APU 101: Driving the Pace of Innovation
By virtually any measure, the arrival of the AMD Accelerated Processing Unit (APU) constitutes a major architectural achievement in computing history—a giant leap forward versus previous processing platforms.
Combining high performance serial and parallel processing cores to enable exhilarating breakthroughs in visual computing, security, performance-per-watt and device form factor—in other words, uniting the CPU and GPU on a single die in order to harness the best of new-age computing—the industry has scarcely seen anything like an Accelerated Processing Unit, and will be forever changed as a result.
APUs in the Context of Heterogeneous Computing
Heterogeneous computing refers to systems that use more than one kind of processor to perform computing functions. These heterogeneous, multi-core systems gain performance not just by adding cores, but also by incorporating specialized processing capabilities to handle particular tasks.
The AMD APU is a heterogeneous architecture- meaning they combine general-purpose x86 CPU cores with programmable vector processing engines – thus bringing together sophisticated CPU scalar processing with the large-scale, parallel based vector processing that is traditionally associated with GPUs.
The appeal of AMD APUs is that they include both scalar and vector hardware as full-fledged processing elements. In AMD APUs, a programmable x86 CPU and the vector processing architecture of a GPU are joined together on a single piece of silicon by a high-performance bus. Both have local access to high-speed memory and on-die cache. AMD APUs also include a variety of other system elements such as memory controllers, I/O controllers, specialized video decoders, display outputs, support for USB 3.0, and high-speed bus interfaces.
AMD designed its APUs to combine CPU and GPU strengths, thus providing software developers with unprecedented flexibility to utilize whichever approach is best suited to the task when they develop new applications.
Innovating a Balanced Solution
Other processor designers have attempted to lash together a CPU and a very basic “integrated” graphics unit in a single package, but the result has been an imbalance in which the GPU lacks the performance and power to fully take on today’s graphics processing challenges—until the AMD APUs, that is.
No other processor company has attempted such a marriage of CPU and GPU using truly programmable, discrete-class GPUs, let alone GPUs that can be programmed using high-level industry-standard tools like OpenCL, Direct-Compute and of course support Microsoft’s complex DirectX® 11 API.
AMD, with its extensive patents and engineering expertise in x86 processor technology and its industry-leading GPU technology, is best situated to deliver on the full potential of combined CPU and GPU strengths.
Arrival of the AMD A-Series APU
With the AMD A-Series APUs, AMD marks a significant technology shift as the company transitions its entire mainstream client computing processor offerings to APUs.
AMD A-Series APUs are designed to enable visually brilliant experiences and high performance gaming and multimedia for end users. Software applications developed for these platforms can use AMD A-Series APUs to improve video quality regardless of the source – delivering extremely sharp HD, and bringing HD clarity and stability to online video. AMD A-Series APUs make it easier than ever to shoot, edit and store HD video, and to quickly search and tag images.
In addition, AMD A-Series APUs include AMD Turbo CORE technology, which varies the performance of the CPU and GPU to respond to the demands of a particular application, and AMD Power Gating, which improves power efficiency by turning on and off CPU and GPU resources based upon application loads.