Google’s machine-learning cloud pipeline explained

When Google first told the world about its Tensor Processing Unit, the strategy behind it seemed clear enough: Speed machine learning at scale by throwing custom hardware at the problem. Use commodity GPUs to train machine-learning models; use custom TPUs to deploy those trained models.

The new generation of Google’s TPUs is designed to handle both of those duties, training and deploying, on the same chip. That new generation is also faster, both on its own and when scaled out with others in what’s called a “TPU pod.”

To read this article in full or to leave a comment, please click here

InfoWorld Cloud Computing

Google’s new TPUs are here to accelerate AI training

Google has made another leap forward in the realm of machine learning hardware. The tech giant has begun deploying the second version of its Tensor Processing Unit, a specialized chip meant to accelerate machine learning applications, company CEO Sundar Pichai announced on Wednesday.

The new Cloud TPU sports several improvements over its predecessor. Most notably, it supports training machine learning algorithms in addition to processing the results from existing models. Each chip can provide 180 teraflops of processing for those tasks. Google is also able to network the chips together in sets of what are called TPU Pods that allow even greater computational gains.

To read this article in full or to leave a comment, please click here

CIO Cloud Computing

Google’s Espresso networking tech takes SD-WAN to internet scale

Google is working to accelerate the performance of its applications over the internet by building out a software-defined network at broad scale. On Tuesday, the company announced Espresso, a system that provides increased network performance to users of the company’s applications.

It works by applying software-defined networking to the edge of the tech titan’s network, where Google connects to the peer networks of other internet service providers. Rather than rely on individual routers to figure out the best way to direct internet traffic, Espresso hands off that responsibility to servers running in the data centers that Google operates at the edge of its network.

To read this article in full or to leave a comment, please click here

Computerworld Cloud Computing

Google’s Espresso networking tech takes SD-WAN to internet scale

Google is working to accelerate the performance of its applications over the internet by building out a software-defined network at broad scale. On Tuesday, the company announced Espresso, a system that provides increased network performance to users of the company’s applications.

It works by applying software-defined networking to the edge of the tech titan’s network, where Google connects to the peer networks of other internet service providers. Rather than rely on individual routers to figure out the best way to direct internet traffic, Espresso hands that responsibility off to servers running in the data centers Google operates at the edge of its network.

To read this article in full or to leave a comment, please click here

Network World Cloud Computing

‘Meet’ Google’s new videoconferencing service for the enterprise

Google appears to have accidentally revealed its new group videoconferencing service for businesses on Tuesday, a week before a big user conference.

The service, called Meet, appears to be its offering for businesses that want to do group meetings over the Internet. According to a saved iOS App Store listing captured by AppAnnie, it will support high-definition video meetings with up to 30 participants. That’s an upgrade over the company’s Hangouts instant messaging and video calling service, which only allows meetings of up to 10 people.

google meet merged Google/AppAnnie

A trio of screenshots shows Google Meet’s functionality on iOS

To read this article in full or to leave a comment, please click here

Computerworld Cloud Computing

Nvidia chief downplays challenge from Google’s AI chip

Nvidia has staked a big chunk of its future on supplying powerful graphics chips used for artificial intelligence, so it wasn’t a great day for the company when Google announced two weeks ago that it had built its own AI chip for use in its data centers.

Google’s Tensor Processing Unit, or TPU, was built specifically for deep learning, a branch of AI through which software trains itself to get better at deciphering the world around it, so it can recognize objects or understand spoken language, for example.

To read this article in full or to leave a comment, please click here

CIO Cloud Computing

The method behind Google’s machine learning madness

First there was TensorFlow, Google’s machine learning framework. Then there was SyntaxNet, a neural network framework Google released to help developers build applications that understand human language. What comes next is anyone’s guess, but one thing is clear: Google is aggressively open-sourcing the smarts behind some of its most promising AI technology.

Despite giving it away for free, however, Google is also apparently betting that “artificial intelligence will be its secret sauce,” as Larry Dignan details. That “sauce” permeates a bevy of newly announced Google products like Google Home, but it’s anything but secret.

To read this article in full or to leave a comment, please click here

InfoWorld Cloud Computing

Google’s CEO just called the next wave in computing, and it’s not VR

Every decade or so, a new era of computing comes along that shapes everything we do. Much of the 90s was about client-server and Windows PCs. By the aughts, the Web had taken over and every advertisement carried a URL. Then came the iPhone, and we’re in the midst of a decade defined by people tapping myopically into tiny screens.

So what comes next, when mobile gives way to something else? Mark Zuckerberg thinks it’s VR. There’s likely to be a lot of that, but there’s a more foundational technology that makes VR possible and permeates other areas besides.

To read this article in full or to leave a comment, please click here

Network World Cloud Computing

10 big announcements from Google’s Cloud Conference

In San Francisco this week at Pier 48, overlooking the Giants’ AT&T Ballpark, Google Cloud Platform (GCP) executives are holding a user conference to introduce products and services they hope will help make the case for choosing Google in the cloud.

Sam Charrington, a cloud and big data analyst and advisor, summed up Google executives’ pitch best this week on Twitter: “GCP exec team’s operating thesis: ‘Cloud’s not done. The industry’s just beginning the journey.”

