Intel’s transition to a hybrid architecture for its CPUs began with the Alder Lake series in 2021, a significant shift that combined performance and efficiency cores within a single package. This approach mirrors the design principles of ARM-based chips and has seen gradual refinements since its inception. According to Robert Hallock, Intel’s Vice President, the hybrid architecture has reached a level of maturity from a hardware perspective, although a “unified core” is still in development.
Performance Insights from Intel’s VP
In a recent discussion with PC Games Hardware, Hallock attributed some of the performance challenges associated with hybrid chips to software limitations. He noted that many users disable the efficiency cores (E-cores) on modern Intel CPUs, believing it enhances performance. However, Hallock clarified that the performance difference between using only performance cores (P-cores) and utilizing both types of cores is minimal, estimated at around 1%.
Hallock elaborated that the early implementation of Intel’s Thread Director, which manages core allocation, contributed to improved performance when only P-cores were engaged. The Windows task scheduler, without the guidance of the Thread Director, struggles to optimize processes effectively across the available cores. With advancements in software optimization, the E-cores, while less powerful, can still significantly enhance overall performance when properly utilized. Tools like Intel’s Advanced Power Optimization (APO) are designed to assist in this optimization process.
He emphasized the critical role of software in enhancing the PC gaming experience, stating that the gaming community, particularly enthusiasts, may be underestimating its importance. Hallock believes that software optimization represents a new frontier in performance enhancement, allowing for greater efficiency and performance extraction from existing silicon resources. The introduction of features like binary optimization in the upcoming Arrow Lake refresh chips exemplifies this potential, even if current applications and games have yet to fully leverage it.
Hallock pointed out that while upgrading hardware can yield immediate performance gains, a significant portion of potential performance—estimated between 10% to 30%—remains untapped due to suboptimal game optimization for specific CPUs. In contrast, AMD has approached this challenge by integrating additional SRAM through its 3D V-cache technology, which meets the L3 cache demands more swiftly, thereby improving gaming frame rates.
Looking ahead, Nova Lake is set to introduce a similar solution with its Big Last Level Cache (bLLC), although this remains a hardware-centric approach. Hallock’s insights suggest that the substantial performance waiting to be unlocked through improved software optimization is not merely a theoretical concept but a pressing reality. He subtly indicated that developers and engineers have predominantly optimized for AMD’s more traditional silicon, which inadvertently limits the full potential of Intel’s innovative hybrid architecture.
