November 01, 2012
When mobile devices become as smart as your PC, will you still need your PC?
What could you do with a 48-core smart phone? If Intel has its way, you won't have to wait long to find out. The device could be a reality in just 5 to 10 years, as the company's research arm preps its 48-core processor for the mobile market.
When you can harness the power of a PC in the palm of your hand, the distinctions between computers and tablets and smart phones start to break down, notes a recent piece at Computerworld.
This opinion is shared by Patrick Moorhead, an analyst with Moor Insights and Strategy, who was interviewed by Computerworld. With a chip like the one Intel is designing, "the phone would be smart enough to not just be a computer but it could be my computer," he stated.
Intel first announced its stamp-sized 48-core processor, the so-called "Single-chip Cloud Computer" (SCC), in 2009. The test chips clocked in under 2.0 GHz, and were suitable for linear algebra, fluid dynamics and Web serving. At the time, though, they were not intended for commercial deployment. Instead, they were to be used as a testbed for multicore experimentation and application porting.
Intel has since changed course and is now pushing the cluster-on-a-chip technology toward the mobile space. The company says we might not have long to wait: the super-smart phones and tablets could be here as early as 2018.
Low-power multicore chips have gained prominence as the Web era's emphasis on processing many small tasks pushes the boundaries of the traditional x86 architecture. Tilera debuted a 100-core chip in 2009, while Intel, AMD, Dell, HP and Penguin Computing are all looking to microservers, a very similar concept, to meet the demands of the cloud and mega-datacenter space.
The current crop of mobile devices already uses multicore chips, but they're of the dual and quad-core varieties. Putting a 48-core processor in a handheld device is a definite game-changer. The additional cores would allow multiple applications to be processed in parallel. A few cores to process email, a few to surf, a few to check email, and so on. Current designs have some capacity for task-sharing, but there's often a noticeable lag, resulting in a less-than-optimal user experience.
Multicore architecture provides an efficient way to handle many small workloads and it offers an attractive compute per watt profile as well. In the proposed Intel model, the cores act like a dynamic computing mesh with just the right amount of cores allocated to a particular app. Such a device could operate in tandem with a remote system (aka "the cloud"), offloading compute-intensive tasks to further optimize on-board resources.
However, there is a potential obstacle. As with the HPC space, application parallelization is not easy and requires the will of the developer community. Enric Herrero, a research scientist at Intel Labs called the lack of suitable software a "limiting factor."
"We need to modify how operating systems and apps are developed, making them far more parallel. Now, [having] cores doesn't matter if I can't take advantage of it," he told Computerworld.
Another analyst interviewed for the piece, Rob Enderle, agreed, saying we still have a ways to go writing for 6- to 8-core machines, and when it comes to getting code to work across massive multicore, "we haven't even really started to do that yet."
Moorhead is more hopeful, though. He believes the software will come when the hardware is ready. When it does, PCs as we know them may become a thing of the past. We'll carry our personal computers with us wherever we go, connecting wirelessly to peripherals like displays, keyboards and mice as needed.
Full story at Computerworld
10/30/2013 | Cray, DDN, Mellanox, NetApp, ScaleMP, Supermicro, Xyratex | Creating data is easy… the challenge is getting it to the right place to make use of it. This paper discusses fresh solutions that can directly increase I/O efficiency, and the applications of these solutions to current, and new technology infrastructures.
10/01/2013 | IBM | A new trend is developing in the HPC space that is also affecting enterprise computing productivity with the arrival of “ultra-dense” hyper-scale servers.
Ken Claffey, SVP and General Manager at Xyratex, presents ClusterStor at the Vendor Showdown at ISC13 in Leipzig, Germany.
Join HPCwire Editor Nicole Hemsoth and Dr. David Bader from Georgia Tech as they take center stage on opening night at Atlanta's first Big Data Kick Off Week, filmed in front of a live audience. Nicole and David look at the evolution of HPC, today's big data challenges, discuss real world solutions, and reveal their predictions. Exactly what does the future holds for HPC?