Search This Blog

Thursday, July 6, 2017

NVIDIA will Power Self-Driving Cars in China

As reported by Engadget: NVIDIA has already forged self-driving alliances with big car manufacturers like Audi, Toyota and Volvo, but its latest is a particularly big deal -- at least if you live in China. The chip designer has unveiled a partnership with Chinese internet giant Baidu that will see the two work together to boost the use of AI. Most notably, NVIDIA's Drive PX tech will find its way into Baidu's Apollo self-driving car platform and autonomous vehicles from "major" Chinese firms. The automotive pact is important enough that Baidu chief Robin Li traveled to the event in one of his company's driverless rides -- even though it was against the law.


The pact will also make NVIDIA's deep learning Volta GPUs available to Baidu Cloud customers, optimize Baidu's deep learning platform (PaddlePaddle) for those Volta processors and use Baidu's conversational AI, DuerOS, for voice commands on NVIDIA's Shield TV.

It's a business win for NVIDIA, of course, but it'll be particularly important for Baidu and anyone in China eager to take their hands off the wheel. Baidu is determined to catch up to Waymo, GM and other companies that have spent years developing self-driving cars, and creating an open platform with NVIDIA's help (Baidu has announced over 50 partners including Ford, Intel and Microsoft) could help it make up for lost time. This leg up, in turn, could make autonomous driving relatively commonplace in China without needing as much help from foreign brands.

NVIDIA Volta GPU

Monday, July 3, 2017

First Production Tesla Model 3's Expected Friday, Elon Musk Says

Ramping up to 20,000 per month by December 2017.
Tesla CEO Elon Musk announced that the first production model of the hotly anticipated Model 3, the company’s attempt to bring an electric car to the masses, is expected on Friday. The first 30 Model 3 customers will receive their new Teslas on the 28th at a handover party, according to a second tweet from Musk. Production is then expected to grow exponentially: 100 cars in August, more than 1,500 by September, and then 20,000 per month by December.
News of SN1 (Serial Number 1) came in a tweet on Sunday night:
Musk then confirmed the production ramp-up in two followup tweets.
Tesla is expected to dramatically increase Model 3 production in 2018 with total Tesla vehicle production approaching 500,000 units annually. The Model 3 already has over 400,000 pre-orders with Tesla’s rabid fanbase clamoring for any glimpse or tidbit of news related to the upcoming vehicle.
The Model 3 was first unveiled over a year ago at a lavish event at Tesla’s headquarters in Hawthorne, California. At the time, Musk said he was "fairly confident" that deliveries will begin by the end of 2017, and "you will not be able to buy a better car for $35,000, even with no options."
At a shareholder meeting last month, Musk said the first Model 3 customers would be limited in their ability to customize their orders — basically just color and wheel type. “I should say that we’ve kept the initial configurations of the Model 3 very simple,” Musk said. “A big mistake we made with the X, which is primarily my responsibility — there was way too much complexity right at the beginning. That was very foolish.”
Tesla’s sky-high valuation — it recently surpassed BMW’s market cap — depends largely on Musk’s ability to sell his vision of sustainable, battery-powered driving to a much broader population. The two current Tesla vehicles, the Model S and Model X, are both extremely expensive. Even with tax incentives, both cars easily push $100,000. The Model 3 will start at $35,000, making it the cheapest in Tesla’s range.
In order for Tesla to sell ten times as many cars as it does now, it needs a much cheaper automobile. That's the Model 3. It's the future of the company.

Friday, June 30, 2017

Watch the Faraday Future FF 91 EV Tackle Pikes Peak

It set a record in the process, which is a nice feather for the Faraday's cap.
When Faraday Future went to Pikes Peak, it came back with a record. Now, you can watch the whole record run from start to finish.
The Pikes Peak YouTube page recently put up a video of Faraday Future's record-setting run. Its 11-minute, 25.083-second time put the all-electric FF 91 more than 20 seconds ahead of the previous production-EV record holder, a lightly modified Tesla Model S.
Pikes Peak wasn't just for fun. It helped Faraday Future figure out what vehicle components need strengthening, and according to the FF 91's pilot -- who also happens to be Faraday Future's principal engineer -- Faraday Future did find room for improvement in its battery pack relays and system seals.
The Faraday Future was stripped and caged for safety, but it carried the same hardware and software that will be present on the production model... if it gets to production. Faraday Future claims that the production FF 91 will pack 1,050 all-electric horsepower, sport a range north of 300 miles, and possess the capability for some degree of semi-autonomous driving.

