Stanford Engineering
Tambe Lab

Chip Gallery

Reasoning LLM Inference with Adaptive Block-Wise Mixed-Format 4-Bit Quantization and Decoupled Access-Execute

Tapeout details: TSMC 7nm, 2025

Publications: VLSI Symp'26

LLM Finetuning with 4T Embedded DRAM and 2D Block Floating-Point

Tapeout details: TSMC 16nm, 2023

Publications: CICC'26

Heterogeneous RISC-V Many-Accelerator SoC w/ Distributed Fine-Grained Power Management

Tapeout details: GF 12nm, 2022

Publications: ISSCC'24, JSSC'26

Transformer Inference with Adaptive Per-Vector FP4/FP8 and Early Exit

Tapeout details: GF 12nm, 2022

Publications: ISSCC'23, OJSSC'26

Attention-Based LSTM/RNN Inference w/ Adaptive Per-Tensor FP8

Tapeout details: TSMC 16nm, 2020

Publications: ISSCC'21, JSSC'22