A Dive into Deep Learning Chip Startup Graphcore’s Software Stack

The last two years have delivered a new wave of deep learning architectures designed specifically for tackling both training and inference sides of neural networks. We have covered many of them extensively, but only a few have seen major investment or acquisition—the most notable of which was Nervana Systems over a year ago.

Among the string of neural network chip startups, Graphcore stood out with its manycore approach to handling both training and inference on the same manycore chip. We described the hardware architecture in detail back in March and while its over $30 million in funding from a …

A Dive into Deep Learning Chip Startup Graphcore’s Software Stack was written by Nicole Hemsoth at The Next Platform.

Liveblog: OpenStack Summit Keynote, Day 2

This is a liveblog of the day 2 keynote of the OpenStack Summit in Boston, MA. (I wasn’t able to liveblog yesterday’s keynote due to a schedule conflict.) It looks as if today’s keynote will have an impressive collection of speakers from a variety of companies, and—judging from the number of laptops on the stage—should feature a number of demos (hopefully all live).

The keynote starts with the typical high-energy video that’s intended to “pump up” the audience, and Mark Collier (COO, OpenStack Foundation) takes the stage promptly at 9am. Collier re-iterates a few statistics from yesterday’s keynote (attendees from 63 countries, for example). Collier shares that he believes that all major challenges humanity is trying to solve counts on computing. “All science is computer science,” according to Collier, which is both great but also represents a huge responsibility. He leads this discussion by pointing out what he believes to be the fundamental role of open source in machine learning and artificial intelligence (ML/AI). Collier also mentions a collection of “composable” open source projects that are leading the way toward a “cloud-native” future. All of these projects are designed in a way to be combined together in a “mix-and-match” Continue reading

Liveblog: AT&T’s Container Strategy and OpenStack’s Role in it

This is a liveblog of the OpenStack Summit session titled “AT&T’s Container Strategy and OpenStack’s Role in it”. The speakers are Kandan Kathirvel and Amit Tank, both from AT&T. I really wanted to sit in on Martin Casado’s presentation next door (happening at the same time), but as much as I love watching/hearing Martin speak, I felt this like presentation might expose me to some new information.

Kathirvel kicks off the session with some quick introductions, then sets the stage for the session. Naturally, Kathirvel starts out by describing AT&T’s cloud deployment. (I say “naturally” because it seems that every presentation starts out with describing how great and how awesome the presenter’s company’s OpenStack cloud is.)

Following the discussion of AT&T’s cloud, Kathirvel launches into a discussion of container trends and demands. He indicates that he believes container usage (or demand?) for enterprise IT applications is huge (and will continue to be large), but doesn’t believe that will hold true for virtual network functions (VNFs) in telco clouds.

As for how containers and OpenStack may be coming together, Kathirvel describes three different use cases:

  1. The first use case has OpenStack managing the infrastructure, with Kubernetes (or another container Continue reading

Liveblog: Kuryr Project Update

This is a liveblog of an OpenStack Summit session providing an update on the Kuryr project. The speakers are Antoni Segura Puimedon and Irena Berezovsky. Kuryr, if you recall, was a project aimed at making OpenStack Neutron functionality available to Docker containers; it has since expanded to also offer Cinder and Manila storage to Docker containers, and has added support for both Docker Swarm and Kubernetes as well.

According to Puimedon, the latest release of Kuryr has a diverse base of contributors, with over 45 active contributors.

So, what will be in the Pike release? For the Kubernetes-specific support:

  • This will be the first release
  • Support for Kubernetes Services (this leverages LBaaS v2)
  • Client- and server-side SSL support
  • RDO packaging

What’s planned for Pike, but may not actually make it? (Again, this is for Kubernetes support.)

  • Token support
  • Resource pools
  • Improved support for Services defined as LoadBalancer type

On the Docker side, the following new features and enhancements will arrive in Pike:

  • Support for Swarm mode
  • IPv4 and IPv6 networking
  • TLS support between Docker and the libnetwork plugin

On the Fuxi side, Kuryr is adding support for Manila shares.

At this point, Berezovsky takes over to discuss the release Continue reading

Liveblog: OpenStack Summit Keynote, Day 2

This is a liveblog of the day 2 keynote of the OpenStack Summit in Boston, MA. (I wasn’t able to liveblog yesterday’s keynote due to a schedule conflict.) It looks as if today’s keynote will have an impressive collection of speakers from a variety of companies, and—judging from the number of laptops on the stage—should feature a number of demos (hopefully all live).

The keynote starts with the typical high-energy video that’s intended to “pump up” the audience, and Mark Collier (COO, OpenStack Foundation) takes the stage promptly at 9am. Collier re-iterates a few statistics from yesterday’s keynote (attendees from 63 countries, for example). Collier shares that he believes that all major challenges humanity is trying to solve counts on computing. “All science is computer science,” according to Collier, which is both great but also represents a huge responsibility. He leads this discussion by pointing out what he believes to be the fundamental role of open source in machine learning and artificial intelligence (ML/AI). Collier also mentions a collection of “composable” open source projects that are leading the way toward a “cloud-native” future. All of these projects are designed in a way to be combined together in a “mix-and-match” Continue reading

Liveblog: Deploying Containerized OpenStack: Challenges & Tools Comparison

This is a liveblog for an OpenStack Summit session on containerized OpenStack and a comparison of the tools used for containerized OpenStack. The speaker is Jaivish Kothari, from NEC Technologies. Two other speakers were listed on the title slide, but were apparently unable to make it to the Summit to present.

