2018 Will Be the Year of the GPU-Accelerated Everything
Andrew Wheeler posted on January 19, 2018 | 4050 views

To make predictions for the changes made to the hardware covered regularly in this section for engineers working in various fields and disciplines, I spoke with industry experts from Lenovo and NVIDIA and got some information from HP, who generally refrains from making predictions. 

Digital Fabrication, Reality Capture and Virtual Reality

On the topic of what 2018 will look like in architecture, engineering and construction industries as well as product development, Cassidy Lammers, worldwide workstation manager for Lenovo, showed me how Lenovo is honing its focus on several topics for the year and sees a few trends continuing to emerge and develop.

Lenovo sees reality capture (i.e. photogrammetry and laser scanning) gaining more acceptance for documenting spaces and capturing as-built information on jobsites, whether it’s a new factory or a construction site. The importance of integrating captured 3D models into projects using powerful CAD software will continue.

When it comes to digital fabrication, Lenovo sees that advances will be made in robotic printing and free-form construction and is keeping its eyes on the space, underscoring its investment in MX3D.

Engineers at MX3D partner with Lenovo and use ThinkStation P910s to run topology optimization applications like Fusion 360, and MX3D designers use the ThinkPad P40 Yoga, which sketches like a tablet but is powerful like a workstation. (Video courtesy of Lenovo.)

Lenovo also sees that immersive experiences like virtual reality (VR), mixed reality (MR) and augmented reality (AR) will be used more for visualization and simulation to reduce project errors, open communication and influence design for the better.

Cloud Computing, Virtualization and IoT

Anne Hecht, senior director product marketing for NVIDIA GRID, submitted four predictions for 2018 about the future of the digital workspace. Of course, graphics processing unit (GPU) computing is central to her predictions and the adoption of GPUs by enterprise entities will continue to grow in 2018.

Pictured here is the NVIDIA Jetson TK1, which is built around NVIDIA Tegra K1 SoC and uses the same NVIDIA Kepler computing core, which is currently baked into many supercomputers around the world. (Image courtesy of NVIDIA.)
Pictured here is the NVIDIA Jetson TK1, which is built around NVIDIA Tegra K1 SoC and uses the same NVIDIA Kepler computing core, which is currently baked into many supercomputers around the world. (Image courtesy of NVIDIA.)

Over the past five years, enterprise applications that require computer graphics have doubled. According to Hecht, over 60 percent of enterprise employees now rely on at least one app that mandates accelerated graphics.

As Hecht says of NVIDIA’s outlook, “We predict that every application developed and deployed in 2018 will require a GPU for maximum performance. Consequently, we anticipate that by 2020 nearly all enterprise employees will rely on an application that requires computer graphics to do their jobs. This will make virtualized GPUs a requirement for every enterprise cloud and virtualized data center.”

The GPU-accelerated cloud is important to consider when accounting for resources and scalability of graphics and computing workloads. In 2018, NVIDIA sees the GPU-accelerated cloud coming down to benefit small- and medium-sized businesses, as well as individuals like creatives, technical professionals and knowledge workers.

It doesn’t matter whether the application is accessed from a data center or the cloud, it’ll be GPU accelerated. And users will benefit from a seamless experience across multiple endpoints, no matter where the app or data is stored. This will accelerate organizations’ move from a single end-user compute solution to the adoption of a portfolio of solutions, whether on-premises, a hybrid or completely in the cloud.

VDI is a type of virtualization technology that hosts a desktop operating system on a centralized server in a data center. Innovative approaches and new tools will give enterprises the ability to measure user experience, which is an oft-noted factor in successful VDI projects. NVIDIA’s GPU SDK will help companies (who know what users want and need) create user experiences that best fit their users.

Data centers will continue in the global march towards 24/7 automation. Workloads won’t be sticky and, with hardware moving away from being special purpose, will allow enterprises to attain the highest levels of infrastructure utilization. For example, they could implement virtual desktop infrastructure (VDI) during the day and high performance computing (HPC) at night.

NVIDIA also predicts that the siloed lines of IT will dissipate as new technologies cross the lines between different teams and functions. Synchronies and intersections will develop between IT departments, especially between HPC and end-user computer administrators, to build data centers that are more collaborative and convergent.

This echoes the thoughts of Lenovo’s Cassidy Lammers, who sees collaboration becoming more virtual and global expertise becoming more distributed. Lammers also predicts advances in the Internet of Things (IoT), where it sees huge new sets of data becoming available through connected products and hardware that will allow manufacturers to capture, analyze, control and manage improvements to various products.

