Microsoft

LUT Tensor Core: Lookup Table Enables Efficient Low-Bit LLM Inference Acceleration

As large language model (LLM) inference demands ever-greater resources, there is a rapid growing trend of using low-bit weights to shrink memory usage and boost inference efficiency. However, these ...
IEEE

An 18.39 fJ/Conversion-Step 1-MS/s 12-bit SAR ADC With Non-Binary Multiple-LSB-Redundant and Non-Integer-and-Split-Capacitor DAC

Abstract: A low-power 12-bit successive approximation register (SAR) analog-to-digital converter (ADC) with split-capacitor, nonbinary-weighted, and multiple-least-significant-bit (LSB)-redundant ...
IEEE

A 32 Gb/s, 4.7 pJ/bit Optical Link With −11.7 dBm Sensitivity in 14-nm FinFET CMOS

Abstract: This paper presents a 32 Gb/s non-return-to-zero optical link using 850-nm vertical-cavity surface-emitting laser-based multi-mode optics with 14-nm bulk FinFET CMOS circuits. The target ...
GitHub

GPTQ-for-LLaMA

Quantization requires a large amount of CPU memory. However, the memory required can be reduced by using swap memory. Depending on the GPUs/drivers, there may be a difference in performance, which ...