Showing posts with label high-performance SoC. Show all posts
Google’s Making Its Own Chips Now. Time for Intel to Freak Out
The
Internet’s most powerful company sent a few shock waves through the tech world
yesterday when it revealed that a new custom-designed chip helps run what is
surely the future of its vast online empire: artificial intelligence.
In
building its own chip, Google has taken yet another step along a path that has
already remade the tech industry in enormous ways. Over the past decade, the
company has designed all sorts of new hardware for the massive data centers
that underpin its myriad online services, including computer servers,
networking gear, and more. As it created services of unprecedented scope and
size, it needed a more efficient breed of hardware to run these services. Over
the years, so many other Internet giants have followed suit, forcing a seismic
shift in the worldwide hardware market.
With
its new chip, Google’s aim is the same: unprecedented efficiency. To take AI to
new heights, it needs a chip that can do more in less time while consuming less
power. But the effect of this chip extends well beyond the Google empire. It
threatens the future of commercial chip makers like Intel and
nVidia—particularly when you consider Google’s vision for the future. According
to Urs Hölzle, the man most responsible for the global data center network that
underpins the Google empire, this new custom chip is just the first of many.
No,
Google will not sell its chips to other companies. It won’t directly compete
with Intel or nVidia. But with its massive data centers, Google is by far the
largest potential customer for both of those companies. At the same time, as
more and more businesses adopt the cloud computing services offered by Google,
they’ll be buying fewer and fewer servers (and thus chips) of their own, eating
even further into the chip market.
Indeed,
Google revealed its new chip as a way of promoting the cloud services that let
businesses and coders tap into its AI engines and build them into their own
applications. As Google tries to sell other companies on the power of its AI,
it’s claiming—in rather loud ways—that it boasts the best hardware for running
this AI, hardware that no other company has.
Google’s
Need for Speed
Google’s
new chip is called the Tensor Processing Unit, or TPU. That’s because it helps
run TensorFlow, the software engine that drives the Google’s deep neural
networks, networks of hardware and software that can learn particular tasks by
analyzing vast amounts of data. Other tech giants typically run their deep
neural nets with graphics processing units, or GPUs—chips that were originally
designed to render images for games and other graphics-heavy applications.
These are well-suited to running the types of calculations that drive deep
neural networks. But Google says it has built a chip that’s even more
efficient.
According
to Google, it tailored the TPU specifically to machine learning so that it
needs fewer transistors to run each operation. That means it can squeeze more
operations into the chip with each passing second.
For
now, Google is using both TPUs and GPUs to run its neural nets. Hölzle declined
to go into specifics on how exactly Google was using its TPUs, except to say
that they handle “part of the computation” needed to drive voice recognition on
Android phones. But he said that Google would be releasing a paper describing
the benefits of its chip and that Google will continue to design new chips that
handle machine learning in other ways. Eventually, it seems, this will push
GPUs out of the equation. “They’re already going away a little,” Hölzle says.
“The GPU is too general for machine learning. It wasn’t actually built for
that.”
That’s
not something nVidia wants to hear. As the world’s primary seller of GPUs,
nVidia is now pushing to expand its own business into the AI realm. As Hölzle
points out, the latest nVidia GPU offers a mode specifically for machine
learning. But clearly, Google wants the change to happen faster. Much faster.
The
Smartest Chip
In
the meantime, other companies, most notably Microsoft, are exploring another
breed of chip. The field-programmable gate array, or FPGA, is a chip you can
re-program to perform specific tasks. Microsoft has tested FPGAs with machine
learning, and Intel, seeing where this market was going, recently acquired a
company that sells FPGAs.
Some
analysts think that’s the smarter way to go. An FPGA provides far more
flexibility, says Patrick Moorhead, the president and principal analyst at Moor
Insights and Strategy, a firm that closely follows the chip business. Moorhead
wonders if the new Google TPU is “overkill,” pointing out that such a chip
takes at least six months to build—a long time in the incredibly competitive
marketplace in which the biggest Internet companies compete.
But
Google doesn’t want that flexibility. More than anything, it wants speed. Asked
why Google built its chip from scratch rather than using an FPGA, Hölzle said:
“It’s just much faster.”
Core
Business
Hölzle
also points out that Google’s chip doesn’t replace CPUs, the central processing
units at the heart of every computer server. The search giant still needs these
chips to run the tens of thousands of machines in its data centers, and CPUs
are Intel’s main business. Still, if Google is willing to build its own chips
just for AI, you have to wonder if it would go so far as to design its own CPUs
as well.
Hölzle
plays down the possibility. “You want to solve problems that are not solved,”
he says. In other words, CPUs are a mature technology that pretty much works as
it should. But he also said that Google wants healthy competition in the chip
market. In other words, it wants to buy from many sellers—not just, say, Intel.
After all, more competition means lower prices for Google. As Hölzle explains,
expanding its options is why Google is working with the OpenPower Foundation,
which seeks to offer chip designs that anyone can use and modify.
