Exploring CPU Power Scaling On Core i9-13900K and Ryzen 9 7950X

One of the biggest running gags on social media and Reddit is how hot and power-hungry CPUs have gotten over the years. Once upon a time, a flagship x86 CPU didn’t even need a heatsink, but now it can saturate an entire radiator. Thankfully, it’s not on the level of a nuclear reactor, as the memes say, but it’s not a nuclear reactor either, as kids these days say. nothing burgerDesigning for higher TDPs and greater power consumption allowed chip makers to continue pushing the envelope in terms of performance. This is no easy task in the post Dennard world, but it certainly introduced some new headaches around power consumption and heat in the process. For better or worse, AMD and Intel’s latest flagship chips are examples.

But despite these general trends, this doesn’t mean high-performance desktop CPUs should waste power. In AMD’s Ryzen 9 7950X review, testing showed that even though it’s capped at 65 Watts for a modern day pedestrian, the 7950X can deliver a decent amount of performance at less than half the power consumption of its normal counterpart.

Pardon the pun, power efficiency is a hot topic these days, with enthusiasts looking to save energy bills while enjoying fast CPU performance (especially in Europe). Along with the silicon capabilities of AMD’s Raphael and Intel’s Raptor Lake-S platforms, it packs as many Joules into the chip as possible. All the while, the small form factor market remains a strong outpost for high-efficiency chips, and cooler chips are needed to build smaller, more compact systems that can avoid the need for massive cooling systems. Essential.

All that said, while it’s nice to see the envelope being pushed in terms of peak performance, the typical focus on how unlocked chips scale when overclocked (pushing up CPU frequency and CPU VCore voltage) is , is just one way to look at the overall CPU. performance. So today we’re going to do the opposite and look at the user’s overall energy efficiency and see what happens when you aim for his spot on the voltage/frequency curve suite. To that end, today we explore how the Intel Core i9-13900K and AMD Ryzen 9 7950X perform at different power levels and see what benefits power scaling can offer compared to stock settings. doing.

The ever-changing definition of TDP (Thermal Design Power)

One of the most frustrating topics around processors has to do with TDP. Specifically, what is TDP? The term has a fixed definition (thermal design power), but what it actually means in terms of product specifications varies by hardware manufacturer. Even Intel and AMD define TDP slightly differently (despite sharing the x86 CPU market), making for a complex and sometimes confusing term. In short, TDP is no longer a hard and fast guide to the maximum power consumption of a particular processor or how much heat energy is dissipated from it.

This is an issue that we have given many column inches over the years, so here are our previous articles on power and TDP.

Starting with Intel, the company assigns official TDP numbers to their processors, but the reality is that unlocked processors aren’t limited to one assigned power number. At best, Intel provides a range of TDP values ​​based on the base frequency of the chip, but Power Level 1 (PL1) and Power Level 2 (PL2) ratings sit on top of it and define very important power parameters. To do. (and all advertised) turbo clock speeds.

On the other hand, seeing how AMD defines its TDP numbers has become more difficult with the release of its Ryzen 7000 series processors and accompanying AM5 platform. AMD puts out his default TDP numbers for his Ryzen 7000 series, which isn’t as tight a cap as Intel’s. Another factor is what AMD calls Package Power Tracking, or PPT for short. A defined TDP value is assigned to a processor based on its default frequency level, but PPT is based on what a real socket can provide in terms of power (and what a flagship chip tuned for maximum performance is going to consume). defined as the limit of

In short, on virtually any consumer-level motherboard paired with a suitable cooler, Any Unlocked processors, including the two we’re eyeing today, circumvent the chip’s base TDP rating to offer the highest possible turbo clock speeds, limited only by thermal and electrical supply limitations. increase. Take the Ryzen 9 7950X as an example, it has a default TDP of 170W, but his actual PPT wattage is defined as 230W (sustained current 160A).

The good news, however, is that both AMD and Intel platforms allow users to easily adjust these power limits. Many motherboards for both platforms offer a number of adjustable values ​​to increase or decrease these limits, and even within Windows, AMD’s Ryzen Master and Intel’s XTU’s allow you to make the necessary changes. I have a software package like

Finding the sweet spot balance of performance, power and heat

AMD offers a new feature called ECO Mode for their latest series of processors. This reduces a 170 W TDP rated processor to 105 W (and a 105 W TDP rated processor to 65 W). We reviewed the AMD Ryzen 9 7950X processor and found some interesting results when using it at 65W instead of the default value of 170W..

The most important thing we observed was that at 65W, the Ryzen 9 7950X processor outperformed the Intel Core i9-12900K on default settings in the CineBench R23 multithreaded test. Since then, Intel has released its 13th Gen Core series under the codename Raptor Lake, thus leveling the overall playing field somewhat. Still, the 7950X was able to deliver 80% of peak performance while consuming less than half the power, demonstrating the potential value of dialing down the power consumption of flagship CPUs.

These findings, in turn, inspired the paper behind this article. I wondered how both the Core i9-13900K and the Ryzen 9 7950X perform when used with different power caps applied. Run at power levels to test just that.

I have selected and tested several “TDP”s including:

• Intel Core i9-13900K Default Settings (125 W Base, 253 W Turbo)
• AMD Ryzen 9 7950X with default settings (170 W base, 230 W PPT)
• both 125W
• both 105W
• both 65W
• both 35W

Through all of this, it’s important to note that the Intel Core i9-13900K and the AMD Ryzen 9 7950X have an important difference in terms of construction: chiplets. (IMC), and iGPU all built within a single die, while AMD’s Ryzen 9 7950X is built on a chiplet design. As a result, the 7950X uses three different chiplets for all major functions, with two 8-core Core Complex Dies (CCDs) and one IO Die (IOD) that contains the memory controller, PCIe, and graphics. increase.

Screenshot of AMD’s Ryzen Master Overclocking software during testing

This means that setting a defined power limit on the Core i9-13900K will cap the power from all the previously mentioned variables. Doing the same thing on the Ryzen 9 7950X won’t have the exact same effect, but that’s because the IOD and CCD work semi-independently, albeit via AMD’s Infinity Fabric interconnect. In order to properly limit the power of the Ryzen 9 7950X, it is necessary to apply a package power tracking limit, such as setting a value of 125W for the PPT and not just the CPU cores to properly limit the power limit.

Test bed and setup

Intel and AMD both use different platforms and sockets, so the Core i9-13900K uses the MSI MPG Z790 Carbon WIFI and the Ryzen 9 7950X uses the GIGABYTE X670E Aorus Master. To maintain a level playing field, we use SK Hynix Platinum P41 2TB NVMe storage drives and AMD Radeon RX 6950 XT graphics cards.

Normally we would rely on the JEDEC memory settings defined by each platform, but in this case we chose to test both platforms with the same SK Hynix DDR5-5600B CL46 memory configuration. This is the memory we used to review the Intel Core i9-13900K, but it’s faster than the highest official JEDEC settings supported by AMD’s Ryzen 7000 series platform (DDR5-5200). In any case, I wanted to make as many apples-to-apples comparisons as possible, including exploring the benefits, if any, of keeping memory at a constant performance and power level to limit system power consumption. .

In terms of benchmarks, we have selected a subset of rendering and encoding tests that we believe highlight differences in processor performance at various power levels. For games, I chose Total War: Warhammer 3. This is a title that is notorious for being heavily CPU dependent and Borderlands 3 is more graphically demanding. We thought it would be wise to test each title individually, but if you’d like to request more titles from our CPU 2023 benchmark suite, feel free to add them as we find time to gather and collate the data.