Apa yang berikutnya Dari Fusion AMD?

What’s next for AMD Fusion?

What a year it has been already for AMD and its APU products – we have now announced top-to-bottom families of processors that support everything from low-power tablets to performance notebooks and desktops. All of which integrate DirectX™ 11™-capable graphics with new “Bobcat” or 32nm “Stars” x86 CPU cores. Bringing the GPU and the CPU together on a single chip was a critical step for AMD, and the resulting processors are finding a welcome home with OEMs and end-users. We are just getting started.

At this week’s AMD Fusion Development Summit in Bellevue, Washington, I spoke to more than 600 attendees about where we plan to take the industry next with the AMD Fusion System Architecture (FSA). The audience was primarily software developers, recognizing that they are critical to our success and that we want their participation during the development of the platform. I have been developing 3D graphics and parallel computation software for more than 20 years, so I understand why total platform design is required to fully enable a programmer’s creativity and productivity. 

In steering the architectural direction of FSA in my role as Corporate Fellow, my primary concern has been how to make heterogeneous (i.e., APU) programming easier, more natural and accessible to the largest possible community of software developers.

So what does that mean, really? We aim to make the unprecedented parallel processing capability of the GPU on the APU as accessible to programmers as the CPU is today. To do that there are a series of simplifying steps we plan to take that will improve on what is already a great foundation:

Add support for C++ features that more fully leverage the GPU as a parallel processor

Unify the memory address space shared by the CPU and GPU, and make it coherent, so they operate seamlessly together.

Add user mode scheduling, to dramatically reduce the time it takes for the CPU and GPU to dispatch work to each other. There are others, but these are big ones, resulting in the biggest leap forward. Once the AMD Fusion System Architecture is realized, the GPU is a true peer processor to the CPU, with direct access by software. In the meantime, the benefits of the integration step are readily apparent: dramatic improvements in battery life for AMD platforms; smaller form factors through reduction in the silicon footprint; acceleration of applications that leverage OpenCL and DirectCompute via the GPU, just to name a few.