Tag Archives: Microsemi

Embedded AI (artificial intelligence) with a variant of a memristor?

I don’t entirely get how ReRAM (resistive random access memory) is a variant of a memristor but I’m assuming Samuel K. Moore knows what he’s writing about since his May 16, 2018 posting is on the Nanoclast blog (hosted by the IEEE [Institute of Electrical and Electronics Engineers]), Note: Links have been removed,

Resistive RAM technology developer Crossbar says it has inked a deal with aerospace chip maker Microsemi allowing the latter to embed Crossbar’s nonvolatile memory on future chips. The move follows selection of Crossbar’s technology by a leading foundry for advanced manufacturing nodes. Crossbar is counting on resistive RAM (ReRAM) to enable artificial intelligence systems whose neural networks are housed within the device rather than in the cloud.

ReRAM is a variant of the memristor, a nonvolatile memory device whose resistance can be set or reset by a pulse of voltage. The variant Crossbar qualified for advanced manufacturing is called a filament device. It’s built within the layers above a chip’s silicon, where the IC’s interconnects go, and it’s made up of three layers: from top to bottom—silver, amorphous silicon, and tungsten. Voltage across the amorphous silicon causes a filament of silver atoms to cross the gap to the tungsten, making the memory cell conductive. Reversing the voltage pushes the silver back into place, cutting off conduction.

“The filament itself is only three to four nanometers wide,” says Sylvain Dubois, vice president of marketing and business development at Crossbar. “So the cell itself will be able to scale below 10-nanometers.” What’s more, the ratio between the current that flows when the device is on to when it is off is 1,000 or higher. …

A May 14, 2018 Crossbar news release describes some of the technical AI challenges,

“The biggest challenge facing engineers for AI today is overcoming the memory speed and power bottleneck in the current architecture to get faster data access while lowering the energy cost,” said Dubois. “By enabling a new, memory-centric non-volatile architecture like ReRAM, the entire trained model or knowledge base can be on-chip, connected directly to the neural network with the potential to achieve massive energy savings and performance improvements, resulting in a greatly improved battery life and a better user experience.”

Crossbar’s May 16, 2018 news release provides more detail about their ‘strategic collaboration’ with Microsemi Products (Note: A link has been removed),

Crossbar Inc., the ReRAM technology leader, announced an agreement with Microsemi Corporation, the largest U.S. commercial supplier of military and aerospace semiconductors, in which Microsemi will license Crossbar’s ReRAM core intellectual property. As part of the agreement, Microsemi and Crossbar will collaborate in the research, development and application of Crossbar’s proprietary ReRAM technology in next generation products from Microsemi that integrate Crossbar’s embedded ReRAM with Microsemi products manufactured at the 1x nm process node.

Military and aerospace, eh?

Intel to produce Panasonic SoCs (system-on-chips) using 14nm low-power process

A July 8, 2014 news item on Azonano describes a manufacturing agreement between Intel and Panasonic,

Intel Corporation today announced that it has entered into a manufacturing agreement with Panasonic Corporation’s System LSI Business Division. Intel’s custom foundry business will manufacture future Panasonic system-on-chips (SoCs) using Intel’s 14nm low-power manufacturing process.

Panasonic’s next-generation SoCs will target audio visual-based equipment markets, and will enable higher levels of performance, power and viewing experience for consumers.

A July 7, 2014 Intel press release, which originated the news item, reveals more details,

“Intel’s 14nm Tri-Gate process technology is very important to develop the next- generation SoCs,” said Yoshifumi Okamoto, director, Panasonic Corporation SLSI Business Division. “We will deliver highly improved performance and power advantages with next-generation SoCs by leveraging Intel’s 14nm Tri-Gate process technology through our collaboration.”

Intel’s leading-edge 14nm low-power process technology, which includes the second generation of Tri-Gate transistors, is optimized for low-power applications. This will enable Panasonic’s SoCs to achieve high levels of performance and functionality at lower power levels than was possible with planar transistors.

“We look forward to collaborating with the Panasonic SLSI Business Division,” said Sunit Rikhi, vice president and general manager, Intel Custom Foundry. “We will work hard to deliver the value of power-efficient performance of our 14nm LP process to Panasonic’s next-generation SoCs. This agreement with Panasonic is an important step in the buildup of Intel’s foundry business.”

Five other semiconductor companies have announced agreements with Intel’s custom foundry business, including Altera, Achronix Semiconductor, Tabula, Netronome and Microsemi.

Rick Merritt in a July 7, 2014 article for EE Times provides some insight,

“We are doing extremely well getting customers who can use our technology,” Sunit Rikhi, general manager of Intel’s foundry group, said in a talk at Semicon West, though he would not provide details. …

He suggested that the low-power variant of Intel’s 14nm process is relatively new. Intel uses a general-purpose 22nm process but supports multiple flavors of its 32nm process.

Intel expects to make 10nm chips without extreme ultraviolet (EUV) lithography, he said, reiterating comments from Intel’s Mark Bohr. …

This news provides an update of sorts to my October 21, 2010 posting,

Paul Otellini, Chief Executive Officer of Intel, just announced that the company will invest $6B to $8B for new and upgraded manufacturing facilities to produce 22 nanometre (nm) computer chips.

Now, almost our years later they’re talking about 10 nm chips. I wonder what 2018 will bring?