# Cnvlutin2: Ineffectual-Activation-and-Weight-Free Deep Neural Network Computing

arXiv: https://arxiv.org/abs/1705.00125 · HTML: https://mlsanity.com/paper/1705.00125/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 29 Apr 2017.

## Built on by (most cited first, 9 of 9)

- [Pruning and Quantization for Deep Neural Network Acceleration: A Survey](https://mlsanity.com/paper/2101.09671/) (2021)
- [DistriFusion: Distributed Parallel Inference for High-Resolution Diffusion Models](https://mlsanity.com/paper/2402.19481/) (2024)
- [Efficient Spatially Sparse Inference for Conditional GANs and Diffusion Models](https://mlsanity.com/paper/2211.02048/) (2022)
- [Energy-Latency Attacks via Sponge Poisoning](https://mlsanity.com/paper/2203.08147/) (2022)
- [PFA: Privacy-preserving Federated Adaptation for Effective Model Personalization](https://mlsanity.com/paper/2103.01548/) (2021)
- [Accelerating Text-to-Image Editing via Cache-Enabled Sparse Diffusion Inference](https://mlsanity.com/paper/2305.17423/) (2023)
- [SGCN: Exploiting Compressed-Sparse Features in Deep Graph Convolutional Network Accelerators](https://mlsanity.com/paper/2301.10388/) (2023)
- [Slice-and-Forge: Making Better Use of Caches for Graph Convolutional Network Accelerators](https://mlsanity.com/paper/2301.09813/) (2023)
- [Mixture-of-Rookies: Saving DNN Computations by Predicting ReLU Outputs](https://mlsanity.com/paper/2202.04990/) (2022)

## Abstract (arXiv)

We discuss several modifications and extensions over the previous proposed Cnvlutin (CNV) accelerator for convolutional and fully-connected layers of Deep Learning Network. We first describe different encodings of the activations that are deemed ineffectual. The encodings have different memory overhead and energy characteristics. We propose using a level of indirection when accessing activations from memory to reduce their memory footprint by storing only the effectual activations. We also present a modified organization that detects the activations that are deemed as ineffectual while fetching them from memory. This is different than the original design that instead detected them at the output of the preceding layer. Finally, we present an extended CNV that can also skip ineffectual weights.