Intel Prepares "Razor Lake" in Three Flavors: S, U, and L
Intel appears to have prepared three versions of its next-generation "Razor Lake" platform, designated as S, U, and L, which provides a clearer picture of the company's future lineup, according to the latest set of Linux…
Archer Tale2 min read

Intel appears to have prepared three versions of its next-generation "Razor Lake" platform, designated as S, U, and L, which provides a clearer picture of the company's future lineup, according to the latest set of Linux kernel patches. Additionally, there is a separate "Razor Lake" entry without a suffix, bringing the total to four processor identifiers. Intel engineer Tony Luck submitted the patch on November 13, following the publication of model numbers in Intel's updated instruction set reference. All four client entries belong to CPU Family 18. The desktop models use identifiers 0x06 and 0x0A for "Razor Lake" and "Razor Lake-S," respectively, while the mobile entries assign 0x07 to "Razor Lake-L" and 0x09 to "Razor Lake-U." There is no "Razor Lake-H" definition in this submission. However, these internal labels do not determine the final product range, and Intel has yet to explain what differentiates the underlying configurations.
Compiler work published on the same day provides another detail: GCC initially treats "Razor Lake" as an alias of "Nova Lake." Both targets share the instruction capabilities listed in the patch, including AVX10.2 and APX. This informs developers about which instructions the compiler can generate but reveals little about potential changes to core designs, cache capacity, or performance. A shared compiler profile alone cannot establish how closely the finished processors will resemble one another. Read full story
Compiler work published on the same day provides another detail: GCC initially treats "Razor Lake" as an alias of "Nova Lake." Both targets share the instruction capabilities listed in the patch, including AVX10.2 and APX. This informs developers about which instructions the compiler can generate but reveals little about potential changes to core designs, cache capacity, or performance. A shared compiler profile alone cannot establish how closely the finished processors will resemble one another. Read full story
GDC November 2026 survey shows 40% of developers using generative AI
403 hyperscale US data centers used staggering 82 terawatts of electricity
Medaka Kuroiwa is Impervious to My Charms Season 2 to Stream on Crunchyroll
Discord's New App For Linux Shown Off With Windows Feature Parity