AMD's Kaveri chips bring console-like experience to PCs
Kaveri chips will go into laptops starting at $400
IDG News Service - Thin-and-light laptops have so far had to compromise on graphics to extend battery life, but Advanced Micro Devices wants to change that with its new laptop processors called Kaveri.
Kaveri chips will go into laptops starting at around $400, and be an alternative to Intel's Core processors code-named Haswell, which dominate the market today. With 818 gigaflops of performance on the fastest chip, AMD claims Kaveri can outperform Haswell processors.
That level of performance is akin to bringing a "console experience to PCs," said Kevin Lensing, senior director of mobility solutions at AMD.
The new chips -- available in the A-series and FX lineups -- will have up to four CPU and eight GPU cores, and draw between 17 and 35 watts of power. Laptops will offer up to nine hours of battery life , Lensing said.
Lenovo will be one of the first to offer laptops with Kaveri chips. In the coming months, the company will sell a 15.6-inch Flex 2 starting at $429, 14-inch Flex 2 starting at $799, B50 and G50 laptops starting at $399, and Z-series laptops starting at $579. The chips were officially announced at the Computex trade show in Taipei.
Most buyers prefer Intel chips for laptops, but AMD has implemented a new chip design in Kaveri. Kaveri has a new "Steamroller" processor core and Radeon graphics processors similar to the ones found in the PlayStation 4 and Xbox One.
The new chips replace the older chips code-named Richland, with the CPU getting a speed boost of up to 20 percent. AMD has put a "full GPU inside a chip," not just a watered down version of a graphics core like in predecessor chips, Lensing said.
Beyond better gaming, users will be able to get Kinect-like gesture recognition capabilities in laptops. For example, users can move a webcam around just by waving fingers, something that could not be easily achieved by earlier AMD chips. The CPU and GPU shaders are able to process the images of moving fingers, process them via algorithms, which then results in the webcam moving. This could be done with a basic webcam, Lensing said, adding that it wouldn't require expensive 3D cameras like Intel is recommending for gesture and face recognition on laptops.
But it are some of the software level improvements that will make the performance gains more visible, Lensing said. The chip will provide a better gaming experience with support for Mantle, which is a low-level toolkit to write some of the most popular games.
Kaveri's biggest architectural achievement is the pooling of CPU and GPU memory resources to increase the overall system memory available to all processors. As a result, tasks for CPUs and GPUs could be lined up for execution in a single queue, which reduces bottlenecks. On earlier chips CPU and GPU memory were partitioned -- CPUs would allocate memory for code execution, with information being copied into GPU memory for code execution.
- Mobile First: Securing Information Sprawl Learn how the partnership between Box and MobileIron can help you execute a "mobile first" strategy that manages and secures both mobile apps...
- AIIM Trendscape: The New Mobile Reality This AIIM Trendscape report shares data, expert opinions, and a unique perspective on the impact of cloud and mobility in the enterprise, surfacing...
- Empowering Your Mobile Workers A modern mobile IT strategy is no longer an option, it is an absolute necessity. Here's how some of the nation's most progressive...
- Mobile Content, Collaboration & IDC's 3rd IT Platform: The Next Frontier for the Mobile Enterprise IDC focuses this article on talks about the new IT platform. This 3rd IT Platform will be the new wave for about the...
- On Demand: Mastering the Art of Mobile Content Management Mobile device usage in the enterprise has skyrocketed, and it continues to escalate. IT must answer to users who demand access to their...
- Why do you need an enterprise mobile platform? Today companies must offer great apps that run on a range of devices, and connect to an exploding set of backend data. Appcelerator... All Mobile/Wireless White Papers | Webcasts