Kothari provides a brief overview of the session, then jumps into a discussion of deployment tools. As illustrated by one of his slides, there’s a huge collection of tools that are used to deploy OpenStack; some are “pure” deployment tools, others are configuration management tools. In this presentation, Kothari says he will focus specifically on OpenStack deployment tools, like Juju (Canonical), Fuel (Mirantis), Crowbar (Dell), and PackStack/TripleO (Red Hat), but I’m not sure how this relates to containerized OpenStack (per the session title).

According to Kothari, some of the challenges in “traditional” (non-containerized) deployment tools are best understood by looking at the challenges in deploying OpenStack:

  • Difficulty related to deployment (conflicts due to services configuration, deployment still prone to failures)
  • Ongoing lifecycle management of OpenStack components

This whole first section of the presentation was setting up the argument that containerizing your OpenStack control plane will help address these challenges. Continue reading

DockerCon 2017: all the session videos are now live!

We’re happy to announce that all the breakout session video recordings from DockerCon 2017 are now available online! Special shoutout to all the amazing speakers for making their sessions informative and insightful. All the videos are published on the Docker Youtube channel and the presentation slides available from the Docker Slideshare account.

Here are the links to the playlists of each track:  

Use Case Track

Use case talks are about practical applications of Docker and are heavy on technical detail and implementation advice. Topics covered during this track were related to high availability and parallel usage in the gaming industry, Cloud scale for e-commerce giants, Security compliance and system, protocols legacy in financial and health care institutions.

Black Belt Track

Black Belt talks were deeply technical sessions presented by Docker experts. These sessions are code and demo heavy and light on the slides. From container internals to advanced container orchestration, security and networking, this track is a delight for the container experts in the room.

Docker Deep Dive

This track focuses on the technical details associated with the different components of the Docker platform: advanced orchestration, networking, security, storage, management and plug-ins. The Docker engineering leads walk you through the best way to Continue reading

New Windows build brings more new features

It was just a week ago when Microsoft issued an update to Windows 10 for Insiders to test, and now, four build numbers later, it has another for testing with some new features added.Microsoft is currently full-on in development of “Redstone 3,” the next major update to its operating system. It has said it plans to issue two major updates per year to the OS, with the Creators Update in March being the most recent. The next update is planned for September.To read this article in full or to leave a comment, please click here

Getting a jump on private LTE networks

Vendor partnerships usually don't raise too many eyebrows or pulse rates around here, but when I came across an announcement by Lemko and Federated Wireless promoting a joint effort on the Citizens Broadband Radio Service (CBRS) it did get my attention.CBRS, if you aren't familiar, refers to shared 3.5 GHz spectrum recently opened by the FCC for commercial applications, and everyone from the big carriers to the likes of Cisco and Google are throwing their weight behind it. One possible application would be LTE services, which could come in the form of extended carrier networks, new cable service provider networks, and even private enterprise networks for IoT or other connectivity.To read this article in full or to leave a comment, please click here

FCC hit with DDoS attacks after John Oliver takes on net neutrality

The U.S. Federal Communications Commission's website slowed to a crawl after comic and political commentator John Oliver urged viewers to flood the agency with comments in support of net neutrality, in what appeared to be a repeat of a 2014 incident.With the FCC headed toward a repeal of net neutrality rules it passed in early 2015, Oliver on Sunday echoed his "Last Week Tonight" commentary on the topic from three years ago. (Note to viewers: The link to Oliver's new diatribe is not safe for work.) As in 2014, the FCC's website seemed to buckle under the load late Sunday and early Monday, but the cause may have been more sinister than a flood of people expressing their support for net neutrality rules.To read this article in full or to leave a comment, please click here

FCC hit with DDoS attacks after John Oliver takes on net neutrality

The U.S. Federal Communications Commission's website slowed to a crawl after comic and political commentator John Oliver urged viewers to flood the agency with comments in support of net neutrality, in what appeared to be a repeat of a 2014 incident.With the FCC headed toward a repeal of net neutrality rules it passed in early 2015, Oliver on Sunday echoed his "Last Week Tonight" commentary on the topic from three years ago. (Note to viewers: The link to Oliver's new diatribe is not safe for work.) As in 2014, the FCC's website seemed to buckle under the load late Sunday and early Monday, but the cause may have been more sinister than a flood of people expressing their support for net neutrality rules.To read this article in full or to leave a comment, please click here

