Conclusion

This is probably the last of our Alder Lake special investigation articles. We looked at DDR4 vs. DDR5, Alder Lake & Windows 10, E-cores Only Performance and DDR5 Performance Scaling so far. Today, we've done a round of benchmarking at various power limits. Intel's 12th Gen Alder Lake processors are fabricated on the company's 10 nm node, which they recently renamed "Intel 7." It alone brings significant efficiency improvements compared to previous 14 nm processors. In our launch day review, we found that AMD still is more efficient, partly because of their 7 nanometer TSMC production process, but also since Intel dialed the power limits of their unlocked 12th Gen K-Series CPUs up way high. With 241 W, these CPUs use a lot of power, which not only hurts overall efficiency, but also increases the cooling requirements because all this energy is converted to heat.

After our initial launch-day coverage, I received many requests for testing at different power settings to determine whether Alder Lake is really as power hungry as it seems, and what the outlook is for non-K SKUs, which will come at lower power-limit settings. Testing this is easy as Intel provides full control over the PL1 and PL2 power limit values that govern how much power the CPU is allowed to consume. What's important here is that even the locked non-K models support these power-limit adjustments. As we've seen in our Core i9-10900 non-K review back in 2020, this can help unlock a lot of additional performance.

I tested a large range of settings with the default of PL1=PL2=241 W, and then PL1=125 W PL2=241 W, which used to be the default settings of previous K-model processors. I also tested what happens if you don't allow boosting beyond the default TDP (PL1=PL2=125 W). With that done, I went down to 50 W in steps of 25 W to see how performance and efficiency are affected. Of course, it's completely unrealistic for such a processor to run at 50 W, but I feel this still provides some insight into what could happen for mobile and ultramobile models.

If we look at application performance, it becomes clear why Intel switched to this new power-limit configuration. Had they used the classic PL1=125, PL2=241 setting, the Core i9-12900K would end up significantly slower in reviews. Nearly all reviewers use various workloads that fully load the processor, like rendering. Here, the difference between both configurations is quite big. Also, most reviewers use Cinebench, which in its latest version has been changed to run a 10 minute warm-up phase so that PL2 benefits are eliminated completely. While this might be relevant for some heavy workstation tasks, I feel like it completely ignores the reality of today's consumer workloads. Some might even say Intel switched to PL1=PL2=241 W only to beat AMD's Zen 3 Ryzen processors in Cinebench. If we look at other application tests, there's pretty much no benefit from such a high power-limit setting. The classic 125/241 does very well here, even 125/125 runs great, and differences are slim. As soon as you drop below 100 W, the effects on performance become serious. At 75 W, there's a 35% performance loss, and 50 W runs only half as fast as the 241 W CPU. On the other hand, 75 W is only a third of 241 W, yet you still get two thirds of the performance. The reason is that some of our tests are low-threaded or even single-threaded—with just one core fully loaded, the power limit doesn't matter at all.

For gaming, things are completely different. There's virtually no performance difference between the 241 W and 125 W power limit settings, no matter the render resolution. The reason is that games are fairly light in their CPU usage; most aren't even able to run as many concurrent threads as the CPU would be able to. What's really impressive here is that even when the Core i9-12900K is constrained to 100 W power consumption, it's still faster than the processors in AMD's Zen 3 lineup. If Intel wanted to give us "the world's best gaming CPU," they could have configured the i9-12900K to run at 125 W, not 241 W.

We also looked at power consumption and efficiency at these TDP limits and found that there's A LOT of efficiency to be gained at lower power limits. While the default 241/241 configuration is less efficient than all Zen 3 CPUs, the Ryzen 7 5800X is beat as soon as you go below the 200 W limit. At 190/190, the 5600X can no longer keep up. The most energy-efficient configuration turns out to be 75 W, which would make the Core i9-12900K the second most efficient CPU in our test group, only beaten by the Ryzen 9 5950X. Of course, such low limits will drastically reduce performance—you're trading longer runtime for lower overall power usage. The sweet spot is near 125 W, I'd say, but it also depends on the application.

The results from this review are very encouraging for the non-K Alder Lake CPU models that will be released in Q1 2022. It is likely that these processors will offer much better efficiency than the K-models because they run closer to the silicon's maximum efficiency point due to their lower power-limit setting. It's expected that these new models will also tick at lower frequencies, which will further improve efficiency. Last but certainly not least, the price points for these models will definitely be much more interesting to many users, which could be tempting for the value crowd, especially when paired with the lower-priced chipsets that will release next year, too.