Overclockers rejoice: 12th Gen Intel® Core™ processors introduce new options for tuning performance

Author: Dan Ragland, Principal Engineer, Performance Tuning & Overclocking Architecture

Published in

Intel Tech

9 min readNov 18, 2021

Whether you overclock to squeeze more value from your CPU, complete time-consuming tasks quicker, or pursue the glory that comes from breaking benchmark records, you need a flexible system built for tuning. That’s why unlocked versions of our 12th Gen Intel® Core™ processors support all the knobs and dials overclockers use to optimize their desktop PCs, plus some new features unique to this new generation.

Otherwise known as Alder Lake, the 12th Gen Core architecture introduces significant enhancements engineered to pull more information onto the SoC and process it faster than ever. PCI Express 5.0 connectivity is one example. DDR5 memory support is another. Larger caches keep additional data closer to the compute resources, while wider, deeper, and smarter cores can complete more work per clock cycle. Tasks are also split between two entirely different types of cores managed by the innovative Intel® Thread Director.

Right out of the box, 12th Generation Core CPUs are a big step up from the previous generation. Their advantage grows as you start tweaking, thanks to a combination of capabilities and tools that continue our commitment to delivering the best overclocking experience¹.

A new architecture creates fresh overclocking opportunities

Most overclockers begin their journey by pursuing higher core frequencies. Fundamentally, the process is simple: use a base clock (BCLK) and ratio multiplier combination that yields more performance than the CPU’s stock settings. Intel Extreme Tuning Utility (XTU) software even has a one-click overclocking feature called Intel Speed Optimizer that does this for you automatically.

Alder Lake’s hybrid architecture integrates two types of cores, giving overclockers a couple of different targets for extracting headroom. The big Performance-cores are optimized for high throughput and low latency. They can be controlled independently, allowing you to apply one ratio to all the P-cores through XTU’s Package tuning mode or switch to the Per-Core pane for more granular adjustments.

The Efficient-cores serve a different purpose. Occupying a much smaller footprint on the SoC, they deftly offload background tasks from the P-cores at much lower power levels. Using die area efficiently allows E-cores to scale with multithreaded workloads. They’re deployed in clusters of four cores, so you can apply one multiplier to all the E-cores on a package or overclock each foursome separately.

P-core and E-core overclocking are enhanced by several advanced settings accessible through Intel XTU software and certain motherboard BIOS interfaces. For example, the voltage applied to each P-core and cluster of E-cores can either be set at a constant value or configured as adaptive, which adjusts automatically in response to workload intensity. For further fine-tuning, voltage overrides are also available in Per-Core mode, though the highest voltage setting takes precedence when more than one core is active.

The tuning options get even more focused from there, allowing enthusiasts to tweak the platform’s behavior for specific scenarios, such as overclocking under liquid nitrogen cooling. Sliders establish the power levels and time windows of Turbo Boost Technology. Special offsets kick in when thermally intensive AVX instructions are detected. The Hyper-Threading Technology available on P-cores can be turned on or off on a per-core basis. And additional controls expose multipliers for independently altering the ring/cache and on-die graphics frequencies.

In all, there are more than 20 different knobs for dialing in overclocks on unlocked 12th Gen Core CPUs with K and KF suffixes. (The former include on-die Intel UHD Graphics 770, while the latter require a discrete graphics card.)

K-series processors are unlocked to unleash overclocking potential

At the top of the stack, the Core i9–12900K and Core i9–12900KF feature eight Hyper-Threaded P-cores and eight E-cores that combine to work on 24 software threads concurrently. The P-cores operate at a 3.2GHz base frequency that Turbo Boost Technology ramps up to 5.1GHz when one core is active. Turbo Boost Max Technology 3.0 adds another 100MHz, bumping the ceiling to 5.2GHz on the fastest cores. The E-cores start at 2.4GHz to optimize power consumption and accelerate up to 3.9GHz with a single core active.

Unlocked Core i7–12700K and Core i7–12700KF models come with eight Hyper-Threaded P-cores and four E-cores. The P-cores get a slightly higher 3.6GHz base clock rate and a 4.9GHz peak Turbo Boost frequency that jumps to 5GHz with Turbo Boost Max Technology 3.0 enabled. The E-cores start at 2.7GHz and reach as high as 3.8GHz with one active core. These Core i7s have a 125W thermal design power and 190W maximum Turbo Boost, both adjustable in XTU.

The Core i5–12600K and Core i5–12600KF combine a total of 10 unlocked cores into a package able to operate on 16 threads simultaneously. Six Hyper-Threaded P-cores have a 3.7GHz base frequency that Turbo Boost spins all the way up to 4.9GHz. The four E-cores offer a 2.8GHz base and 3.6GHz peak single-core frequency. Like the other unlocked 12th-Generation Core CPUs, these Core i5s are rated for 125W at their base clock rates. Turbo Boost can push them up to 150W unless you override those limits.

Motherboards with the Intel Z690 chipset complete the picture

Taking full advantage of Alder Lake’s overclocking features requires a platform based on the Intel Z690 chipset. That includes support for BCLK adjustments, which motherboard makers can implement in two ways. The CPU has an internal base clock source that simplifies integration for mainstream designs. That source can be tweaked through the motherboard BIOS, but it’s set and locked during boot-up, which prevents on-the-fly adjustments from the OS.

