Month: April 2018

Enlarge / Some of the equipment (and a team member) that made this new work possible. (credit: Holger Müller lab)

Measurements in physics are funny things. You'd hope that attempts to quantify some of the fundamental properties of the Universe would follow a simple pattern: they'd start with large error bars, but, over time, measuring technology improves and the error bars shrink. Ideally, the value would then remain nicely within the previous error.

It almost never really works like that. In many cases, measurements cluster together for a while before a new set makes a leap to somewhere else, outside the error bounds. And, even as technology improves, some sets of error bars stubbornly refuse to overlap. A new paper out this week indicates that this is the case with the Fine Structure Constant, which describes the strength of the electromagnetic force. But instead of chalking it up to the vagaries of measurement, the researchers suggest that the difference could be real—and it tells us something about what physics might lie beyond the Standard Model.

Mighty fine

The Fine Structure Constant is a measure of electromagnetic force, and that force shows up in a large number of phenomena. This means there are plenty of ways to do measurements that tell us something about the value of the Fine Structure Constant. When it comes to high-precision measurements, researchers have come up with two different ways of doing it. The first relies on particle physics and direct measurements of the magnetic properties of the electron. The second has been to study how atoms interact with light.

Enlarge (credit: Mazda)

I don't think I'm exaggerating when I say it has never been easier to get into motorsport. A plethora of grassroots endurance racing series have sprung up in recent years, making wheel-to-wheel racing cheaper than it has been in decades. However, even this route isn't exactly cheap; in addition to your safety gear you'll still need to either build a car to the rules or rent a seat in one. For those without the mechanical skills, space, or budget to build their own race car, there are still other options: video games. After all, unlike most other games, a good racing game teaches you skills that do translate well to the real thing.

Arguably, it all started in 2008 with the first Nissan Playstation GT Academy, which used Gran Turismo to find new racing talent. GT Academy proved the concept, and, as racing games have improved, more and more motorsports programs have started taking the idea seriously. Perhaps the best example is McLaren's World's Fastest Gamer competition, the winner of which got a contract as the Formula 1 team's latest simulator test driver.

Mazda is the latest company to give it a go; the winner of its new Hot Lap Challenge will get a test next Spring in the Global Mazda MX-5 Cup car. There are two ways to win. Starting on Friday (April 13) at the Long Beach Grand Prix in California, Mazda is holding seven "at-track" qualifying rounds, bringing two simulators running iRacing to each race for that purpose. Whoever sets the fastest overall lap across those seven weekends wins a trip to the 2019 Rolex 24 at Daytona and becomes a finalist for the race car test.

Enlarge / Ryzen die shot. (credit: AMD)

The second-generation AMD Ryzen desktop processors are open for preorders today. Shipping on April 19th, the new chips start at $199 for a six-core, 12-thread part running at a base of 3.4GHz and a turbo of 3.9GHz; the prices goes up to $329 for an eight-core, 16-thread processor at 3.7/4.3GHz.

Details on the new chips are a little light, with the full reveal, including performance numbers, coming on release day. We know that the second-generation processors are an incremental improvement over the first-generation Zen architecture. They keep the same basic layout—groups of four cores/eight threads are arranged into "core complexes" (CCXes), and a Ryzen chip has two CCXes joined together. Each core has 512KB of level 2 and 2MB of level 3 cache.

The second generation increases clock speeds (the previous high-end part had clocks of 3.6/4.0GHz) and makes the processor's turbo boosting smarter. On first-generation parts, the clock boosting could happen to a pair of cores, or all cores together. This meant that if you needed, say, four fast cores, they were constrained to the "all core" turbo speed. On the second-generation chips, that turbo boosting is now available with any number of cores, just as long as there's power and thermal headroom. Workloads with more than two cores, but fewer than all of them, should be able to use more of the available power budget and hence run faster.