Welcome to Our Department
Published:
If you are beginning Computer Engineering in 2017, welcome to a field full of attractive buzzwords. Read more
Published:
If you are beginning Computer Engineering in 2017, welcome to a field full of attractive buzzwords. Read more
Published:
Apple announced the iPhone X (pronounced “ten”) on September 12, 2017 at the Steve Jobs Theater — the first event at the new Apple Park campus in Cupertino — pricing it at $999 for 64 GB and $1,149 for 256 GB. The iPhone X replaced the physical Home button (present on every iPhone since 2007) entirely, enabling a 5.8-inch OLED Super Retina display (2436×1125, 458 ppi) that covered almost the entire front face of the device with a notch at the top for the TrueDepth camera system. The TrueDepth camera array comprised seven components: a dot projector (projecting 30,000 invisible infrared dots in a structured pattern onto the user’s face), an infrared camera (capturing the distortion of the dot pattern to build a precise 3D depth map), a flood illuminator (illuminating the face with IR light to enable recognition in the dark), a 7MP visible-light front camera, an ambient light sensor, a proximity sensor, and a speaker. The system worked in darkness where no 2D camera could function, and Apple quoted a false-acceptance rate of 1 in 1,000,000 for Face ID versus 1 in 50,000 for Touch ID — a 20× improvement in the probability of a random person’s face unlocking the phone. Read more
Published:
In 2017, choosing Computer Engineering means entering a field that is expanding far beyond traditional software development. Read more
Published:
AMD launched Ryzen Threadripper on August 10, 2017, with the Threadripper 1950X (16 cores, 32 threads) at $999 and the Threadripper 1920X (12 cores, 24 threads) at $799. Both used AMD’s TR4 socket, a physically massive 4094-pin LGA package on the X399 chipset. The 1950X was priced directly against Intel’s new Core i9-7900X (10 cores, $999 launched two months earlier) and offered 60% more cores for the same price — a competitive statement rather than a niche positioning. Read more
Published:
Learning software development can easily become overwhelming because there is always another topic to study. A programming language leads to a framework. The framework leads to databases, networking, testing, operating systems, design, and security. Each topic reveals three more. Read more
Published:
Intel launched Skylake-X in June 2017 for the LGA 2066 socket (Basin Falls platform), replacing Broadwell-E. The lineup opened with the 10-core Core i9-7900X at $999 and the 8-core i7-7820X at $599, then extended in August to 12-, 14-, 16-, and 18-core models topped by the i9-7980XE at $1,999. All chips supported quad-channel DDR4, 44 PCIe 3.0 lanes (on the top SKUs), and Intel’s X299 chipset. Core i9 was itself a new brand tier Intel created with this launch because i7 no longer distinguished the highest-end parts. Read more
Published:
If I were building a gaming PC on July 1, 2017, this is the machine I would actually want. I am not simply choosing the most expensive part in every category. The priority is gaming performance, then reliability, cooling and enough headroom to keep the system useful for years. Read more
Published:
Apple introduced Core ML at WWDC on June 5, 2017, shipping with iOS 11 (September 19, 2017), macOS High Sierra (September 25, 2017), watchOS 4, and tvOS 11. Core ML provided a unified on-device inference framework: developers converted pre-trained models from common training frameworks (Keras, Caffe 1, scikit-learn, XGBoost, libsvm) into Apple’s .mlmodel format using the coremltools Python package, then added the .mlmodel file to an Xcode project where Xcode automatically generated a Swift/Objective-C class wrapping the model’s inputs and outputs. At runtime, Core ML automatically routed computation to the most appropriate hardware: the CPU (using BNNS — Basic Neural Network Subroutines — a low-level optimized library), GPU (via Metal Performance Shaders, for models with GPU-friendly layer types), or the Neural Engine on devices that had one. The A11 Bionic (iPhone X, September 2017) shipped the first dedicated Neural Engine — a 2-core block running at approximately 600 GOPS, used exclusively for Core ML inference and not exposed to general Metal computation — enabling face recognition for Face ID and Animoji at low power. The A12 Bionic (iPhone XS, September 2018) expanded this to an 8-core Neural Engine capable of 5 TOPS, an 8× improvement, and established the Neural Engine as a permanent dedicated accelerator on all subsequent Apple SoCs. Read more
