple of slides when the turbo issue with Ryzen CPUs came out. Slides about how it’s processors offer the Turbo speeds it promises and how it’s processors are more reliable. Fast forward a few weeks and Intel is coming out with a CPU that comes with only one year warranty and, not only for turbo, but even for base frequency the indication “Up to” before the frequency number.

          Right now in mainstream desktop CPUs Intel reminds me of AMD. What is AMD doing all those last years, when it sees that it’s new products are behind Nvidia and Intel products? Overclock them from the factory throwing power efficiency out the window. Intel is doing the same with the 9900KS and probably we will see more factory overclocked models from Intel until it’s 7nm series of processors arrives on shelfs.

          • Jeremy Hellstrom

            We were gently mocking the cooler Intel used on an overclocked chip at a show not that long ago … and that was before we could blame it on the ginger. I have a feeling you may be right, though some parts may see a higher TDP because of the new GPU. On the flip side they have been focusing on power efficiency to a huge degree over the past few generations as opposed to large jumps in processing power. Since neither AMD nor NVIDIA seem to be doing that, and fears of ARM’s imminent arrival in the server room seem to have been premature (at this point) why not boost power consumption and performance at the same time.

            Cutting the warranty is a dick move, as I said, and could hurt them if people actually realize what has happened. On the plus side … maybe AMD is about to start reminding you of Intel in the desktop CPU market!

        • JohnGR

          @Jeremy Hellstrom

          I can’t reply to your last post. I don’t get a reply button under that post. Maybe there is a limit somewhere on how many replies a post can have? to put a limit maybe, in case two people having an argue and keep replying to each other.

          Anyway, Intel can thank ARM for the making of the Atom line and AMD can blame itself for selling it’s mobile GPU devision to Qualcomm and latter stopping any R&D on little gems like Temash and Mullins. Intel is years ahead on power efficiency compared to AMD. The same probably is true for Nvidia. AMD being almost all the time behind in performance and with empty pockets, never really have the opportunity to target power efficiency. 4 out of 5 times they where trying to create products with higher performance, to keep themselves close to the competition and only 1 out of 5 times to improve efficiency.

          • Jeremy Hellstrom

            Hmm, I made the comment inside the admin dashboard. .. wonder if that was the reason. This one is on the normal interface.

        • JohnGR

          @ Jeremy

          No it’s not the admin panel or anything with your account compared to any other account like mine.

          I replied under my own post and I don’t have a reply button under that “Test” post.

          Hope this link at imgur works
          https://imgur.com/vI3AHnl

          • bspadmin

            Reply limit was set to 5 levels, bumped it to 6…do I hear 7?

            • Jeremy Hellstrom

              what’s ASCII for infinite again?

            • bspadmin

    • Sebastian Peak

      The TDPs of these CPUs don’t directly correlate to anything when we actually put them under all-core loads, though total system draw is not the best way to measure or compare differences. I’m sure both companies would disagree with posting all-core Cinebench loads as ‘unrealistic’ for whatever thermal scenario they used to calculate the TDP.

      As to the x264 test, the text output does not include duration, and I intended to add that info once I actually sat down and did the math. I’ve added an encode time chart now and the total times – which absolutely favor the R9 3900X in that test thanks to the far lower second pass times. The test is a 11812 frame HD encode, so I just used average FPS to calculate the duration in the chart above.

      • JohnGR

        (TDP)Yes they don’t, that’s why I don’t find impressive that 127W from Intel. If it was a number that could mean something, then yes. But i think it doesn’t really say anything these days. (x264) That overall chart is something missing from other reviews too, but it is much more informative. When you have to do a job that have two parts, you want to know the overall time, not how fast the first part will do to finish. It was really nice of you adding that. Thanks.

Leave a reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Latest Podcasts

Archive & Timeline

Previous 12 months
Explore: All The Years!