Thursday, June 29, 2017

Elon Musk's Boring Machine 'Godot' Completes the First Section of an LA Tunnel

Godot is here, and operational.
As reported by The Verge: Serial entrepreneur Elon Musk says his ambitious tunnel-boring endeavor, aptly named The Boring Company, has officially started digging underneath Los Angeles. Musk announced the news on Twitter, where he said “Godot,” the Samuel Beckett-inspired name of the company’s tunnel boring machine, had completed the the first segment of a tunnel in the Southern California metropolis. Prior to today, it was unclear how long it would take Musk to convince the city to allow him to move the experimental effort beyond the SpaceX parking lot in Hawthorne.
Musk has made a rather public showing of his offbeat tunnel-digging venture over the last few months, with an occasional flurry of Twitter announcements and Instagram posts to commemorate each new milestone. In May, Musk posted videos of test runs of the electric sled mechanism that would theoretically ferry cars at speeds up to 124 mph. Back in April, the company also put out a pretty neat concept video showing what an interconnected tunnel network might look like in a decade or two when it’s fully built out.
Along the way, Musk has attracted a ton of interest from his usual fans, as well as and the expected crowd of naysayers who think his idea of an underground tunnel network to cut down on traffic congestion is just a pipe dream. Proving his critics wrong, as Musk seems engineered to do, The Boring Company has made substantial progress since it became a real company late last year.
We don’t have details on what Musk hammered out with the city of LA. But he did tweet earlier this month about a meeting with L.A Mayor Eric Garcetti to lay the groundwork for the necessary permits and regulatory approvals he’d need to start digging with Godot, which weighs about 1,200 tons and runs about 400 feet long. Musk said last month that the first tunnel would run from LAX to Culver City, Santa Monica, Westwood, and Sherman Oaks, with later tunnels covering more of the greater LA area. Now, it looks like the LAX to Culver City route appears underway.

Monday, June 26, 2017

SpaceX Is Making Commercial Space Launches Look like Child’s Play

Two successful launches in a weekend, one using a recycled rocket—and the booster landed successfully each time.

As reported by MIT Technology Review: Late Friday, SpaceX launched a satellite into orbit from Florida using one of its refurbished Falcon 9 rockets. Then on Sunday, for good measure, it lofted 10 smaller satellites using a new version of the same rocket, which it launched from California. The feat is a sign that the private space company seems more likely than ever to turn its vision of competitively priced, rapid-turnaround rocket launches into reality.

Ever since it was established, SpaceX has sought to turn commercial spaceflight into a profitable venture. A fundamental part of that vision is its reusable rockets, which were one of our 10 Breakthrough Technologies of 2016. By launching, landing, and refurbishing boosters, it hopes to massively cut the cost of putting satellites into space. But it hopes to crank up the frequency of launches, too, in order to generate as much revenue as possible.

It showed in March that the first of those goals was achievable. But numbers from the weekend’s launch activity show that it’s doing a very good job on the second, too. The successful missions notch the firm’s current launch count for 2017 up to nine—the most it’s ever achieved in a calendar year, and we’re still only six months in. The firm has now managed to land 13 of its rockets after launch, and Friday’s mission was the second time it’s successfully reused a rocket.

Meanwhile, the rest of the space industry looks on. While plenty of organizations are trying their hand at developing ways to reduce the cost of launches, none can match SpaceX’s success to date. Its president, Gwynne Shotwell, said earlier this year that the cost of flying a refurbished booster was potentially less than half that of building a new one for each launch. And its CEO, Elon Musk, knows from experience that those successes are positioning SpaceX to become an incredibly powerful force in the future of space exploration.

“Imagine if we were an aircraft company selling aircraft that could be flown many times, and everyone else was selling aircraft that could be flown once,” he said after the first successful reuse of a SpaceX rocket booster. “I mean, you know, that’s not a very competitive position to be in.” And this point, it looks more than ever as if the commercial space race is SpaceX’s to lose.





Friday, June 23, 2017

The Air Force and IBM are Building an AI Supercomputer


Supercomputers today are capable of performing incredible feats, from accurately predicting the weather to uncovering insights into climate change, but they still by and large rely on brute processor power to accomplish their tasks. That's where this new partnership between the US Air Force and IBM comes in. They're teaming up to build the world's first supercomputer that behaves like a natural brain.