Published:
Google announced first-class Kotlin support for Android at Google I/O on May 17, 2017, endorsing Kotlin as an officially supported language for Android development alongside Java. Kotlin was created by JetBrains (the company behind IntelliJ IDEA, on which Android Studio was based) and had reached version 1.0 in February 2016 after years of development that started in 2010. The announcement included immediate support in Android Studio 3.0 (released October 2017) — Kotlin tooling was built into the IDE at the same level as Java, with full debugging support, code completion, refactoring, and a Java-to-Kotlin converter that could automatically translate existing Java files. Kotlin compiled to JVM bytecode (like Java), meaning Kotlin and Java source files in the same project compiled to interchangeable .class files, and Kotlin code could directly call Java APIs and vice versa — including the entire Android SDK (android.app.Activity, android.widget., android.content.), all Jetpack libraries, and any Java library in Maven Central. Read more
Published:
Java has been the main language of Android development since the first Android SDK in 2008. It has worked, but it has also accumulated frustrations that Android developers have complained about for years. Kotlin — designed by JetBrains and reaching version 1.0 in February 2016 — has been gaining serious traction as an alternative, and the community momentum suggests this is no longer a niche experiment. Read more
Published:
Docker announced the Moby Project at DockerCon 2017 in Austin, Texas on April 18, 2017, as a structural reorganization of Docker’s open-source components. Solomon Hykes, Docker’s founder and CTO, introduced Moby as an open framework of modular components from which container systems — including Docker itself — could be assembled. Moby encompassed the building blocks Docker had developed over four years: containerd (container runtime, donated by Docker to the CNCF in March 2017), runc (the OCI-compliant low-level runtime, originally submitted as the reference implementation when Docker co-founded the Open Container Initiative in June 2015), LinuxKit (a toolkit for building minimal, immutable Linux distributions optimized for container workloads), BuildKit (a next-generation image build backend with parallel layer execution and improved caching), SwarmKit (orchestration primitives), and Notary (content trust and signing). Docker simultaneously rebranded its products: what had been “Docker” (free) became “Docker CE” (Community Edition), and “Docker CS” (commercially supported) became “Docker EE” (Enterprise Edition), clarifying the commercial/open-source boundary. The docker/docker GitHub repository was renamed to moby/moby, which caused immediate confusion among the large number of contributors who had contributed to that repository expecting to contribute to the Docker product. Read more
Published:
The word “serverless” is a little funny because the servers do not actually disappear. They are still there — running in AWS data centers in Virginia or Oregon or wherever the platform routes your traffic. The servers become less visible to the developer, which is different from not existing. Read more
Published:
AMD launched Ryzen on March 2, 2017, with three models: Ryzen 7 1800X (8 cores/16 threads, 3.6 GHz base/4.0 GHz boost, $499), Ryzen 7 1700X ($399), and Ryzen 7 1700 ($329). All used AMD’s new AM4 socket and the X370 or B350 chipset. The Ryzen 7 1800X was positioned directly against Intel’s Core i7-6900K (8 cores, $1,089) — AMD offered the same core count at less than half the price. Geekbench multi-core scores placed the 1800X within 5% of the i7-6900K while costing $590 less. Read more
Published:
Cloudflare disclosed Cloudbleed on February 23, 2017, after Google Project Zero researcher Tavis Ormandy reported the bug to Cloudflare five days earlier on February 17 and Cloudflare deployed a fix within hours. The vulnerability had existed in production since September 22, 2016 — 5 months before disclosure — and the worst-case period of highest data leakage ran from February 13 to 18, 2017. Cloudflare estimated it had leaked roughly 1 in every 3,300,000 HTTP responses over the worst period. Read more
Published:
