Google has quietly acquired Provino Technologies, a start-up developing network-on-chip (NoC) systems for machine learning, an IEEE Spectrum investigation has discovered.

The latest processors for AI applications are home to thousands—or even hundreds of thousands—of cores, each of which needs to move vast swathes of data.

NoC technologies could accelerate communications on such “many-core” chips by replacing traditional buses and direct wires with an architecture familiar from large computer networks and the Internet, based on routers directing packets of data.

“Technology for communication simply hasn’t improved in the same way as that for computation,” says Md Farhadur Reza, assistant professor of Computer Science at the University of Central Missouri. “NoC’s decentralized architecture can have applications running multiple tasks in parallel and communicating with each other at the same time. And that means performance will improve, throughput will improve, and your wires will be shorter.”

Ex-Apple engineer Shailendra Desai founded Provino in 2015 to provide a platform called iFabric for developing NoC chips. The start-up was based in Silicon Valley with a second office in Ahmedabad, India.

In 2018, Provino raised $8 million in a Series A funding round led by Dell Technologies Capital, the investment arm of the computing multinational. At the time, the company identified chip development for “machine learning/artificial intelligence, consumer and automotive applications” as its focus.

“Provino’s technology has a fresh approach to system-on-chip design, addressing the challenging requirements of next generation chip design in the burgeoning artificial intelligence and machine learning markets,” wrote Daniel Docter, Managing Director of Dell Technologies Capital at the time.

Not only are NoC architectures faster and less prone to data bottlenecks than traditional chips, they are also inherently scalable, reconfigurable and fault tolerant. “There are multiple paths between the two nodes, so even if one link is down, you can still route a packet another way,” says Reza. “This makes it the most efficient architecture for neural networks.”

Neural networks are famously computationally intensive, particularly during training runs that rely on frequent communications between the “neurons” and memory.

Google started developing its own application-specific integrated circuits (ASICs) for neural networks in 2015. The hardware, called tensor processing units (or TPUs), are deployed within Google’s data centers to power AI products like Translate, Photos, Search, Assistant, and Gmail.

The precise specifications of Google’s TPUs are unknown, although some Google researchers have been studying NoC technologies for years.

In early February, Google bought 20 patents and patent applications for NoC communications and power control from Provino, for an undisclosed sum. This appears to be Provino’s entire portfolio of intellectual property.

At around the same time, the company shuttered its website, and manyofitsengineers in India now describe themselves as working for Google.

While neither Desai nor Provino responded to requests for comment, Google later confirmed to Spectrum that it had bought Provino. It provided no further details on the acquisition, but the purchase could signal a move within Google to adopt NoC technologies.

A wholesale transition to NoC is unlikely to happen overnight, says Reza: “There are still a lot of challenges from the architectural and the algorithmic point of view. Routing is at the heart of NoC and there are many questions about how to design the routers, the algorithms they use, their buffers and the capacity of the links.”

Nevertheless, anything that promises to improve the efficiency—and reduce the staggering power usage—of machine learning systems, especially at the scale at which Google operates, can only be a good thing for the future of sustainable AI.

The Conversation (0)

The Spectacular Collapse of CryptoKitties, the First Big Blockchain Game

A cautionary tale of NFTs, Ethereum, and cryptocurrency security

8 min read
Vertical
Mountains and cresting waves made of cartoon cats and large green coins.
Frank Stockton
Pink

On 4 September 2018, someone known only as Rabono bought an angry cartoon cat named Dragon for 600 ether—an amount of Ethereum cryptocurrency worth about US $170,000 at the time, or $745,000 at the cryptocurrency’s value in July 2022.

It was by far the highest transaction yet for a nonfungible token (NFT), the then-new concept of a unique digital asset. And it was a headline-grabbing opportunity for CryptoKitties, the world’s first blockchain gaming hit. But the sky-high transaction obscured a more difficult truth: CryptoKitties was dying, and it had been for some time.

The launch of CryptoKitties drove up the value of Ether and the number of transactions on its blockchain. Even as the game's transaction volume plummeted, the number of Ethereum transactions continued to rise, possibly because of the arrival of multiple copycat NFT games.

That perhaps unrealistic wish becomes impossible once the downward spiral begins. Players, feeling no other attachment to the game than growing an investment, quickly flee and don’t return.

Whereas some blockchain games have seemingly ignored the perils of CryptoKitties’ quick growth and long decline, others have learned from the strain it placed on the Ethereum network. Most blockchain games now use a sidechain, a blockchain that exists independently but connects to another, more prominent “parent” blockchain. The chains are connected by a bridge that facilitates the transfer of tokens between each chain. This prevents a rise in fees on the primary blockchain, as all game activity occurs on the sidechain.

Yet even this new strategy comes with problems, because sidechains are proving to be less secure than the parent blockchain. An attack on Ronin, the sidechain used by Axie Infinity, let the hackers get away with the equivalent of $600 million. Polygon, another sidechain often used by blockchain games, had to patch an exploit that put $850 million at risk and pay a bug bounty of $2 million to the hacker who spotted the issue. Players who own NFTs on a sidechain are now warily eyeing its security.

Remember Dragon

The cryptocurrency wallet that owns the near million dollar kitten Dragon now holds barely 30 dollars’ worth of ether and hasn’t traded in NFTs for years. Wallets are anonymous, so it’s possible the person behind the wallet moved on to another. Still, it’s hard not to see the wallet’s inactivity as a sign that, for Rabono, the fun didn’t last.

Whether blockchain games and NFTs shoot to the moon or fall to zero, Bladon remains proud of what CryptoKitties accomplished and hopeful it nudged the blockchain industry in a more approachable direction.

“Before CryptoKitties, if you were to say ‘blockchain,’ everyone would have assumed you’re talking about cryptocurrency,” says Bladon. “What I’m proudest of is that it was something genuinely novel. There was real technical innovation, and seemingly, a real culture impact.”

This article was corrected on 11 August 2022 to give the correct date of Bryce Bladon's departure from Dapper Labs.

This article appears in the September 2022 print issue as “The Spectacular Collapse of CryptoKitties.”

Keep Reading ↓Show less
{"imageShortcodeIds":[]}