IBM and the USAF announced on Friday that the machine will run on an array of 64 TrueNorth Neurosynaptic chips. The TrueNorth chips are wired together like, and operate in a similar fashion to, the synapses within a biological brain. Each core is part of a distributed network and operate in parallel with one another on an event-driven basis. That is, these chips don't require a clock, as conventional CPUs do, to function.



What's more, because of the distributed nature of the system, even if one core fails, the rest of the array will continue to work. This 64-chip array will contain the processing equivalent of 64 million neurons and 16 billion synapses, yet absolutely sips energy -- each processor consumes just 10 watts of electricity.

Like other neural networks, this system will be put to use in pattern recognition and sensory processing roles. The Air Force wants to combine the TrueNorth's ability to convert multiple data feeds -- whether it's audio, video or text -- into machine readable symbols with a conventional supercomputer's ability to crunch data.

This isn't the first time that IBM's neural chip system has been integrated into cutting-edge technology. Last August, Samsung installed the chips in its Dynamic Vision Sensors enabling cameras to capture images at up to 2,000 fps while burning through just 300 milliwatts of power.

Tuesday, June 20, 2017

It's Time to Embrace the eSIM

Goodbye, and good riddance, to SIM cards.
As reported by Engadget: It's a downright shame that eSIMs aren't commonplace by now. Embedded-SIM technology has the potential to make getting connected to cellular networks much more convenient, but there hasn't been a consumer-friendly set of specifications for it since its 2013 introduction. That is, until last year, when the GSM Alliance (GSMA) released updated guidelines to add support for multiple profiles and devices (more on that later). Since then, thanks to partnerships between Microsoft, Intel and Qualcomm on a new generation of connected PCs with eSIMs onboard, we're going to see the technology feature in all sorts of gadgets over the next few years.

For the uninitiated, SIM stands for subscriber identity module, and it's generally a tiny, fingernail-size piece of plastic that you slide into a tray on your phone, laptop, tablet or smartwatch. Typically, it's found in your phone and contains a unique reference number for your account so that your mobile service provider knows whom to charge and how much access to grant you. The card also has some onboard memory to store a small number of your contacts and SMS messages.




But fiddling with a tiny physical card is archaic and frustrating (who wants to carry around a SIM ejector?), and eSIMs can alleviate that pain. Embedded SIMs integrate the identification technology of the plastic card into the device's processor or modem itself. For Intel-branded chips, this will be supported in its existing XMMTM 7260 modem and upcoming XMM 7360 model, while Qualcomm offers it in the Snapdragon 835 chipset. But that doesn't mean you'll have to buy a new device. If your machine already has a SIM card tray, you'll be able to slide in an adapter.

And don't worry about being locked to one carrier. Thanks to the new version of what's called "remote SIM provisioning (RSP)," eSIMs can store and adopt different profiles (or accounts) so you can simply switch carriers without having to get a new card (we'll get to what this looks like on your device in a bit). This means you could get wireless plans when you're traveling or buy specific amounts of data without having to visit a carrier's physical store. The GSMA, which represents the interests of about 800 mobile operators worldwide, updated its RSP specifications in March to extend this capability to gadgets other than phones.


The Samsung Gear S3 pictured above is one of the first consumer devices to use an eSIM.

Carriers will have to support this technology before you can access their networks over eSIM. So far, T-Mobile, AT&T and about 20 providers worldwide have said they'll work with Microsoft to let eSIM-connected PCs buy data from the Windows Store. What that process will look like isn't clear yet. Microsoft said it's working on making the carrier-selection process part of the Windows 10 interface. It could be as easy as picking a wireless operator the way you select a WiFi network, then going to the Windows Store to buy the amount of data you think you'll need.

Based on a presentation made by Deutsche Telecom/T-Mobile at the GSMA's March event, it could also involve entering an activation code provided by the operator in your device's settings.

Not only will eSIMs save you trips to physical stores and remove the need to fiddle with a tiny piece of plastic, the space saved by eliminating the card tray could also make for smaller gadgets. Because SIM cards don't take up all that much room in laptops and phones, the space gained is most significant in things like wearables and connected devices. If they didn't have to accommodate physical SIM trays, device makers could design LTE-capable smartwatches that are a bit slimmer.

We're still at least a few months away from seeing the benefits of eSIM and the resulting carrier flexibility. But given the support from big brands like ASUS, Lenovo and HP, which have all signed on to make eSIM-enabled PCs, it's clear we'll see much more of the feature soon. Plus, who knows? In a few more years, we could even say goodbye to physical SIM cards altogether. And good riddance.