Month: November 2019

• 

  Last year, Chevrolet took a 550kW electric Camaro hot rod to SEMA, so this year Ford brought Mustang Lithium, which boasts 671kW. That's 900hp. [credit: Ford ]

This week, Las Vegas is playing host to SEMA, an annual trade show and celebration of the custom car. And although multiple reports claim that the show is nothing but wall-to-wall Toyota Supras, each different to the last, a pair of customs from Ford and Chevrolet caught my eye. (The fact that Ford and Chevrolet both have PR machines that let me know they have pictures to share is entirely not coincidental.) One is a Mustang with even more power than the Shelby GT500 we tested recently. The other is a restored Chevy farm truck. The thing they have in common? Both have been converted to electric power.

Restomodding electric motors and batteries into cars is not a new thing. San Diego's Zelectric has been converting air-cooled German classics over to the way of the electron for some time now. Jaguar developed a drop-in electric powertrain for the E-Type which it will sell from next year for an unspecified (but I'm guessing six-digit) price. Chevy even did the "converted electric muscle car to SEMA" thing 12 months ago, with the eCOPO Camaro, which used an 800V architecture and a pair of motors to send 550kW (737hp) and 813Nm (600lb-ft) to the rear wheels.

Talk about a mood stabilizer

Ford worked with Webasto on its SEMA special, called Mustang Lithium. It's similar in concept to last year's eCOPO from Ford's deadliest rival but with everything turned up just that little bit more. The Lithium is based on a current Mustang fastback, minus the internal combustion stuff. It also uses an 800V architecture, with batteries supplied by Webasto. A single electric motor from Phi-Power sends "more than" 671kW (900hp) and 1,356Nm (1,000lb-ft) to the rear wheels, which should generate plenty of heat, which is good because the Mustang Lithium is destined to be a testbed for things like battery-management strategies.

Enlarge / MLPerf offers detailed, granular benchmarking for a wide array of platforms and architectures. While most ML benchmarking focuses on training, MLPerf focuses on inference—which is to say, the workload you use a neural network for after it's been trained. (credit: MKLPerf)

When you want to see whether one CPU is faster than another, you have PassMark. For GPUs, there's Unigine's Superposition. But what do you do when you need to figure out how fast your machine-learning platform is—or how fast a machine-learning platform you're thinking of investing in is?

Machine-learning expert David Kanter, along with scientists and engineers from organizations such as Google, Intel, and Microsoft, aims to answer that question with MLPerf, a machine-learning benchmark suite. Measuring the speed of machine-learning platforms is a problem that becomes more complex the longer you examine it, since both problem sets and architectures vary widely across the field of machine learning—and in addition to performance, the inference side of MLPerf must also measure accuracy.

Training and inference

If you don't work with machine learning directly, it's easy to get confused about the terms. The first thing you must understand is that neural networks aren't really programmed at all: they're given a (hopefully) large set of related data and turned loose upon it to find patterns. This phase of a neural network's existence is called training. The more training a neural network gets, the better it can learn to identify patterns and deduce rules to help it solve problems.

Enlarge / Poor piggy. (credit: Getty Images | Aurich Lawson)

AT&T is adding $10 to the monthly bills of customers with certain grandfathered mobile-data plans and not letting them switch back to their older packages. AT&T is pitching the change as a "bonus" because it's also adding 15GB to the customers' monthly data allotments.

"Enjoy more data," AT&T says in a support document. "Starting with your October 2019 bill, you'll get an additional 15GB of data on your Mobile Share plan. This bonus data comes with a $10 price increase."

Paying an extra $10 for another 15GB isn't a bad deal as far as US wireless prices go, but that's only true if you actually need the extra data. The plans getting the data-and-price increases already had between 20GB and 60GB of data a month at prices that ranged from $100 to $225. Now those plans have 35GB to 75GB and cost $110 to $235. (The data allotments can be shared among multiple people on the same family plan.)

Enlarge

Greetings, Arsians! The Dealmaster is back with another batch of deals and discounts to share. Topping our list today is a sale on a three-month Audible membership. All-new subscribers can get three months of the audiobook subscription service for just $6.95 per month—that's a total of $20.85 for three months.

Typically, Audible costs $14.95 per month, making a three-month subscription $45. This deal shaves almost 50% off the normal price, and it's likely one of the best deals we'll see on Audible memberships until the end of the year. Audible discounts are infrequent, and it's unlikely that Amazon (Audible's parent company) will discount the service even more for Black Friday or Cyber Monday. It's always possible—but we don't expect it.

The only caveat is that you must be a totally new Audible member to get this deal, so those who have subscribed in the past and have cancelled cannot renew their subscriptions at the discounted rate. Audible is clearly aiming this sale at new subscribers in the hopes that they will continue their memberships well after the three-month discounted rate expires.

Enlarge (credit: Cloudhead Games)

A review of the 2018 virtual reality sensation Beat Saber can be boiled down to one or two sentences: wave lightsabers to the rhythm of intense, catchy music. It's a thin elevator pitch, yet all of its pieces add up to something addictive, inventive, and—based on what I've seen from other recent VR apps—hard to replicate.

That changes this week with Pistol Whip, which both evokes the simple genius of Beat Saber yet actually delivers on that rare combination of familiar and fresh. Its single-sentence pitch is just as fun: pretend you're John Wick and get into gun-fu battles against hypercolor hitmen to the rhythm of thudding techno. (No, this isn't a licensed John Wick game, but rather an obvious homage to the house that Reeves built.)

In its current state, on PC-VR systems and the standalone Oculus Quest, Pistol Whip is already an impressive trip of a "rhythm shooter," and it blends some of VR gaming's best qualities—tracked hands, body presence, and quick reactions—to deliver a body-filling sensation of badassery. Yet in its current state, it also sits on the boundary of an "early access" game, in spite of not being advertised as one. The issues are a bit annoying, but I'm having too good of a time to not otherwise recommend this gem of a 2019 VR game.

Guess the tragic acronym between kills

• 

  Every screen in this gallery was captured from live PC gameplay during our review period. Each level is a three-color auto-glide through a hallway of doom.

In Pistol Whip, you stand on what is essentially a virtual motorized sidewalk—the kind you might find at an airport terminal—and slowly glide through trippy, distorted environments. An underground bunker. A city celebrating Dia De Los Muertos. A seemingly random explosion of geometric shapes. We never quite learn why we're in these places, nor why they're full of pistol-wielding men in suits. LSD? PTSD? Some intense acronym is probably at play.

More than a year after undergoing a 14-hour operation to transplant a penis, scrotum, and lower abdominal wall, a severely injured veteran reports that he has regained normal sensation and function of his new body parts.

The young man now has “near-normal” erections, the ability to achieve orgasms, and normal sensation in the shaft and tip of his transplanted penis, according to his medical team at Johns Hopkins School of Medicine. He urinates while standing up with a “strong stream” and no issues with urgency or straining. (The transplant did not include donor testicles, to avoid the possibility of fathering non-genetically related children.)

The doctors reported the update on their patient November 7 in the New England Journal of Medicine.