Executive Review: How Does Intel’s Core i9 11900K Compare To AMD’s Ryzen 5800X

Intel's Rocket Lake is one of the most anxiously anticipated launch in almost a decade. It marks the first move away from the wider Skylake macro-architecture (since Haswell) and the first major revision for Intel on the 14nm process. This is the first platform from Intel that supports PCIe 4.0 and also the first of the "Jim Keller parts" on Desktop - created by the same guy that designed AMD's Zen. Intel has managed to deliver an outstanding IPC of 19% while staying on the same process (which is a phenomenal gain) and will be competitive with AMD's Zen again for the first time in almost a year.

This time around we will be publishing two reviews, one will contain a core for core comparison against AMD with an executive overview of Rocket Lake while the second will be a deep dive and contain extensive comparisons. You can find the latter over here. So without any further ado, let's get started. For the purposes of this executive review, we are running the Rocket Lake parts on the integrated GPU (you can check out gaming reviews with NVIDIA's unattainable RTX 3000 series in our deep dive).

All testing of the Rocket Lake platform was conducted using the ASUS Maximus XIII Hero motherboard (with the 0703 update) and the integrated GPU (yes I know there is a graphics card in the picture below but please ignore it). To get a better idea of real-world conditions, we opted to go with a fairly average NZXT 240mm AIO instead of a balls-to-the-wall 360mm heavy-flow configuration. This is important because it will impact how Adaptive Boost ( a feature we left on) functions.

Most of these specifications are what you would expect from the typical setup and the motherboard is the one that came with our review kits. The entire test was conducted using the stock settings that came with the motherboard (which notably has max PL1 and Pl2 limits) with one major change: enabling Adaptive Boost in the bios.

Since Adaptive Boost does not count as overclocking according to Intel and we could see no stability issues associated with it, there is no reason not to turn it on. Keep in mind that it requires your CPU to stay under 75c but considering we have a 240mm AIO which is nothing special - it should still represent the performance that an average person has access to.

First up we have the bread and butter of rendering benchmarks: Cinebench. This has been the industry standard for testing the brute force of CPUs for ages past and will continue to be so as far as the eye can see. Cinebench uses a real workload (from Cinema 4D) and is able to showcase the maximum rendering potential of any CPU.

Let's start with the single thread score. This particular number is able to demonstrate how much raw power a single core of any processor has (well technically, a single thread, but let's not get into that right now). It is a very good indicator of determining the architectural prowess of any given processor among other things. This is why it is all the more impressive that Intel, with their Core i9 11900K processor was able to hit a single-thread score of 639 - which is more than AMD's Zen 3 based Ryzen 5800X. This is even more impressive when you consider that Rocket Lake is roughly 1 full process (+ optimization) behind AMD's offering. The Ryzen 5800X follows just behind at 625 points and the Ryzen 5600X in our testing beat out the 11600K with 600 points.

But of course, the era of single-threaded applications is far gone. Up next, we have multi-threaded scores for both AMD and Intel offerings and once again, the Core i9 11900K comes out on top with a score of 6389 versus 6115 from AMD. It is worth noting here that this win is something that came quite late in the development phase. Our configuration utilizes Adaptive Boost and before the update, the Core i9 11900k was trading blows in Cinebench and not achieving a clear win as you see here. It is also worth adding that since we are using a 240mm AIO, our CPU did occasionally exceed 75c, so you are seeing a very realistic, real-world implementation of Adaptive Boost as opposed to a chiller scenario (ahem).

The real world, however, is not as well optimized as Cinebench and that is where Geekbench comes in. Geekbench runs a ton of different algorithms with varying levels of optimization and gives a much more realistic idea of how your processor is going to fare across a very large set of software.

In single-threaded results, the Core i9 11900k maintains its lead with 1859 points and even the Core i5 11600k pulls ahead of the Ryzen 5600X at 1774 points. The tables, however, turn when it comes to multi-threaded results. The Ryzen 5800X is able to beat back the Core i9 11900K by roughly 200 points and take the lead with a score of 10469. The Core i5 11600K on the other hand, maintains its lead over the Ryzen 5600X.

Now let's talk a bit about gaming. If you want detailed gaming results on a Core i9 11900k that has a MAX PL2, be sure to head over to our deep dive published by Hassan. But if you want the overview, TimeSpy is as good a synthetic as it gets.

The Intel Core i9 11900k performs almost identical to the Ryzen 5800X (the difference is inside the margin of error and can change with more runs) while the Ryzen 5600X comfortably beats the Core i5 11600k. That said, we have seen that in actual game testing (which is heavily dependant on clock speeds) both Rocket Lake processors had an edge over their AMD counterparts - so be sure to check out our deep dive.

