AMD APU

Lore & Background

The AMD Fusion project began in 2006 with the aim of developing a system on a chip combining a CPU with a GPU on a single die, a goal advanced by AMD's acquisition of graphics chipset manufacturer ATI in 2006. The project required three internal iterations of the Fusion concept before a product deemed worthy of release was created, with delays attributed to technical difficulties of combining CPU and GPU at a 45 nm process and conflicting views on the roles of CPU and GPU within the project. The first generation APUs, Llano for high-performance and Brazos for low-power devices, were announced in January 2011 and launched on June 14, 2011. Llano featured K10 CPU cores and a Radeon HD 6000 series GPU on the FM1 socket, while Brazos used the Bobcat microarchitecture and a Radeon HD 6000 series GPU. The second generation, Trinity for high-performance and Brazos-2 for low-power, was announced in June 2012, with Trinity featuring Piledriver CPU cores and Radeon HD 7000 series GPU cores on the FM2 socket. A subsequent APU based on Piledriver, codenamed Richland, was released in March 2013 for laptops and June 2013 for desktops. The third generation Kaveri for high-performance devices launched in January 2014, based on the Steamroller architecture, while low-power Kabini and Temash, based on Jaguar, were announced in summer 2013. Following the Zen microarchitecture, Ryzen and Athlon APUs released as Raven Ridge on the DDR4 platform, after Bristol Ridge a year prior. AMD also supplied semi-custom APUs for consoles starting with the Sony PlayStation 4 and Microsoft Xbox One eighth generation video game consoles, and subsequently for the Xbox Series X|S and PlayStation 5.

Reader's Guide

The AMD APU line represents a significant architectural shift by integrating CPU and GPU on a single die, enabling heterogeneous computing through the Heterogeneous System Architecture (HSA), of which AMD is a founding member. This integration allowed for graphics accelerators like OpenCL to be used with the integrated graphics processor, with the goal of eventually featuring 'heterogeneous cores' capable of processing both CPU and GPU work automatically. The APU's impact is seen across multiple market segments: from high-performance desktop and laptop chips like Llano, Trinity, and Kaveri, to low-power platforms like Brazos and Kabini for netbooks and tablets. The semi-custom APUs for major video game consoles demonstrated the architecture's scalability and versatility. However, the project faced delays due to technical challenges and internal disagreements, and AMD later rebranded the Fusion platform as HSA following a trademark infringement lawsuit by the Swiss company Arctic over the name 'Fusion'. With the introduction of Zen-based processors, AMD renamed their APUs as Ryzen with Radeon Graphics and Athlon with Radeon Graphics, using a G suffix on desktop model numbers to distinguish them from regular processors and to move away from the Bulldozer-era A-series APUs.

Did You Know?

• The AMD Fusion project started in 2006 and required three internal iterations before a product was deemed worthy of release.
• AMD's acquisition of ATI in 2006 advanced the Fusion project.
• AMD supplied semi-custom APUs for the Sony PlayStation 4, Microsoft Xbox One, Xbox Series X|S, and PlayStation 5.
• The Fusion name was dropped after a trademark infringement lawsuit by the Swiss company Arctic.

From Fusion to Reality: The Long Road to a Unified Chip

The concept of merging a general-purpose processor with a 3D graphics engine onto a single silicon die traces back to 2006, when AMD launched what it called the Fusion project. That same year, the company's acquisition of graphics chipset maker ATI provided the critical GPU expertise needed to make the vision tangible. However, turning the blueprint into a shippable product proved far more difficult than anticipated. AMD reportedly cycled through three internal iterations of the Fusion concept before settling on a design it considered release-worthy. The primary obstacles were twofold: the sheer technical challenge of integrating CPU and GPU logic at a 45-nanometer process node, and internal disagreements over how much responsibility each half of the chip should carry. After years of development, the first generation finally materialized. Llano, aimed at high-performance desktops and laptops, and Brazos, targeting low-power devices, were unveiled on January 4, 2011, at the Consumer Electronics Show in Las Vegas, with actual products hitting shelves on June 14 of that year.

A Decade of Iteration: From Llano to the Zen Era

The APU lineup evolved through several distinct generations over more than a decade. The initial Llano paired K10 CPU cores with Radeon HD 6000-series graphics on a 32-nanometer process, offering two to four cores at 65 to 100 watts and supporting DirectX 11, OpenGL 4.2, and OpenCL 1.2. The second generation, Trinity, shifted to Piledriver cores and Radeon HD 7000-series GPU on the FM2 socket, arriving in October 2012, followed closely by the Richland refresh in 2013. The third generation, Kaveri, debuted on January 14, 2014, with Steamroller cores and deeper CPU-GPU integration, while low-power variants Kabini and Temash used the Jaguar architecture. A significant rebranding occurred with the arrival of the Zen microarchitecture: AMD now markets these chips as Ryzen with Radeon Graphics or Athlon with Radeon Graphics, using a G suffix on desktop models such as the Ryzen 5 3400G and Athlon 3000G to distinguish them from standard processors. Bristol Ridge preceded Raven Ridge on the DDR4 platform in this new era.

Powering the Living Room: APUs in Game Consoles

One of the most consequential applications of AMD's APU technology extends well beyond the PC desktop and laptop. Beginning with the eighth generation of home video game consoles, AMD designed semi-custom APUs that became the central processing and graphics backbone of both Sony's PlayStation 4 and Microsoft's Xbox One. These chips brought the same fundamental philosophy—CPU and GPU sharing a single die and a unified memory bus—into the living room, enabling the complex rendering and game-logic workloads that define modern console titles. The partnership did not end with that generation. AMD's semi-custom silicon continued into the ninth generation, powering the PlayStation 5 and the Xbox Series X|S consoles. This sustained console presence underscores how the APU architecture, originally conceived for consumer PCs, became a foundational building block for the entire interactive entertainment industry, shaping the hardware landscape of two successive console generations and embedding AMD's design philosophy into the machines millions of gamers rely on daily.

Heterogeneous Computing: The Architectural Vision

At its core, the AMD APU represents a bet on heterogeneous computing—the idea that a single chip should handle both sequential CPU tasks and massively parallel graphics workloads without forcing data across a discrete bus. In practice, the APU places CPU modules, cache, and a discrete-class graphics processor on the same die, all connected by the same bus. This arrangement lets graphics accelerators such as OpenCL run directly on the integrated GPU, and AMD's stated long-term goal is a fully integrated unit whose heterogeneous cores automatically route work to whichever core type is best suited, regardless of whether the task is CPU-bound or GPU-bound. AMD is a founding member of the Heterogeneous System Architecture Foundation, collaborating with other industry members on software and hardware implementations. The memory subsystem reflects this integration: a DDR3 controller arbitrates between coherent and non-coherent requests, partitioning physical memory so the GPU can claim up to 512 MB while the CPU uses the remainder. Additional features like the Unified Video Decoder and Eyefinity multi-monitor support further cement the APU as a self-contained computing platform.