AMD has just released a set of developer tools for Accelerated Processing Units (APUs) ahead of their launch of ‘Berlin’, the first ever server chip based on Heterogeneous System Architecture (HSA). The rollout is expected in the first half of 2014. The software includes ArrayFire 2.0, CodeXL 1.3, clMath, PGI Accelerator Compiler, Project GCC/HSA and Project Sumatra.
Suresh Gopalakrishnan, corporate vice president and general manager of the Server Business Unit at AMD, made the announcement at the APU13 event in San Jose, California, USA. “Servers must be efficient, scalable and adaptable to meet the compute characteristics of new and changing workloads. Software applications that leverage server APUs are designed to drive highly efficient, low-power, dense server solutions optimized for highly parallel and multimedia workloads,” he said.
This year AMD introduced two product families crafted for the data centre; the Opteron 3300, and the Opteron X-Series. These are the world’s first server-class x86 APU’s to be used in HP’s Moonshot software-defined servers.
APUs from AMD, were first launched back in 2011. They act as a CPU and a GPU on a single die, this improves data transfer rates between these components while reducing power consumption. “It’s not like the idea is new, but now we have the technology to put that powerful GPU back on the same piece of silicon as the CPU.” said Margaret Lewis, director of Server ISVs and software planning at AMD.
The company is looking towards a difficult year, as it needs to maintain two separate software environments, one for APU servers and one for its ARM-based servers. AMD will launch ‘Warsaw’ chips, in the first quarter of 2014 which are built to Open Compute specifications and offer better performance per watt than Opteron 6300. Then, in the second half of the year, we should see ‘Seattle’, the company’s first 64-bit ARM System-on-a-Chip.
We should be keeping our eyes out for the ‘Berlin’ series, though. These are a family of cost effective, low power APUs for ultra-dense, fabric-based servers. AMD will aim these at enterprise customers that run Web applications, gaming services and media content. The processors can also be used to provide hosted desktops and could run High Performance Computing applications, such as simulations.
In recent times, AMD has worked with the open source community to develop software tools for these purposes. The developer toolkit for server-side APUs now includes a new version of ArrayFire for OpenCL, this is a fast math library by AccelerEyes that uses clMath for GPU computing and offers an easy-to-use API for Linux and Windows developers. Next up is CodeXL 1.3, AMD’s own suite for Linux and Windows, which features remote profiling and debugging. Project Sumatra is a joint open source project managed by AMD and Oracle which enables developers to code in Java. AMD has also cooperated with SUSE on GCC/HAS, enabling the open source Linux compiler, GCC, to support HSA. Meanwhile the OpenCL math libraries, contributed to open source in August under the name ‘clMath’, enables developers to pick up the pace in common scientific and engineering computations on AMD APUs.