That’s
a powerful idea, and a potentially powerful threat to the world’s biggest chip
makers. According to Shane Rau, an analyst with research firm IDC, Google buys
about 5 percent of all server CPUs sold on Earth. Over a recent year-long
period, he says, Google bought about 1.2 million chips. And most of those
likely came from Intel. (In 2012, Intel exec Diane Bryant told WIRED that
Google bought more server chips from Intel than all but five other
companies—and those were all companies that sell servers.)
Whatever
its plans for the CPU, Google will continue to explore chips specifically
suited to machine learning. It will be several years before we really know what
works and what doesn’t. After all, neural networks are constantly evolving as
well. “We’re learning all the time,” he says. “It’s not clear to me what the
final answer is.” And as it learns, you can bet that the world’s chip makers
will be watching.
Today, Intel owns the data center market. The only challenger in the x86 space, AMD, once claimed a significant share of that market, but has been all-but eliminated after years of noncompetitive CPU architectures. AMD has been driven to single-digit market share, though the company hopes to take back some of it with its upcoming Zen processor, due next year. Other vendors, like IBM or ARM, have an even smaller market share than AMD. That could change in the next few years, however, and Google has flung its support behind a new interconnect standard, OpenCAPI, and IBM’s POWER9 CPU architecture.
In a blog post on
Friday, Google announced that it had joined the OpenCAPI consortium, a group
dedicated to developing a next-generation set of interconnects for servers and
data centers. If this is giving you a sense of déjà vu, never fear — the Gen-Z
announcement we covered last week also
concerned a large group of companies that are developing a next-generation
interconnect, and most of the same companies are members. Gen-Z aims to develop
an interconnect standard for storage devices, heterogeneous accelerators, and
pooled memory using memory semantic fabric, while OpenCAPI uses DMA semantics.
Google and Nvidia are the only two members of OpenCAPI that aren’t also members
of Gen-Z.
In its blog post, Google documents a
new server it has developed, the Zaius P9 (which implements the OpenCAPI
standard).
Zaius is designed to use two IBM POWER9
LaGrange CPUs with support for DDR4 (16 DIMM slots per CPU, 32 total), along
with two 30-bit buses handling inter-CPU communication. POWER9 will include
support for PCI Express Gen 4, with 84 lanes spread between the two processors.
PCIe 4.0 isn’t expected to be finalized until 2017, and there’s no word on when
consumer hardware will actually be available. Power9 is expected in 2017, but
we don’t know when Google‘s Zaius
specifically will debut. The chips themselves will target a 225W TDP, well
above most of Intel’s hardware.
The goal of these new interconnect
initiatives is to challenge Intel’s dominance in this space. OpenCAPI is a
project Nvidia has prominently planned to support with the enterprise version
of its Pascal architecture, and AMD has its own reasons for cooperating with
such efforts. If it wants to win back space for Zen, it may have decided
throwing its own lot in with competitors working on new interconnects is the
right way to do that. There’s precedent for doing this — back in 2003, it was
AMD’s HyperTransport bus and its support for “glueless” multi-socket systems
that gave the company a prominent advantage over Intel in the multi-socket
server market. Even after dual and quad-core chips were available, Opteron
continued to outperform some of its Core 2-equivalents in multi-socket
configurations, at least for a little while.
The threat to Intel is in the last line of Google’s blog
post, where the company writes: “We look forward to a future of heterogeneous
architectures within our cloud. And, as we continue our commitment to open
innovation, we’ll continue to collaborate with the industry to improve these
designs and the product offerings available to our users.”
That might seem like a mild sentence,
but it’s a shot across the bow. Google is prominently backing Intel’s chief
competitors, and given the consistent downturn in the PC industry, you can bet
that Intel is taking any and all threats to its data center market extremely seriously.
Moving deeper into computer
vision, Intel is acquiring Silicon Valley startup Movidius for an undisclosed
price.
Remi El-Ouazzane, CEO of Movidius (left), and Josh Walden of Intel.
The deal will help Intel move into hot new markets with technology for drones, robots, virtual reality headsets, security cameras, and more. Intel said the move will also help it with deep learning solutions — from devices to the cloud.
“We’re entering an era where devices must be smart and connected,” said Intel senior vice president Josh Walden, in a blog post. “When a device is capable of understanding and responding to its environment, entirely new and unprecedented solutions present themselves.”
Remi El-Ouazzane, CEO of San Mateo, Calif.-based Movidius, said he was excited about the acquisition.
“Movidius’ mission is to give the power of sight to machines,” he said in a blog post. “As part of Intel, we’ll remain focused on this mission, but with the technology and resources to innovate faster and execute at scale. We will continue to operate with the same eagerness to invent and the same customer-focused attitude that we’re known for, and we will retain Movidius talent and the start-up mentality that we have demonstrated over the years.”
Movidius is working with customers like DJI, FLIR, Google, and Lenovo to give sight to smart devices, including drones, security cameras, AR/VR headsets, and more.