+MORE AT NETWORK WORLD: Is Google pushing the cloud envelope too far? +

To read this article in full or to leave a comment, please click here

Network World Cloud Computing

10 big announcements from Google’s Cloud Conference

In San Francisco this week at Pier 48, overlooking the Giants’ AT&T Ballpark, Google Cloud Platform (GCP) executives are holding a user conference to introduce products and services they hope will help make the case for choosing Google in the cloud.

Sam Cherrington, an analyst and cloud consultant, summed up Google executives’ pitch best this week on Twitter: “GCP exec team’s operating thesis: ‘Cloud’s not done. The industry’s just beginning the journey.”

Google is seen by many as being behind Amazon Web Services, Microsoft Azure, and even IBM in the IaaS cloud market. In a new research note, Deutsche Bank investment analysts predicted that GCP is on a $ 400 million revenue run rate, which is roughly 20 times less than AWS’s.

To read this article in full or to leave a comment, please click here

InfoWorld Cloud Computing

Google’s new managed containers are brought to you by Red Hat

A new incarnation of Red Hat’s OpenShift Dedicated service for running containers will be available on Google Cloud Platform and could further Google’s plans to create a genuinely open-source hybrid cloud.

OpenShift Dedicated was originally hosted on Amazon EC2, but it’s based on technology that can theoretically allow it to run anywhere. Now that Google and Red Hat are teaming up, instances of OpenShift Dedicated will be available on Google Cloud Platform. There are no details about pricing or availability as yet. 

To read this article in full or to leave a comment, please click here

InfoWorld Cloud Computing

Google’s cloud benchmarking tool ups its game

Google’s PerfKit toolset for benchmarking cloud environments was originally released earlier this year in a pre-1.0 version. Today, it’s officially been bumped to a 1.0 release, with expanded support for various cloud providers and automation of 26 different benchmarks, up from the 20 originally provided.

Given how tough it can be to reliably benchmark any cloud, having an open source, cloud-agnostic toolkit to help make it happen is a net boon.

To read this article in full or to leave a comment, please click here

InfoWorld Cloud Computing

Google’s infrastructure czar predicts cloud business will outpace ads in 5 years

Google may lag behind its two biggest rivals in the public cloud, but Urs Hölzle, the technology titan’s senior vice president for technical infrastructure, sees Google’s cloud revenue eclipsing its ad business in five years.

“I think cloud will actually turn out to be a huge business because it’s a service business,” he said during an interview on stage at the Structure conference in San Francisco on Wednesday. 

Asked what he would say to enterprises that are skeptical about using Google’s cloud, Hölzle pointed to the company’s long track record providing enterprise services through its Apps productivity suite, and through the Google Search Appliance, which it’s been selling to enterprises since 2002. 

To read this article in full or to leave a comment, please click here

Network World Cloud Computing

Google’s Android Based Brillo Has the Potential to Take IoT Automation to Next Level

brillo1 300x155 Google’s Android Based Brillo Has the Potential to Take IoT Automation to Next LevelWith the acquisition of Nest last year, Google has demonstrated its interest in the field of smart home. At recently concluded Google I/O annual developer conference, the group of Mountain View celebrates a further step forward, talking openly about the Internet of Things.

Born Brillo, a project to connect any device used, not only smartphones, tablets, computers and smartwatch, but also those that are part of everyday life such as home appliances, cars, surveillance systems etc.

Brillo is the ecosystem through which Google intends to play a leading role in the IoT. It is a platform derived from Android, and reduced to essentials to be performed on devices with minimum system requirements, therefore, suitable to be fitted for example in lamps for smart intelligently manage the lighting system of the house. The strength of Brillo is the ability to recognize these devices in an entirely automatic way in smartphones and tablets, as well as simplify the configuration process, making it accessible even to beginners.

It will be able to connect devices of all kinds, through the use of sensors from the extremely low power consumption, enabling them to communicate with each other and enabling users to interact with it such as centralized refrigerators, equipment for monitoring of home, lighting and much more talking to each other.

In addition to home automation, Brillo is also designed for industrial use. Thus, a plant could, for example, use it to connect its sensors and manufacturing equipment.

Google’s another project Weave will be used as the cross-platform protocol, based on JSON (JavaScript Object Notation), through which developers can put in communication between their devices and objects compatible with Brillo, thereby taking advantage of the enormous potential of synchronization of cloud platforms and Mobile application versatility.

As regards the technical specifications, it seems that the software developed by Google can run on devices with a small quantity of RAM, even if only 32 or 64 MB. It supports Wi-Fi connectivity and Bluetooth low energy, does not require particularly powerful processors to run and the Thread protocol used by equipment designed by Nest, a Google property company specializing in intelligent thermal control systems.

Google Brillo IoT is based on a kernel that is derivative of the Android system; naturally it compact the bone to be unified with devices of very small size and devices not too capable on the hardware side. Given the market share of Android and the open source nature, Brillo has the potential to reach the same level as Android. The choice of keeping popular Android mobile OS caters especially to the simplification of procedures developed by device manufacturers.

One thing is sure – one linked to the Internet of Things is a new territory, but which have already staked their eyes for all big technology industries. Microsoft recently announced the arrival of a specially developed IoT version of the Window 10 operating system. Huawei has presented an IoT platform called LiteOS weighing just 10 kB and Samsung has already launched the chip design intended specifically for this sector.

The IoT will come soon in our lives every day without making too much noise with a number of interconnected devices that will grow dramatically in the coming years, and it is obvious that all the big names are getting ready to new market requirements.


CloudTimes