Artificial Intelligence Will Pervade Everything in 2018

Artificial intelligence (AI) will creep into every industry through 2018, and Lenovo is keeping its eyes focused on machine learning within manufacturing and ray tracing and rendering within immersive experiences. Another facet of AI Lenovo sees growing is generative design. The ability for individuals and teams to meet predetermined goals and with predetermined constraints will benefit designers and engineers more in major industries like automotive and aerospace.

With surprising wins by AlphaGo’s deep learning over top human players like Lee Sedol at a more intuitive game of Go and strangely sealed warnings from Elon Musk and Stephen Hawking, artificial is set for a landmark year. (Image courtesy of YouTube.)
With surprising wins by AlphaGo’s deep learning over top human players like Lee Sedol at a more intuitive game of Go and strangely sealed warnings from Elon Musk and Stephen Hawking, artificial is set for a landmark year. (Image courtesy of YouTube.)

If you want to follow the advance of machine-learning and deep-learning AI in 2018, which is the most exciting development in computing hardware, follow NVIDIA and its partners. NVIDIA has partnered with FANUC to change the way robots manage other robots who are manufacturing robots, which has perhaps cost as many as 5,000 people their jobs, according to an anonymous source.

There is so much going on in the computing hardware space around the world, and it touches pretty much every industry you can think of, and nearly every facet of day-to-day life. Computing hardware has evolved to a point where we do not even think about its ubiquity in a separate sense from ourselves.

The debate will rage between those who believe GPU-accelerated AI will converge with a host of converging and advancing technologies, including the Industrial Internet of Things (IIot) and robotics, and begin to replace more workers, and those who believe that racing towards automation will increase jobs and create new fields for people to explore.

The Importance of Recycling Computing Devices Will Increase

There are 7 billion people in the world. Approximately 2 billion of them are in the position where they use computer electronics in some way daily. The most popular computing device of all time is the smartphone. There are landfills packed with computer electronics because companies try to turn as much of a profit as possible as a general priority and they make the most of planned obsolescence to create stable growth and profit margins.

Smartphones waiting to be processed at Sims Recycling Solutions’ facility in Roseville, Calif. (Image courtesy of Jim Merithew/Cult of Mac.)
Smartphones waiting to be processed at Sims Recycling Solutions’ facility in Roseville, Calif. (Image courtesy of Jim Merithew/Cult of Mac.)

Companies like Apple and Samsung employ long-term supply chain strategies that center around encouraging customers to buy new versions of the product they already have with the promise of more interesting new features.


Cybersecurity Will Have to Do Better in 2018

Mix these sales strategies in with the fact that you produce the data for companies offering free software service for free (i.e. email software such as Gmail): They collect the data you provide, sell it to advertisers who are looking for a set of demographics to target with customized ads and monetize the digital record of your behavior—anonymized or not. 

The fact is, there is a huge amount of digital data out there, and we don’t even question that ours is among the billions of others out there. We have a digital world packed with huge data sets of people’s personal digital information, and we are accepting the likelihood of it getting into the wrong hands.

For consumers in 2017, the year ended with news that the hardware they own makes their personal information vulnerable. The cybersecurity gatekeepers of enormous amounts of personal data looked bad throughout the year: Equifax compromised the personal financial information of hundreds of users and admitted it months after the fact. Yahoo lost an even higher quantity of less crucial data, but, again, hundreds of millions of compromised personal accounts were taken by hackers. And the year ended with the Spectre and Meltdown exploits making Intel, AMD, ARM, Apple, Google and especially Microsoft look like a strange mixture of foolish and vigilant.

HP doesn’t really engage in the business of making predictions, but it did mention that the Z Desktop Workstations (HP Z8, Z6 and Z4) are its primary focus when it comes to hardware we cover in this section.

HP Z Workstations like the Z8 pictured here include two Intel Skylake-SP Xeon CPUs (56 cores) up to 3 TB of DDR4 RAM and up to nine PCIe slots with terabytes of storage as well. (Image courtesy of HP.)
HP Z Workstations like the Z8 pictured here include two Intel Skylake-SP Xeon CPUs (56 cores) up to 3 TB of DDR4 RAM and up to nine PCIe slots with terabytes of storage as well. (Image courtesy of HP.)

HP is wisely placing a big emphasis on security, and its HP Z Desktop Workstations have features likeHP SureStart, which is purportedly the industry’s first self-healing BIOS with intrusion detection. The HP Security Suite, in combination with HP SureStart, is supposed to ensure that the IP created on its workstations is as secure as possible.

Recommended For You