Dell takes Intel’s cue on PCs, puts enterprise on top of the agenda

For the umpteenth time, Dell Technologies has reiterated that PCs are important to the company, and it won't quit the market.But PCs occupied only a few minutes of CEO Michael Dell's opening keynote at Dell EMC World in Las Vegas on Monday. PCs are the engine that keep enterprises chugging, he said.Instead, Dell spent time educating attendees about the new Dell Technologies and its products. It's been less than a year since the US$67 billion Dell-EMC merger was finalized, and a lot of focus was on answering burning questions about the company's future.Dell did say the company would offer the PC-as-a-service worldwide by the end of the year, with more details about the program to be shared on Tuesday. HP and Microsoft are offering PC-as-a-service options, allowing customers to buy devices and support and pay on a monthly basis. That option reduces the hardware acquisition and support costs for companies.To read this article in full or to leave a comment, please click here

Dell takes Intel’s cue on PCs, puts enterprise on top of the agenda

For the umpteenth time, Dell Technologies has reiterated that PCs are important to the company, and it won't quit the market.But PCs occupied only a few minutes of CEO Michael Dell's opening keynote at Dell EMC World in Las Vegas on Monday. PCs are the engine that keep enterprises chugging, he said.Instead, Dell spent time educating attendees about the new Dell Technologies and its products. It's been less than a year since the US$67 billion Dell-EMC merger was finalized, and a lot of focus was on answering burning questions about the company's future.Dell did say the company would offer the PC-as-a-service worldwide by the end of the year, with more details about the program to be shared on Tuesday. HP and Microsoft are offering PC-as-a-service options, allowing customers to buy devices and support and pay on a monthly basis. That option reduces the hardware acquisition and support costs for companies.To read this article in full or to leave a comment, please click here

African Regional Internet & Development Dialogue Tackles the World of Education

Dates 8-9 May, Kigali, Rwanda

The first ever African Regional Internet Development Dialogue (RIDD) was launched in Rwanda, Kigali on the 8 of May 2017, placing SDG 4 on Education at the center of the conversation of the first day of the meeting. Delegates had an opportunity to explore how the Internet can provide quick wins for education, but most importantly come up with real solutions that can be implemented immediately.

For Africa a skilled workforce that utilizes ICTs effectively is a key factor in determining its competitiveness in the global digital economy and fully exploiting its potential for sustainable growth. It is the basis for social and economic development, and the foundation of an Internet for everyone.

Dawit Bekele

Core Engine and Windows Updates in Ansible 2.3

Ansible 2.3 Updates

Although the majority of the features added to Ansible 2.3 were networking related, that’s not all folks!

There were several significant changes around module management, the Core engine, and Microsoft Windows support we’d love to show off.

For full details on the release, check out the changelog here.

Module Management

In prior releases, Ansible was organized in two separate module repositories: Ansible-modules-core and Ansible-modules-extras.

The intent was to differentiate the repositories in terms of code quality, feature enablement, and supportability of the modules. We believe we’ve developed a better process.

At the launch of 2.3, Ansible has moved to a metadata-based system for modules. Ansible modules now include an ANSIBLE_METADATA Block which specifies a support category: Core, Curated or Community.

  • Core - supported and maintained by the Ansible engineering team
  • Curated - supported and maintained by the Ansible engineering team and Ansible by Red Hat partners
  • Community - supported and maintained by the Ansible community

In the new system, modules will be following a specific process per category.

Core modules

Modules that the Ansible engineering team directly maintain, and will ship with Ansible. These modules also receive slightly higher priority for pull requests. Any issues that are opened Continue reading

Crunching Machine Learning And Databases Together On GPUs

While it is always best to have the right tool for the job, it is better still if a tool can be used by multiple jobs and therefore have its utilization be higher than it might otherwise be. This is one of the reasons why general purpose, X86-based computing took over the datacenter. Economies of scale trumped the efficiency that can come from limited scope or just leaving legacy applications alone in place on alternate platforms.

The idea of offloading computational tasks from CPUs to GPU accelerators took off in academia a little more than a decade ago, and …

Crunching Machine Learning And Databases Together On GPUs was written by Timothy Prickett Morgan at The Next Platform.

Worth Reading: Be careful what you wish for

After a few minutes of hyperventilating into an empty Dunkin Donuts bag, though, I calmed down enough to appreciate that the guy is trying to provide exactly what I’ve been complaining about the lack of for quite some time now: some kind of logical progression, a path to follow to go from Point BootCamp to Point GotAJob. It’s a helpful roadmap that lays clearly lays out the (many, oh so many) technologies we should be aware of as we work toward professional web dev status. —The New Stack

The post Worth Reading: Be careful what you wish for appeared first on rule 11 reader.

Generalizing a Hardware, Software Platform for Industrial AI

Industrial companies have replaced people with machines, systems analysts with simulations, and now the simulations themselves could be outpaced by machine learning—albeit with a human in the loop, at the beginning at least.

The new holy grail of machine learning and deep learning, as with almost any other emerging technology set, is to mask enough of the complexity to make it broadly applicable without lose the performance and other features that can be retained by taking a low-level approach. If this kind of deep generalization can happen, a new mode of considering how data is used in research and enterprise …

Generalizing a Hardware, Software Platform for Industrial AI was written by Nicole Hemsoth at The Next Platform.

1 … 2,080 2,081 2,082 2,083 2,084 … 3,840