In a nod to the overclocking community, 12th Gen Core processors preserve an external clock generator path to enable alternatives to the internal source. This special signal path is used to create a BCLK that third-party software tools can change from Windows without rebooting. Even if you ultimately wish to commit an overclock to firmware, being able to quickly adjust and test new settings on the desktop saves tons of time.

Faster RAM helps keep the overclocked cores and buses fed with data, so it’s no surprise that Z690 motherboards can also offer extensive controls for dialing in every aspect of memory performance. Like with the previous generation, memory overclocking won’t be limited to this platform.

The memory overclocking options are extensive

New memory standards don’t come along very often. When they do, it usually takes some time for overclockers to figure out how to juggle frequencies, timings, and voltages for optimal performance. Alder Lake takes the guesswork out of balancing DDR5 bandwidth and latency with Extreme Memory Profile (XMP) 3.0 support. When you pair an XMP-certified module with a compatible motherboard and CPU, it’s easy to access profiles stored on that module to set memory overclocking parameters that are already validated.

The new standard builds on the previous generation with more profiles, plus an option for rewritable ones. The old XMP 2.0 specification for DDR4 gives module manufacturers enough storage on each DIMM for only two profiles. XMP 3.0 has a third read-only slot, carving out room for a wider range of pre-baked configurations. It also adds two rewritable profiles that allow full customization. You can populate those from scratch or load up one of the pre-programmed profiles, make your own modifications, and then save the result to a rewritable slot. To further improve usability, XMP 3.0 introduces support for custom profile names up to 16 characters long. Assigning descriptive names makes it easier to identify profiles meant for gaming, low latency, or other scenarios.

Enthusiasts are sure to appreciate the added flexibility once they start experimenting with DDR5’s tuning options. Beyond layers of timings to adjust, the standard moves power management from the motherboard to the memory module for standardized local control. Three voltage rails — VDD, VDDQ, and VPP — are derived from the DIMM itself. You can roll up your sleeves and adjust them manually or let the optimized XMP 3.0 profiles set those voltages automatically.

Alder Lake is smarter about overclocking memory, too. Rather than running aggressive settings all the time, its new Dynamic Memory Boost Technology intelligently switches between the baseline (default) speed and XMP profile of your choice whenever it detects a performance-bound workload. Maximum memory bandwidth is available when it’s needed.

Intelligent memory-boosting isn’t exclusive to 12th Gen systems running XMP 3.0-certified DDR5. The processor’s dual-channel memory controller also supports DDR4, and Dynamic Memory Boost Technology knows how to dynamically control modules with XMP 2.0 profiles as well. Simply enable it in the motherboard BIOS and choose the XMP profile to use when you need turbocharged performance.

A thinner die and thermal interface complete the package

Overclocking invariably produces more heat for your system to dissipate. Transferring this thermal energy away from the processor die is key to maintaining high frequencies, so Alder Lake desktop CPUs feature an enhanced package designed to improve cooling performance.

Like previous generations, the die is covered by a layer of solder thermal interface material that ensures efficient transfer between the CPU and its integrated heat spreader. But the die is 25% thinner for the 12th Gen, and the accompanying thermal interface is 15% thinner. A thicker integrated heat spreader makes up the difference to complete the stack, resulting in a thermally optimized design with increased cooling capacity.

Overclockers are going to love getting their hands on Alder Lake

Intel has an entire team dedicated to enhancing hardware and software with features that benefit enthusiasts. As Alder Lake’s performance hybrid architecture took shape, they made sure its P-cores and E-cores could both be overclocked via unlocked ratio multipliers. They created the XMP 3.0 specification to simplify DDR5 overclocking. Then they implemented Dynamic Memory Boost to intelligently switch between overclocked XMP profiles and lower-power JEDEC configs. They preserved the external clock generator path to enable real-time BCLK adjustments from Windows. And they rolled out a new version of the Extreme Tuning Utility so enthusiasts can start experimenting on day one.

There’s nothing like an innovative architecture to reset the bar on benchmarking world records. Alder Lake takes a big leap forward thanks to significantly improved P-cores plus the addition of E-cores, and extreme overclockers have already set a new XTU 2.0 Global First Place at 12623 Marks and a World Record on Geekbench Single Core 4 with a Score of 11,511 using a 12th Gen processor with liquid nitrogen cooling².

The 12th Gen Core is so new that it’s going to take some time for enthusiasts to experiment with the hardware and establish overclocking best practices. The most hardcore among them has access to more than 20 knobs and dials controlling the processor itself, plus additional settings for memory. Novices can get in on the action as well through the Intel Speed Optimizer and XMP 3.0 profiles. Regardless of your experience level, Alder Lake offers exciting options for unlocking extra performance.

Notices & Disclaimers

Performance varies by use, configuration and other factors. Learn more at www.Intel.com/PerformanceIndex

1. Based on enhanced overclocking ability enabled by Intel’s comprehensive tools and unique architectural tuning capabilities. Overclocking may void warranty or affect system health. Learn more at intel.com/overclocking. Results may vary.

2. To learn more about the overclocking world records, visit https://hwbot.org/benchmarks/world_records.

Intel technologies may require enabled hardware, software or service activation. Your results may vary. No product or component can be absolutely secure.

Code names are used by Intel to identify products, technologies, or services that are in development and not publicly available. These are not “commercial” names and not intended to function as trademarks.