AMD has announced an update to its aging Geode line-up of embedded processors, packing a Bobcat CPU and DirectX 11-capable GPU into a single low-power chip.
Designed for embedded markets that would prefer an x86 chip to a RISC-based ARM CPU, the Geode series has been a successful niche product for the company. However, the new G-Series of APUs promises to trounce its predecessor in all regards.
According to AMD, the new G-Series APUs represent an increase in performance-per-watt terms of three times over the previous generation Geode chips. This performance boost is largely thanks to the GPU portion of the G-Series APU, which supports both DirectCompute and OpenCL.
If you're lucky enough to have supporting software, then this means you can offload parallel computation tasks onto the graphics portion of the hardware, leaving the CPU free to deal with other tasks.
AMD will be releasing five G-Series APUs as a part of the intial product rollout. Let's start with the flagship T56N, which features a pair of 1.6GHz Bobcat cores with 512KB of L2 cache per core, 64KB of L1 cache, Radeon HD 6310-equivalent graphics and a claimed TDP of just 18W. Meanwhile, the T48N drops the speed to 1.4GHz but keeps the same hardware as its bigger brother.
Then we have the low-voltage 1GHz dual-core T40N, which features Radeon HD 6250 graphics, LV-DDR3 memory support and a TDP of just 9W. There are also a couple of single-core chips on offer. These include the 1.5GHz T52R with an 18W TDP, and the bottom-end 1.2GHz T44R, which has an AMD Radeon HD 6250 graphics engine and a claimed TDP of 9W.
Although the chips won't be available via retail, OEMs including Fujitsu, Kontron, Sintrones, Starnet and Wyse - among others - have confirmed that they will be launching G-Series products in the coming months.
Are you pleased to see AMD giving rival embedded systems a run for their money, or are you unconvinced by the company's APU conceptl? Share your thoughts over in the forums