First Look: AMD's New A-Series Fusion Processor Review

If you consider yourself to be a well-informed PC buyer then you probably know about AMD's new FUSION technology. FUSION is what AMD calls it's new Accelerated Processing Unit (APU) technology. The first FUSION processors were the part of the Brazos platform of C-series and E-series processors (code named "Zacate") for the entry-level laptop market. We had a first look of the E-350 processor last year and were eager to see how it would perform in the real world once new AMD-equipped laptops started to appear at CES 2011.

The E-350 APU has already found a home in a number of netbook alternative laptops like the HP Pavilion dm1z and the Lenovo ThinkPad X120e. Those AMD-powered compact laptops delivered video and gaming performance that was radically superior to Intel-based netbooks in the same price range. The big question was, "What does AMD have planned for mid-range and high-end notebooks?"

The answer is the Sabine platform ... formerly known by it's code name, "Llano".

AMD's A-Team
The Sabine platform consists of the A-series APUs (A4, A6, and A8) and will come in a wide variety of dual-core and quad-core options. These new notebook processors feature thermal design packages (TDPs) of 35 and 45 watts ... meaning they are intended for mid-range and high-performance notebooks unlike the C-series and E-series APUs which had TDPs of 9 and 18 watts respectively. Desktop versions of the A-series APUs will soon follow with 65- and 100-watt TDPs, but that's not what we're here to talk about today.

The base clock speeds for the A-series notebook APUs range from 1.4GHz to 2.1GHz and higher clock speeds are reached via AMD's "Turbo Core" technology. Turbo Core works in much the same way as Intel's TurboBoost technology by up-clocking one or more cores depending on the processor demands of the software you're running. Unlike TurboBoost, AMD's Turbo Core bases its clock speed increases on the workload of the APU rather than thermal measurements of processor activity.

Without getting too technical, the new A-series APUs are made using a 32nm manufacturing process and each processor core on the "CPU side" of the APU has 1MB of its own Level 2 cache. More important than the CPU cores, however, is the fact that half the chip die is now taken up by the graphics-processing circuitry. As you can see in the schematic below, that is a lot of room to set aside for graphics.

The A4, A6, and A8 processors each have their own "models" of mobile Radeon graphics which AMD claims to be "discrete-class graphics” at their respective price ranges. As you can imagine, we were eager to put that particular claim to the test since we had mixed feelings about Intel's claim that its new integrated graphics inside the "Sandy Bridge" second-generation Core-series processors also deliver "discrete-class graphics" performance.

One fact that supports AMD's claim that the A-series APUs deliver true "discrete-class graphics" is the fact that all the A-series chips natively support DirectX 11 games ... you can't do that with Intel integrated graphics. Yes, most Intel-based notebooks intended for multimedia or even light gaming will feature a dedicated graphics card from Nvidia or AMD, so the argument that Intel integrated graphics don't support DirectX 11 might sound weak. However, you're missing the point that notebook manufacturers can now build an AMD-based laptop with only a single APU chip that can play DirectX 11 games and you don't have to spend that extra money on a discrete graphics card if you're just a casual gamer.

As if that wasn't good enough, AMD's latest Hybrid Crossfire technology allows the GPU portion of the APU to work with a second discrete graphics card (if your notebook is equipped with dual graphics) and provides an additive performance boost over the discrete graphics alone. In other words, Intel notebooks with dual graphics have the option of using integrated graphics or discrete graphics. AMD dual graphics now has the option of integrated, discrete, or integrated and discrete working together for even greater performance.

I Pity The Fool Who Doesn't Buy Fast Memory
Warning: minor techno-geek tangent ahead. One seemingly insignificant technical detail that I love about the new A-series APU chips is the way the graphics cores make use of system memory. For the sake of general audiences, the graphics side of the APU is first in line to access the system memory. AMD did this because graphics processors prefer the fastest RAM while traditional CPUs aren't so demanding about fast RAM. The A-series chips will work with DDR3 memory at speeds up to 1600MHz.

This is an important little bit of information because most notebook manufacturers will only include 1333MHz RAM inside their laptops. Since graphics processors can take advantage of faster memory, this means that you will see a noticeable performance boost simply by upgrading your notebook with faster RAM!

Now, let's take a closer look at the "white box" test laptop that AMD provided.