Category: Hardware

Posted on

5G reportedly coming to premium iPhones in 2020, all models in 2021

The latest report from renowned Apple leaker Ming-Chi Kuo already has an eye on 2020 and beyond. The news lines up with other reports around future iPhones, noting that the high-end versions of the handset are set to get 5G in the second half of next year. By 2021, all models are set to be on-board with the next-gen wireless standard.

The report is inline with recent rumors that have the company holding off on 5G until 2020. That puts Apple somewhat behind the curve of a number of Android manufacturers who have been racing to get the technology to market. Of course, those companies (including Samsung, LG and even OnePlus) may be putting the cart before the horse, with wireless carriers providing extremely limited access to the tech through the end of 2019.

Apple’s 5G iPhone conundrum

Apple’s push into 5G is believed to be a primary driver behind the company’s recent decision to make nice with Qualcomm, though Kuo believes that the company is shooting for 2022/2023 to begin manufacturing its own wireless chips. That would help Apple further divorce itself on reliance from third party component makers, which seems to have been the plan all along.

Apple and Qualcomm are ending their legal battles

The report has Apple continuing to release three models of iPhone later next year. The list includes a 5.4 inch and 6.7 inch OLED models, making the smaller iPhone even smaller and the larger even larger. The XR successor, meanwhile, would maintain a 6.1 inch display, getting upgraded to OLED next year, while only offering up an LTE modem — a move that could muddy the waters a bit for consumers.

Posted on

Habana Labs launches its Gaudi AI training processor

Habana Labs, a Tel Aviv-based AI processor startup, today announced its Gaudi AI training processor, which promises to easily beat GPU-based systems by a factor of four. While the individual Gaudi chips beat GPUs in raw performance, it’s the company’s networking technology that gives it the extra boost to reach its full potential.

Gaudi will be available as a standard PCIe card that supports eight ports of 100Gb Ethernet, as well as a mezzanine card that is compliant with the relatively new Open Compute Project accelerator module specs. This card supports either the same ten 100GB Ethernet ports or 20 ports of 50Gb Ethernet. The company is also launching a system with eight of these mezzanine cards.

Last year, Habana Labs previously launched its Goya inferencing solution. With Gaudi, it now offers a complete solution for businesses that want to use its hardware over GPUs with chips from the likes of Nvidia. Thanks to its specialized hardware, Gaudi easily beats an Nvidia T4 accelerator on most standard benchmarks — all while using less power.

“The CPU and GPU architecture started from solving a very different problem than deep learning,” Habana CBO Eitan Medina told me.  “The GPU, almost by accident, happened to be just better because it has a higher degree of parallelism. However, if you start from a clean sheet of paper and analyze what a neural network looks like, you can, if you put really smart people in the same room […] come up with a better architecture.” That’s what Habana did for its Goya processor and it is now taking what it learned from this to Gaudi.

For developers, the fact that Habana Labs supports all of the standard AI/ML frameworks, as well as the ONNX format, should make the switch from one processor to another pretty painless.

“Training AI models require exponentially higher compute every year, so it’s essential to address the urgent needs of the data center and cloud for radically improved productivity and scalability. With Gaudi’s innovative architecture, Habana delivers the industry’s highest performance while integrating standards-based Ethernet connectivity, enabling unlimited scale,” said David Dahan, CEO of Habana Labs. “Gaudi will disrupt the status quo of the AI Training processor landscape.”

As the company told me, the secret here isn’t just the processor itself but also how it connects to the rest of the system and other processors (using standard RDMA RoCE, if that’s something you really care about).

Habana Labs argues that scaling a GPU-based training system beyond 16 GPUs quickly hits a number of bottlenecks. For a number of larger models, that’s becoming a necessity, though. With Gaudi, that becomes simply a question of expanding the number of standard Ethernet networking switches so that you could easily scale to a system with 128 Gaudis.

“With its new products, Habana has quickly extended from inference into training, covering the full range of neural-network functions,” said Linley Gwennap, principal analyst of The Linley Group. “Gaudi offers strong performance and industry-leading power efficiency among AI training accelerators. As the first AI processor to integrate 100G Ethernet links with RoCE support, it enables large clusters of accelerators built using industry-standard components.”

Posted on

Habana Labs launches its Gaudi AI training processor

Habana Labs, a Tel Aviv-based AI processor startup, today announced its Gaudi AI training processor, which promises to easily beat GPU-based systems by a factor of four. While the individual Gaudi chips beat GPUs in raw performance, it’s the company’s networking technology that gives it the extra boost to reach its full potential.

Gaudi will be available as a standard PCIe card that supports eight ports of 100Gb Ethernet, as well as a mezzanine card that is compliant with the relatively new Open Compute Project accelerator module specs. This card supports either the same ten 100GB Ethernet ports or 20 ports of 50Gb Ethernet. The company is also launching a system with eight of these mezzanine cards.

