TetraMem shows off ‘Cullinan’ analog-in-ReRAM SoC

TetraMem shows off ‘Cullinan’ analog-in-ReRAM SoC

Technology News |
By Peter Clarke

TetraMem Inc. (Fremont, Calif.), a startup that is developing analog-in-ReRAM compute processor architecture, has published a paper that shows its form of analog computing is capable of executing calculations with arbitrary precision.

The ability to perform high-precision multiplication within single electronic devices that can be readily formed in arrays offers scope to reduce the power consumption of machine learning when based on artificial neural networks.

The publishing of Programming memristor arrays with arbitrarily high precision for analog computing in Science follows on from the news that TetraMem has worked with Synopsys to develop an analog-in-memory compute AI accelerator SoC. That SoC is apparently called Cullinan.

Layout of analog in-memory computing accelerator SoC with A) whole SoC and B) diagram of one of ten identical computing cores. Source: Supplementary materials to TetraMem’s Science paper.

The article provides a technical underpinning to the premise that analog computing can support calculations with arbitrary precision. This plays to TetraMem’s fundamental device a metal-oxide based cross-point switch, which was reported in Nature in April 2023 (see Single memristor provides 11-bit memory device).

The device is made of a mixture of Al3O2, above a layer of HfO2 sandwiched between a tantalum/titanium top electrode and a platinum bottom electrode. Each of the bilayers is less than 1nm thick so that after being laid down they appear to form a mixed layer rather than two separate continuous layers. The device was fabricated in a 240-nm diameter via above the CMOS peripheral circuitry.

Conductive channels are formed in the material by applying a set voltage and that can be at varying levels of strength. The authors propose that the conductive and insulating phases of the metal-oxide material relate to the orthorhombic phase with a high number of oxygen vacancies and the monoclinic phase without oxygen vacancies, respectively.

“We are creating the world’s most efficient AI accelerator chip, built around a RISC-V architecture and powered by our novel analog in-memory computing technology,” said David George, head of global operations for TetraMem, at the time of the Synopsys announcement.

TetraMem was co-founded in 2018 by Joshua Yang, Professor of Electrical and Computer Engineering at the University of Southern California, Professor Qiangfei Xia of the University of Massachusetts, Miao Hu (CTO), and Glenn Ge (CEO). The company had already produced a number of test chips in silicon by mid-2022 and at that time expected to see mass production of chips including its technology by about the end of 2023.

Related links and articles:

The article in Science can be accessd here:

News articles:

Single memristor provides 11-bit memory device

HP helps birth analog-in-memory processor startup

Chinese startup raises funds for in-memory AI

Strong backing for digital-in-memory computing


If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :    eeNews on Google News


Linked Articles