TAIPEI: AMD LAUNCHED its Kaveri A-series APU chip for "ultra thin and high-performance mobile PCs" at Computex on Wednesday, less than 24 hours after Intel announced its Core M chip aimed at convertible 2-in-1 devices.
Touted as the firm's third architectural generation of performance APUs, the new Kaveri mobile chip will power mobile PCs and notebooks, and claims to be "the most advanced APU ever created".
Those who are wondering why AMD has launched its Kaveri chip again should note that this is the mobile version of the desktop Kaveri chip that the firm launched at CES in Las Vegas earlier this year, and not the same thing, although AMD did say that its latest mobile APU shares some physical attributes with the desktop version.
For instance, the mobile version of Kaveri also features AMD's heterogeneous system architecture (HSA), which refers to AMD's unified memory architecture that the firm said allows more workloads to be offloaded to the GPU by mitigating the latency bottleneck caused by repeatedly having to fetch data from system memory.
AMD mobility product line director Kevin Lensing told The INQUIRER in a press briefing ahead of the Kaveri launch today that HSA is an important addition in the APU for AMD because it makes the GPU just as easy and efficient to code to as the CPU.
One of the key aspects of AMD's heterogeneous system architecture is the ability for developers to deploy code that runs on the CPU's on-die GPU and access main memory. Being able to cut down on the number of memory fetches means AMD can make it viable to run more workloads on the GPU.
As with the desktop version, the mobile Kaveri APU also features AMD's Graphics Core Next (GCN) architecture, which enables gamers to experience high-performance video games with Mantle, a tool for alleviating CPU bottlenecks such as API overhead and inefficient multi-threading. Mantle, which is basically AMD's answer to Microsoft's Directx, enables improvements in graphics processing performance. AMD claims that Kaveri teamed with Mantle enables it to offer built-in Radeon dual graphics to provide performance boosts ranging from 49 percent to 108 percent.
"For the first time in our performance APUs we have GCN cores in this device but not only do we have GCN cores we have the most graphics cores we have ever put in an APU," said Lensing.
"If you look back to Trinity and then to Richland we had six graphics CUs but Kaveri scales up to eight graphics CUs, that's why you get this unprecedented amount of single precision floating point operations of 818 [GFLOPS with Kaveri] even at mobile TDP.
"You get this from the combination of the CPU cores and the largest ever set of graphics cores."
The Kaveri mobile APU architecture features up to 12 compute cores - four CPU plus eight GPU - to deliver better performance and responsiveness on devices across various workloads and applications. It also supports Ultra HD (UHD) 4K resolutions and new video post-processing enhancements that can upscale HD 1080p videos to 4K quality on UHD-enabled monitors and TVs.
The Kaveri APU family also features AMD's first FX branded "enthusiast class" APUs which are aimed at bringing better performance for gaming.
The announcement from AMD arrives less than a day after chip rival Intel announced its Core M 14nm fanless mobile chip family at Computex, which will power upcoming slimline laptops, tablets and in particular 2-in-1 convertible devices such as the Transformer Book Chi that Asus announced on Monday.
Intel said Core M devices will offer 20 to 40 percent performance improvements, while system on chip (SoC) power could be cut by up to 45 percent, producing 60 percent less heat, which allows for fanless design.
At the announcement in Taipei today, AMD's Lensing pitted its Kaveri FX chip against Intel's Haswell Core i7 CPU, as you will see in the video below.
It was exactly a year ago that AMD showed off the Kaveri chip for the first time at Computex 2013, during the launch of its third-generation Richland desktop processor.
As part of the announcement on Wednesday, AMD also revealed an x86 embedded G Series SoC and CPU for embedded applications for thin client deployment in various industries such as healthcare and retail. µ
Sign up for INQbot – a weekly roundup of the best from the INQ