On January 31, 2017 at approximately 5:20 PM UTC, a GitLab database administrator accidentally deleted 300 GB of production PostgreSQL data from db1.cluster.gitlab.com while attempting to remove a directory on a staging server. The error occurred while manually remediating a replication lag incident under time pressure: the administrator ran rm -rf targeting a directory path but was connected to the production server rather than staging. GitLab’s CEO Sid Sijbrandij and the engineering team made an unusual decision: they live-streamed the recovery effort on YouTube, opened a public Google Doc tracking the incident state, and posted updates in real time to Twitter — narrating the disaster as it unfolded rather than waiting until service was restored. Read more
Published:
Tim Berners-Lee proposed the Semantic Web in a 2001 Scientific American article written with James Hendler and Ora Lassila. The vision was compelling: an extension of the existing web in which information would have explicit meaning, allowing software agents to reason across data from different sources, make connections automatically, and perform tasks that currently require human judgment to navigate. Read more
Published:
Amazon announced Amazon Go on December 5, 2016 via a public video, describing a 1,800 square-foot convenience store at 2131 7th Avenue in Seattle’s South Lake Union neighborhood (Amazon’s corporate campus) where shoppers could scan a QR code in the Amazon Go mobile app at an entry gate, take items from shelves, and walk out — receiving an automatic receipt and charge to their Amazon account minutes later, with no checkout lines or cashiers required. The store opened to Amazon employees in December 2016 for internal testing; the public launch was delayed over a year to January 22, 2018, reportedly due to technical difficulties tracking large crowds (more than 20 shoppers at once in early tests created tracking accuracy problems) and disambiguating who had taken items in dense crowd scenarios. The underlying technology — which Amazon branded “Just Walk Out” — combined ceiling-mounted depth cameras (hundreds per store), weight sensors embedded in store shelves, computer vision deep learning models, and the Amazon app entry ID to build a real-time “virtual cart” for each shopper. Contrary to widespread speculation, the system did not use RFID tags on individual products; all tracking was camera and sensor fusion based. Read more
Published:
Passwords are structurally weak as the sole mechanism for account security, and the events of the past few years have made this concrete rather than theoretical. In 2012, LinkedIn suffered a breach that exposed 6.5 million password hashes; in 2016, it became clear that the actual number was 117 million. Yahoo’s 2013 breach, disclosed in 2016, exposed credentials for 500 million accounts. Adobe, Dropbox, and dozens of other services have had major breaches. The stolen credentials from these breaches are traded and combined into large credential lists that attackers use in automated “credential stuffing” attacks — testing username and password combinations across many services simultaneously. Read more
Published:
Programmers have a particular relationship with difficult problems. When something does not work, the instinct is to stay. One more test. One more search. One more change to the configuration. An hour later the problem is unchanged, and the mind is worse equipped to solve it than it was when the session started. Read more
Published:
Google Home went on sale November 4, 2016 at $129, announced at Google I/O in May 2016 and launching four months after the Pixel phones introduced Google Assistant (October 4, 2016). The device competed directly with Amazon Echo (which had launched June 23, 2015 at $179.99 for Prime members) after Amazon had built a 1.5-year head start in the smart speaker market with Alexa. Google Home’s hardware included a 2-inch tweeter and passive bass radiator for audio output, a 7-microphone far-field array for voice pickup, and a 360° sound design with a fabric mesh exterior (replacing Amazon’s cylindrical black plastic aesthetic). The far-field microphone array performed beamforming — electronically steering the sensitive direction of the combined microphone pickup toward the detected voice source and suppressing the direction of the device’s own speaker output — to enable voice detection while music played at moderate volume from the same device. Always-on keyword detection (“Ok Google”) ran on-device using a small neural network model; the audio before and after keyword detection was streamed to Google’s servers for the full speech recognition and natural language processing to occur. Read more