China startup Witinmem uses analog flash for compute-in-memory
Analog flash memory from Solid Storage Technology Inc. is being used by Witinmem in its WTM2101 neural networking chip.
Solid State Technology is a subsidiary of Microchip Technology Inc. (Chandler, Ariz.). Witinmem (Beijing, China), founded in 2017 has launched an integrated storage and computing accelerator the WTM1001 and the integrated storage and computing SoC, the WTM2101.
Computing-in-memory is expected to be deployed in artificial intelligence (AI) speech processing at the network’s edge but requires an embedded memory solution that simultaneously performs neural network computation and stores weights.
To achieve this Witinmem has made use of SST’s SuperFlash memBrain technology the memBrain neuromorphic memory product is optimized to perform vector matrix multiplication (VMM) for neural networks. The result is 10 to 20 times lower power consumption than alternative approaches, along with lower overall processor bill of materials (BOM) costs because external DRAM and NOR are not required.
Instant on
The WTM2101 can recognize hundreds of command words, in real time and immediately after power-up on a sub-milliampere current budget, Microchip said, in a statement.
The SoC features computing-in-memory technology for neural networks processing including speech recognition, voice-print recognition, deep speech noise reduction, scene detection, and health status monitoring. WITINMEM, in turn, is working with multiple customers to bring products to market during 2022 based on this SoC.
Permanently storing neural models inside the memBrain array also supports instant-on functionality for real-time neural network processing. Witinmem has leveraged the floating gate nonvolatility to power down its computing-in-memory macros during the idle state to further reduce leakage power in demanding IoT use cases.
“The Witinmem SoC showcases the value of using memBrain technology to create a single-chip solution based on a computing-in-memory neural processor that eliminates the problems of traditional processors that use digital DSP and SRAM/DRAM-based approaches for storing and executing machine learning models,” said Mark Reiten, vice president of the license division at SST, in a statement issued by Microchip.
WTM series chips are used in low-power AIoT applications such as wearable devices and smart terminal devices. Witinmem has obtained five rounds of industrial capital-led investment and financing, with a total financing of 300 million yuan (about US$48 million).
Related links and articles:
News articles:
Analog machine learning chip lowers ‘always-on’ power
Infineon gets ‘always-on’ with Aspinity deal
IP startup launches with ‘new analog’ approach
Agile Analog gains $19 million for global expansion
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
