October 03, 2012Francisco Cheng
As the mobile processor has become the center of digital life for most, the compute demands on them have skyrocketed. Multicore CPUs, strong DSPs and large, parallel GPUs, have become the norm for mobile systems. But the available power has stayed relatively the same. The need for more performance in a fixed power envelope has increased the need to maximize all the capabilities of the platform. Applications have to utilize all the units available and move the workloads to the best processors available.
In a typical system, the CPU and GPU work together to tackle some of the most sophisticated tasks in the world, though the processes aren’t as distributed as some chipmakers would like them to be. In particular, the recent rise in mobile resolutions has increased the need for GPU performance, which has caused dramatic growth in GPU size. Meanwhile, mobile CPUs have seen more traditional growth. But GPUs should no longer be considered good for only running visual applications. Today they are strong parallel processors and can be used to run parallel computing tasks.
In other words, the GPU should not only be tasked for visuals and gaming, but also for general tasks like launching apps and crunching Excel spreadsheets—that’s the whole concept behind “parallel computing”. And if the HSA (Heterogeneous Systems Architecture) Foundation has its way, parallel computing will not only provide a surge in computational capacity for all apps, but it will also help to establish standards for the platform. And these standards, supported by many vendors, can be used by operating systems and applications in a platform agnostic way.
Qualcomm, a leading technology company, is driving this initiative and establishing its place as a Founder member of the HSA Foundation. It’s an ideal marriage, as the Snapdragon family of chipsets is already a heterogeneous environment, utilizing different IP blocks (CPU, GPU, DSP, video and ISP) to deliver the best possible app experience. As a company, Qualcomm is always investing heavily in new CPU, GPU, memory and systems architectures for Snapdragon chipsets. Because of this rich history and experience in distributed processing, Qualcomm can help the HSA foundation achieve an open and widely adopted standard for heterogeneous computing.
Chips based on heterogeneous system architecture will be a lot more than just a CPU slapped next to a GPU; the standard will enable an entire ecosystem, where workloads will run seamlessly on different processing cores in the platform, all in parallel. The important thing to remember is that this is an open platform to developers, agnostic of OS and form factor, unlike other standards. The goal is to have an amazing hardware platform and an entire ecosystem of software and applications to take advantage of it. What this means to the end user is that not only will mobile apps run even smoother and faster, but so will the entire operating systems. All the components within the mobile processor are working in concert to achieve this while using the least amount of power possible.
October 03, 20120