But you know what is the one thing Intel's Rocket Lake CPUs have that AMD does not? AVX512. This is an incredibly controversial instruction set that in cases can truly make Intel processors shine. If you are in the research industry or someone who has an AVX512 accelerated workload, the choice is very simple. For our tests, we used yCruncher set to 1 billion decimals of pi and single thread (multi-thread uses SMT and is not AVX optimized). The results, as you can see were amazing.

The Core i9 11900K and even the Core i5 11600k effortlessly pull ahead of both the Ryzen 5800X and Ryzen 5600X. Unfortunately, there is a dearth of AVX512 accelerated benchmarks so we really don't have more to add on the subject here but Ian Cutress over at Anandtech has an AVX512 which he won't share so be sure that further shows off AVX512 prowess.

Of course, one of the major differences between 5000 series Ryzen and Intel Rocket Lake is the fact that you get a free iGPU. In an environment where GPUs are super hard to get by, an iGPU could turn out to be the saving grace for a business buying PCs (or even gamers that are building a second rig just to work on). In our testing, we found that the iGPU inside Rocket Lake (HD 750) is a very competent little guy that can handle all sorts of everyday graphical workloads and even some light 1080p gaming.

Both the Core i9 11900k and Core i5 11600k perform identically in the TimeSpy GPU test (because they have the same number of EUs).

Interestingly, however, we saw that in some cases, the Core i9 11900k score would pull very slightly ahead of its twin in the Core i5 11600k. The only explanation for this considering everything else was exactly the same is that the former is benefitting very slightly from the larger cache.

The time has finally, come, I think to talk about the elephant in the room. Intel managed to squeeze an architecture meant for the 10nm process onto a 14nm die (barely) and while the cuticle limit will not allow for more cores, it has come at the cost of a massive TDP increase. While most gamers likely won't care if their processors sips an extra 100 watts, Rocket Lake is very power hungry.

In our testing, we noticed that the Intel Core i9 11900K would consume up to 263w if TDP1 was set to MAX. This is with Adaptive Boost and everything else at default levels. The entirety of the system (including the AIO etc) would draw around 336 watts from the wall.

We also decided to run some conformity tests. Intel is claiming a TDP of 125W, so what would happen if we forced the processors to adhere to this spec. Keep in mind however that not even AMD processors adhere to that spec and motherboards typically freely allow these limits to be surpassed. They are only ever followed in business environments or laptops. By setting the PL1 and PL2 to 125W, we saw the following results:

At 125W TDP, the Cinebench R20 score of the 11900K fell from 6389 to 3968. A reduction of 37%. Power efficiency, however, increased from 24.3 points per watt to 31.7 points per watt. The Core i5 11600k on the other hand, only experienced a performance reduction of17% and increased power efficiency from 21 points per watt to 29 points per watt.

Intel's Rocket Lake is an exceptional series in many ways. It is the first backported architecture, the first PCIe 4.0 architecture from Intel and the first Jim Keller part (you know, the guy who made Zen). But it is also in many ways severely limited by the underlying process. This architecture was never meant to exist on the 14nm process and because of the exceptionally large footprint, it cannot be raised to have more than 8 cores on the mainstream die. This is very problematic because Intel's last Core i9 part was actually a 10 core and it also allows AMD to compete on the basis of core count.

The solution? pricing. If Intel can price this series correctly - and maintain that pricing along with supply - it will potentially have a winner on its hands. AMD's Ryzen lineup is prone to severe supply issues (because of the central bottleneck at TSMC) and considering Intel's foundries do not have the same problem, their bane (14nm) could turn out to be their biggest advantage. Because it doesn't matter how good something is (I am looking at you NVIDIA RTX 3000 series), if you cannot buy it. Competition is always good for the consumer and with Pat taking the helm of Intel, we might see the company make a comeback in the coming years.

Depending on how quickly their 10nm volume ramp occurs, RKL might also be the shortest-lived family from the company and if that happens - you would best wait for Alder Lake. As a concluding note, I will say this much: Rocket Lake marks Intel as being competitive with AMD again (up to 8 cores at any rate) and depending on supply, could see the company defend its position with (finally) a PCIe 4.0 platform in its ranks.

An anxiously awaited update from Intel that finally introduces PCIe 4.0 to its lineup and an architectural update that was 6 years in the making. Affordable pricing and large supply could make this a winner.

Subscribe to get an everyday digest of the latest technology news in your inbox

Some posts on wccftech.com may contain affiliate links. We are a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for sites to earn advertising fees by advertising and linking to amazon.com

© 2024 WCCF TECH INC. 700 - 401 West Georgia Street, Vancouver, BC, Canada

You need to login to use this feature.