Intel has cleared up all doubts: its upcoming Nova Lake architecture will support AVX10.2 and APX, confirming that the company won’t fall behind in the 512-bit vector performance race versus AMD. This information comes from Intel’s own “Instruction Set Extensions Reference Manual No. 60,” which details future capabilities of its processors.
For months there was speculation that Nova Lake might drop support for AVX10.2, which would have left AMD as the only company with desktop 512-bit acceleration. However, the new documentation confirms that both the P-cores (“Coyote Cove”) and the E-cores (“Arctic Wolf”) will implement advanced vector processing, ensuring a significant leap in intensive computing tasks.
What This Means and How It Compares
What this implies is that Nova Lake processors, which are expected to include up to 52 cores, will offer a noticeable advantage in AVX-512 operations, improving performance in applications optimized for data science, AI, rendering, or physics simulation. In other words, the CPU will become a more versatile base platform, capable of accelerating both desktop and server software.
Meanwhile, AMD was the first to offer full support for AVX-512 with its Zen 5 architecture and plans to continue this route with Zen 6, making it clear that both companies will compete directly in this field in the coming years.
Clearly, Intel does not want to lag behind AMD, and the confirmation of compatibility of its next architecture with AVX10.2 and APX is a clear sign of that intention.
