Analog-in-memory AI processor startup uses memristors
TetraMem in Calfornia is developing analog-in-memory compute processor architecture using memristor technology.
The company was founded in 2018 by four engineers with common experience at Hewlett Packard working on memristor research, the non-linear two-terminal electrical component that is known more often as the resistive RAM (ReRAM).
Like other analog-in-memory compute companies TetraMem uses the cross-point network for a ReRAM array to accelerate the multiply-accumulate operations in artificial intelligence neural networks.
Other in-memory compute articles:
- Memristor can learn for neuromorphic AI
- ST hints at analog in-memory computing chip
- Axelera shows DIANA analog in-memory computing chip
- NeuroBlade raises $83m for compute in memory chip
- Molecular memristor for new low power computing architectures
- Memristors combine ferroelectrics and graphene
The company claims its patented devices are a scalable technology optimized for low-power edge and endpoint inference can be 100s of times more efficient than Von Neumann architecture devices.
The company is claiming its neural processing unit (NPU) can deliver 20TOPS/W based on the INT8 datatype using mature technology nodes. This is expected to scale to 100TOPS/W at more advanced nodes.
TetraMem is pursuing an IP licensing business model including IP licensing, royalties, and custom design services. The company has achieved three tape outs in silicon and expects mass production of chips including its technology around the end of 2023.
Other related articles
- Cognifiber raises $6m for In-Fibre photonic processing
- sureCore develops in-memory computing for edge AI
- First In Memory Computing chip for GDDR6
Other articles on eeNews Europe
- IQM raises €128m in Europe’s largest quantum computing round
- Fibonacci boost for quantum computer error correction
- ABB pulls out of Russia
- UK publishes alternative to EU Horizon research programme
- Israel to build open architecture quantum computer
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