“When computers can see, they can become autonomous, and that’s just the beginning,” El-Ouazzane said. “We’re on the cusp of big breakthroughs in artificial intelligence. In the years ahead, we’ll see new types of autonomous machines with more advanced capabilities as we make progress on one of the most difficult challenges of A.I.: getting our devices not just to see, but also to think.”
To boost its RealSense 3D depth camera technology, Intel is acquiring critical technologies to help it be a leader in computer vision and “perceptual computing,” where sensing real-world objects is important. Movidius makes what it calls a vision processing unit (VPU).
“Simply put, computer vision enables machines to visually process and understand their surroundings,” Walden added. “Cameras serve as the ‘eyes’ of the device, the central processing unit is the ‘brain’, and a vision processor is the ‘visual cortex’. Upon integration, computer vision enables navigation and mapping, collision avoidance, tracking, object recognition, inspection analytics, and more – capabilities that are extremely compelling in emerging markets.”
Intel needs technology for recognizing objects, understanding scenes, authenticating, tracking, and navigating. Walden said that Movidius offers “massive potential” to accelerate Intel’s own plans.
“With Movidius, Intel gains low-power, high-performance system-on-chip (SoC) platforms for accelerating computer vision applications,” Walden said. “Additionally, this acquisition brings algorithms tuned for deep learning, depth processing, navigation and mapping, and natural interactions, as well as broad expertise in embedded computer vision and machine intelligence. Movidius’ technology optimizes, enhances, and brings RealSense capabilities to fruition.”
He added, “Computer vision will trigger a Cambrian Explosion of compute, with Intel at the forefront of this new wave of computing, enabled by RealSense, in conjunction with Movidius and our full suite of perceptual computing technologies.”
Movidius has focused on the device level, combining advanced algorithms with dedicated low-power hardware.
“At Intel, we’ll be part of a team that is attacking this challenge from the cloud, through the network, and on the device,” El-Ouazzane said. “This is very exciting.”
The deal will help Intel move into hot new markets with technology for drones, robots, virtual reality headsets, security cameras, and more. Intel said the move will also help it with deep learning solutions — from devices to the cloud.
“We’re entering an era where devices must be smart and connected,” said Intel senior vice president Josh Walden, in a blog post. “When a device is capable of understanding and responding to its environment, entirely new and unprecedented solutions present themselves.”
Remi El-Ouazzane, CEO of San Mateo, Calif.-based Movidius, said he was excited about the acquisition.
“Movidius’ mission is to give the power of sight to machines,” he said in a blog post. “As part of Intel, we’ll remain focused on this mission, but with the technology and resources to innovate faster and execute at scale. We will continue to operate with the same eagerness to invent and the same customer-focused attitude that we’re known for, and we will retain Movidius talent and the start-up mentality that we have demonstrated over the years.”
Movidius is working with customers like DJI, FLIR, Google, and Lenovo to give sight to smart devices, including drones, security cameras, AR/VR headsets, and more.
“When computers can see, they can become autonomous, and that’s just the beginning,” El-Ouazzane said. “We’re on the cusp of big breakthroughs in artificial intelligence. In the years ahead, we’ll see new types of autonomous machines with more advanced capabilities as we make progress on one of the most difficult challenges of A.I.: getting our devices not just to see, but also to think.”
To boost its RealSense 3D depth camera technology, Intel is acquiring critical technologies to help it be a leader in computer vision and “perceptual computing,” where sensing real-world objects is important. Movidius makes what it calls a vision processing unit (VPU).
“Simply put, computer vision enables machines to visually process and understand their surroundings,” Walden added. “Cameras serve as the ‘eyes’ of the device, the central processing unit is the ‘brain’, and a vision processor is the ‘visual cortex’. Upon integration, computer vision enables navigation and mapping, collision avoidance, tracking, object recognition, inspection analytics, and more – capabilities that are extremely compelling in emerging markets.”
Intel needs technology for recognizing objects, understanding scenes, authenticating, tracking, and navigating. Walden said that Movidius offers “massive potential” to accelerate Intel’s own plans.
“With Movidius, Intel gains low-power, high-performance system-on-chip (SoC) platforms for accelerating computer vision applications,” Walden said. “Additionally, this acquisition brings algorithms tuned for deep learning, depth processing, navigation and mapping, and natural interactions, as well as broad expertise in embedded computer vision and machine intelligence. Movidius’ technology optimizes, enhances, and brings RealSense capabilities to fruition.”
He added, “Computer vision will trigger a Cambrian Explosion of compute, with Intel at the forefront of this new wave of computing, enabled by RealSense, in conjunction with Movidius and our full suite of perceptual computing technologies.”
Movidius has focused on the device level, combining advanced algorithms with dedicated low-power hardware.
“At Intel, we’ll be part of a team that is attacking this challenge from the cloud, through the network, and on the device,” El-Ouazzane said. “This is very exciting.”