Last year, Habana Labs previously launched its Goya inferencing solution. With Gaudi, it now offers a complete solution for businesses that want to use its hardware over GPUs with chips from the likes of Nvidia. Thanks to its specialized hardware, Gaudi easily beats an Nvidia T4 accelerator on most standard benchmarks — all while using less power.

“The CPU and GPU architecture started from solving a very different problem than deep learning,” Habana CBO Eitan Medina told me.  “The GPU, almost by accident, happened to be just better because it has a higher degree of parallelism. However, if you start from a clean sheet of paper and analyze what a neural network looks like, you can, if you put really smart people in the same room […] come up with a better architecture.” That’s what Habana did for its Goya processor and it is now taking what it learned from this to Gaudi.

For developers, the fact that Habana Labs supports all of the standard AI/ML frameworks, as well as the ONNX format, should make the switch from one processor to another pretty painless.

“Training AI models require exponentially higher compute every year, so it’s essential to address the urgent needs of the data center and cloud for radically improved productivity and scalability. With Gaudi’s innovative architecture, Habana delivers the industry’s highest performance while integrating standards-based Ethernet connectivity, enabling unlimited scale,” said David Dahan, CEO of Habana Labs. “Gaudi will disrupt the status quo of the AI Training processor landscape.”

As the company told me, the secret here isn’t just the processor itself but also how it connects to the rest of the system and other processors (using standard RDMA RoCE, if that’s something you really care about).

Habana Labs argues that scaling a GPU-based training system beyond 16 GPUs quickly hits a number of bottlenecks. For a number of larger models, that’s becoming a necessity, though. With Gaudi, that becomes simply a question of expanding the number of standard Ethernet networking switches so that you could easily scale to a system with 128 Gaudis.

“With its new products, Habana has quickly extended from inference into training, covering the full range of neural-network functions,” said Linley Gwennap, principal analyst of The Linley Group. “Gaudi offers strong performance and industry-leading power efficiency among AI training accelerators. As the first AI processor to integrate 100G Ethernet links with RoCE support, it enables large clusters of accelerators built using industry-standard components.”

Posted on

Meet Hatch Baby’s portable, WiFi-enabled sleep device Rest+

Menlo Park-based Hatch Baby has prided itself on introducing “smart” nursery devices — including Grow, a changing pad with a built-in scale and Rest, a device doubling as a sound machine and night light.

Now, the company is introducing an updated version of Rest with Rest+ as part of an effort to help further establish Hatch Baby in the family sleep space.

The Rest+ device will still have the sound machine, night light and a “time to rise” feature found in the original. But, with feedback from many customers and Amazon reviews, Hatch Baby has now included the addition of an audio monitor and a clock.

The audio monitor is essential for letting parents check in on baby while they sleep without going into the room and potentially waking the baby up.

The clock is also a fantastic addition, in my opinion, especially for those with toddlers who can read numbers. These little people are big enough to get out of their beds but not mature enough to know moms and dads need to sleep at 4 a.m. Often advice passed from parent to parent is to put a clock in the baby room and tell kids not to come out until it shows a certain number.

It also helps establish healthy sleep habits in little ones. Most toddlers (ages one to 3) need about 12 to 14 hours of sleep in a day, spread out between nighttime and naps, according to the National Sleep Foundation. However, as any parent knows, the older a baby gets, the harder it is to get them to want to go to bed.

Rest+ features include:

  • Audio monitor: Parents can now check in on their child in their room without the risk of disrupting their little one’s sleep right from their phone — no extra gadgets necessary.
  • Sound machine: Parents can choose from a range of sound options, from white noise to soft lullabies. They can simply crank up the volume remotely when the dog barks or the neighbors throw a party.
  • Night light: This feature, which stays cool to the touch, provides soft and soothing
    lighting for midnight feeding sessions or bright and reassuring light when the dark feels scary for older kids. Parents and kids can choose from a rainbow of colors to make it their own, but the optional patented toddler-lock setting makes sure that parents are the only ones in control when needed.
  • Time-to-Rise: Green means go! This feature enables parents to teach toddlers and
    preschoolers to stay in bed until it’s time to rise once the light changes color (and enjoy those extra minutes of sleep).
  • Clock: Rest+ features an easy-to-read clock so that parents can stay on track with their busy schedules and can help teach children to read numbers.

Any one of these features could cost parents a good amount of dough when purchased separately. A Phillips Avent audio monitor runs just under $100 on Amazon, for example. However, Rest+ is just $80 (slightly more than the original $60 price tag for the Rest device), for all five features.

Something else that may make the Rest+ attractive to parents — it is WiFi-enabled and portable so you can take it with you when you travel.

Whipping a sound machine, nightlight, audio monitor and clock all into one portable, WiFi-enabled device can also save precious space in the nursery and makes this a must-have item for many parents hoping for just a little bit more sleep.

Hatch Baby co-founder Ann Crady Weiss tells TechCrunch the Rest+ will only be available on the Hatch Baby site and is part of a plan to launch a full line of products aimed at getting parents — and their children — more precious sleep. Though she wouldn’t say what the company was working on next, she did mention we’d hear something about it in the coming months. So stay tuned!