As it has been discussed here many times, Intel has a great fondness for Java.
This affection was again eloquently expressed by Intel’s VP of Intel Software & Services Group, Michael Greene, while discussing the collaboration between one of the world’s best known technology companies and the most popular programming language on the planet.
Understanding the JDK made easy
“With so many contributions to the open JDK, we need to understand what things are being contributed to the codebase day after day,” said Greene at his JavaOne 2016 keynote address. “This will enable Java developers to make informed decisions about changes we’re making to the codebase.” Michael took this opportunity to announce the new performance portal for the JDK, an Intel website where performance metrics and training information are published for each build. Intel engineers will be actively tracking regressions and working with community members to resolve them.
Some of Intel’s own recent contributions include APIs for performant code, lexicographical array comparison, and checksum support. Making multithreaded applications, including those using MapReduce, more scalable is one way Intel hopes to ease the challenges of Java developers. IoT also featured prominently in Greene’s discussion of ways Intel is working hard on behalf of the programming community. “We’ve added more support for sensors and low energy Bluetooth for healthcare, security, and wearables.” Intel is also serving the enterprise space by extending beyond maker platforms into commercial platforms. Robots, drones, and industrial machines are all in line for innovation with these new features.
Java makes science and social more efficient
Special guests from the ultra-serious field of particle physics and the ephemeral field of social media made an appearance to talk with Greene about their own experiences with Java in terms of efficiency. Ben Wolff, a CERN software engineer, spoke about how Java keeps the wheels turning, metaphorically speaking, at the world’s largest particle accelerator—the Large Hadron Collider. Java has been used at CERN since the mid-nineties and is involved in the control system which features about 100,000 devices and 2 million endpoints. That’s not even counting the world-class ERP and BI data warehouse. It takes a lot of code to keep up.
Maintaining, managing, and updating all the code is quite a feat. Ben suggested that the next version of Java could well make this easier. “I’m personally looking forward to seeing the new modularization feature in Java9.” The institute works with a large system that includes internal and public APIs. Managing all the dependencies and inter-dependencies has been a true journey into classpath hell. Jigsaw may prove to ease this pain.
Nandini Ramani, VP of Engineering at Twitter, talked about the difference it has made for the social media giant to transition to the JVM from its monolithic Ruby on Rails stack. The company is now building in Java using Scala, deploying OpenJDK8 on Linux. Back in 2010 during the World Cup, traffic caused the Twitter site to fail over and over. In contrast, Ellen DeGeneres’ viral tweet about Bradley Cooper in 2014 proved the JVM’s ability to allow Twitter activity to scale to 3.3 million events over a period of eleven seconds. Ramani is looking forward to additional OpenJDK improvements that focus on low latency garbage collection and a pause time of less than 10ms.
Java and Intel will help connect the world
Expansion was the final frontier covered in Greene’s keynote. He shared a short, futuristic film that helped audience members visually comprehend the implications of “The Connection Effect”. By 2020, Intel estimates that there will be 50 billion devices and 200 billion sensors deployed throughout the world. This fact has profound implications for software programming. Compute, analytics, and storage will become even more important, and 5G connectivity must become ubiquitous.
Intel and Micron Technology have been collaborating on the 3DXpoint, a revolution in non-volatile memory that is intended to support large datasets. Intel predicts that the Xpoint will be up to 1,000 times faster than NAND with 1,000 times more endurance and ten times the density. Along with next generation field programmable arrays, Intel certainly appears to be gearing up for the expansion that will support a fully connected human ecosystem.