There are 2 repositories under approximate-computing topic.
phoeniX RISC-V Processor
Library of approximate arithmetic circuits
Generator of arithmetic circuits (multipliers, adders) and approximate circuits
Approximate layers - TensorFlow extension
Curated content for DNN approximation, acceleration ... with a focus on hardware accelerator and deployment
A library of VHDL components for Neural Networks
Condor allows for the specification of synopsis-based streaming jobs on top of general dataflow systems. Condor provides a collection of twelve synopses and an integration to Apache Flink.
Implementation of our proposed defense strategy against adversarial attacks "Defensive Approximation (DA)"
Code for the paper "Combining Gradients and Probabilities for Heterogeneours Approximation of Neural Networks"
GPU-accelerated Neural Network layers using Approximate Multiplications for PyTorch
Approximate Printed Machine Learning Classifiers
This repository contains approximate 8-bit multiplier Verilog code.
Carry Cut-Back Adder (CCBA) - An approximate adder circuit with artificially-built false timing paths
DeBAM : Decoder Based Approximate multiplier for Low Power Applications
GeAr low latency adder implementation and error analysis
https://arxiv.org/abs/1910.09876
Approximate Logic Synthesis of Random Forest classifiers
MATLAB and HDL models of ACA-CSU approximate adders
Repository accompanying the paper "Fixed-Posit: A Floating-Point Representation for Error-Resilient Applications" published in IEEE Transactions on Circuits and Systems II
Picorv32-IM with exact and approximate SIMD multiplication extensions. The SIMD modules can be accessed with custom RISC-V Instruction to perform dual/quad 8-bit multiplications.
AxQEMU: a Floating-Point Approximation-aware Emulator
Code for reproducing CoNGA 2023 results on approximate posit arithmetic