Arm and Samsung this week introduced their joint design-technology co-optimization (DTCO) program for Arm’s next-generation Cortex general-purpose CPU cores in addition to Samsung’s next-generation course of expertise that includes gate-all-around (GAA) multi-bridge-channel field-effect transistors (MBCFETs).
“Optimizing Cortex-X and Cortex-A processors on the most recent Samsung course of node underscores our shared imaginative and prescient to redefine what’s attainable in cellular computing, and we sit up for persevering with to push boundaries to satisfy the relentless efficiency and effectivity calls for of the AI period,” mentioned Chris Bergey, SVP and GM, Shopper Enterprise at Arm.
Underneath this system, the businesses intention to ship tailor-made variations of Cortex-A and Cortex-X cores made on Samsung’s 2 nm-class course of expertise for varied purposes, together with smartphones, datacenters, infrastructure, and varied custom-made system-on-chips. For now, the businesses doesn’t say whether or not they intention to co-optimize Arm’s Cortex cores for Samsung’s 1st era 2 nm manufacturing node referred to as SF2 (due in 2025), or the plan is to optimize these cores for all SF2-series applied sciences, together with SF2 and SF2P.
GAA nanosheet transistors with channels which might be surrounded by gates on all 4 sides have plenty of choices for optimization. For instance, nanosheet channels could be widened to extend drive present and increase efficiency or shrunken to scale back energy consumption and price. Relying on the appliance, Arm and Samsung could have loads of design selections.
Protecting in thoughts that we’re speaking about Cortex-A cores aimed toward all kinds of purposes in addition to Cortex-X cores designed particularly to ship most efficiency, the outcomes of the collaborative work promise to be fairly respectable. Particularly, we’re wanting ahead Cortex-X cores with maximized efficiency, Cortex-A cores with optimized efficiency and energy consumption, and Cortex-A cores with decreased energy consumption.
These days collaboration between IP (mental property) builders, resembling Arm, and foundries, resembling Samsung Foundry, is important to maximize efficiency, scale back energy consumption, and optimize transistor density. The joint work with Arm will be sure that Samsung’s foundry companions could have entry to processor cores that may ship precisely what they want.