HSA AMD: AMD’s answer to Intel’s dominance
Only yesterday Intel and Nvidia outlining the benefits of their products focused on massively parallel computing GPU accelerated (GPGPU), the first with Knights Corner and MIC micro-architecture , the second with Tesla and its new API OpenACC . Apparently amd does not intend to stay out of the game and from HardwareZone bring us something about your bet on the future: HSA.
HSA stands for Heterogeneous System Architecture , and is the weapon with which AMD plans to build a whole new ecosystem of hardware and applications based on it. But so far only mentions what AMD achieved HSA, but not as we do, and according to AMD is that HSA is an open standard that involves a set of hardware and software that integrate CPU (not only AMD CPUs) and GPU together to a new high-level language, which provide greater ease of programming with features such as unified access to system memory (HSA Managenment Memory Unit) and its address space, with complete consistency between CPU and GPU, providing an environment that combines sequential programming with parallel programming.
HSAs allow future CPUs, APUs and GPUs offer corporate users and the performance benefits that are able to offer together, and it plans to create a complete open ecosystem with many members, similar to what we have ARM and MIPS, and a certain way it was in its infancy with many manufacturers x86 that contributed to its growth, until Intel began to expel one by one with their practices known to all.
AMD claims that because of its approach to HSA, you can not afford to neglect any aspect of your current graphics products (APUs and GPUs) and their performance in games or GPGPU computing, which they consider a very important point.
Yet little is known about HSA, which will at some point in 2014. AMD will provide more details during your event HSA AMD Fusion Developer Summit , which will be held from 11 to 14 next month.
– AMD’s HSA – A Threat to Intel’s Dominance and Mobile Ambitions? (HardwareZone)
– Answered by the Experts: Heterogeneous Compute and GPU with AMD’s Manju Hegde (Anandtech)