CPUs

Intel 11th Gen Rocket Lake-S PL1, PL2 Power Draw Explained: Turbo Boost, Thermal Velocity Boost, and Adaptive Boost

Intel’s 11th Gen Rocket Lake-S processors feature the first upgrade to the company’s desktop lineup since 2015’s Skylake. Dubbed Cypress Cove, the new CPUs feature a 14nm backport of the 10nm Sunny Cove core. Although much of the IPC advances have been retained, the efficiency is much lower than the original 10nm design. Larger transistors mean a larger die, higher thermal output, and increased power draw. Before we move onto the PL1 (stock) and PL2 (boost) power values, here’s a quick primer on what these and some of the other primarily used terms mean:

PL1, PL2 and Tau with TB 2.0

The marketed TDP of Intel CPUs pertains to the power draw at stock frequencies, otherwise known as the PL1 value. When the CPUs boost to higher “Turbo Frequencies”, the power draw increases considerably. This power state or profile is called PL2 (Power Level 2) while the stock (marketed) power state is called PL1.

Then there’s Tau which determines how long will your chip stay at the boosted “PL2” state. In most single to quad-threaded applications, the CPU power draw will stay under PL2 despite reaching the maximum Turbo Boost frequency. However, when you’re running a more intensive multi-threaded application, say Cinebench or 7-zip, all your CPU cores will kick into action. In such a scenario, the power draw will quickly soar up to the PL2 limit and the processor will reduce the all-core frequency to keep the power draw in check.

Turbo Boost with 10nm Ice Lake

Under standard conditions, the processor would adhere to the PL2 power state and the accompanying boost frequency for a duration “Tau”, usually 56 seconds, and then return to PL1 (stock), thereby reducing the clock speed to the base frequency. This would continue till the intensive workload is completed and the CPU returns to idle. After that, it’d return to PL2 power limits and repeat the above-explained behavior. However, for better or for worse, Intel doesn’t enforce the PL2 and Tau values, and motherboard vendors are free to set them as they choose in the firmware. In most cases, your CPU will run at the maximum Turbo Boost clock (PL2 power state) for infinity or as long as the thermals allow it (while ignoring the Tau time limit). More on this here.

10th Gen ProcessorPL1 Power (W)PL2 Power (W)Tau (Seconds)
Core i9-10900K12525056
Core i7-10700K12522956
Core i5-10600K12518256
Core i9-109006522428
Core i7-107006522428
Core i5-10600, Core i5-10500, Core i5-104006513428
Core i3-10320, Core i3-10300, Core i3-10100659028
Pentium Gold 6500, Pentium Gold 6400, Celeron G5920, Celeron G5900585828
Core i9-10900T3512328
Core i7-10000T3512328
Core i5-10000T359228
Core i3-10000T355528
Pentium Gold G6500T, Pentium Gold G6400T, Celeron 5900T354228

Although the PL1 and PL2 values of the 10th Gen and 11th Gen lineups are more or less the same, there’s more to it than meets the eye. The newer Rocket Lake-S parts support AVX-512 as well which can dramatically speed up certain AVX workloads…at the cost of power, a lot of power.

11th Gen ProcessorPL1 Power (W)PL2 Power (W)Tau (Seconds)
Core i9-11900K12525156
Core i7-11700K12525156
Core i5-11600K12522456
Core i9-119006522428/56
Core i7-117006522428/56
Core i5-11600, Core i5-11500, Core i5-114006515428/56
Core i9-11900T3511528
Core i7-11700T3511528
Core i5-11600T358428
Power Values for the lower-end Core i3 and Pentium and Celeron SKUs are unchanged

For example, when paired with a Z590 board and 16GB of DDR4-3600 memory, the Core i7-11700K drew 339W in AIDA64 and a whopping 394W in Prime95, both AVX tests. Similarly, the 11600K also draws close to 300W with workloads with AVX512 in action. However, in most gaming and rendering workloads, the load power draw should be much closer to the PL2 values of 250W and 224W for the Core i7/i9 and Core i5, respectively.

More here

Turbo Boost vs Thermal Velocity Boost vs Adaptive Boost

The 11th Gen Rocket Lake-S CPUs make Intel’s boost algorithm even more confusing by adding a third, wait, no fourth boost state called Adaptive Boost, in addition to the already existing Turbo Boost 2.0, Turbo Boost Max (3.0), and Thermal Velocity Boost (in ascending order). Turbo Boost 2.0 is the standard boost algorithm that has existed since the early Skylake era. It boosts the two fastest cores to the TB 2.0 clock while the remaining cores hover between 4.6GHz and 5GHz.

AnandTech’s bar chart explains it really well!

Turbo Boost 3.0 or Turbo Boost Max allows all the 10th Gen and 11th Gen CPUs to boost by another 100MHz, but only on the two fastest cores. Then there’s Thermal Velocity Boost which boosts the Core i9 lineup by another 100MHz across all cores, on top of TB 2.0 and 3.0, but only if the CPU temp is below 70 degrees. Finally, there’s Adaptive Boost which allows six of the slowest cores on the Core i9-11900K to boost to the Turbo Boost 2.0 value. As you might have guessed, the maximum boost clock under Adaptive Boost will depend on the available power and thermal headroom and is mainly applicable to gaming workloads. It likely won’t work with heavily threaded tasks like rendering (Cinebench, Blender, Corona) and definitely won’t work with AVX512 where the boost clock will mostly stay under 5GHz despite consuming a lot more power:

Areej

Computer Engineering dropout (3 years), writer, journalist, and amateur poet. I started my first technology blog, Techquila while in college to address my hardware passion. Although largely successful, it was a classic example of too many people trying out multiple different things but getting nothing done. Left in late 2019 and been working on Hardware Times ever since.

Leave a Reply

Your email address will not be published. Required fields are marked *

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

Back to top button