Startup Runs AI in Novel SRAM
Areanna claims 100 TOPS/W in simulation
July 22, 2019
SAN JOSE — In his spare time, an engineer at Tektronix sketched out a novel deep-learning accelerator, and now his two-person startup is the latest example of the groundswell of enthusiasm that deep learning is generating.
Behdad Youssefi defined an SRAM with specialized cells that can handle the matrix multiplication, quantization, storage and other jobs needed for an inference processor. After four years solo work on the concept originally planned as a PhD thesis, he formed startup Areanna with a colleague at Tektronix and a Berkeley professor as an advisor.
In Spice simulations the design delvers more than 100 tera-operations/second/watt when recognizing handwritten digits using 8-bit integer math. Youssefi claims it could beat Google’s TPU in computational density by an order of magnitude.
To read the full article, click here
Related Semiconductor IP
- 5G-NTN Modem IP for Satellite User Terminals
- AXI-S Protocol Layer for UCIe
- HBM4E Controller IP
- 14-bit 12.5MSPS SAR ADC - Tower 65nm
- 5G-Advanced Modem IP for Edge and IoT Applications
Related News
- Startup Slashes SRAM Power With Standard Logic Process
- Israeli AI startup NeuReality raises $35M Series A to bring its novel inferencing chip to the market
- RISC-V Exceeding Expectations in AI, China Deployment
- Qualitas Semiconductor Expands Global Presence with 4nm UCIe and PCIe Gen 6.0 IP Licensing Agreement in the U.S. AI Market
Latest News
- OpenTitan Ships in Chromebooks: First Production Deployment
- Breker Verification Systems Adds RISC‑V Industry Expert Larry Lapides to its Advisory Board
- Weebit Nano’s ReRAM Selected for Korean National Compute-in-Memory Program
- Marvell Extends ZR/ZR+ Leadership with Industry-first 1.6T ZR/ZR+ Pluggable and 2nm Coherent DSPs for Secure AI Scale-across Interconnects
- BrainChip Announces Neuromorphyx as Strategic Customer and Go-to-Market Partner for AKD1500 Neuromorphic Processor