DaVinci: A Scalable Architecture for Neural Network Computing

By H. Liao, Jiajin Tu, Jing Xia, Xiping Zhou. 2019

145 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

PanGu-α: Large-scale Autoregressive Pretrained Chinese Language Models with Auto-parallel Computation (2021, methodology): “imensional parallel functionalities with a carefully designed parallelization strategy and apply them to the largest PanGu-, which is efﬁciently trained on a cluster of 2048 Ascend 910 AI processors [12] and powered by CANN4. We train three PanGu-models on a high-quality 1.1TB Chinese text corpus wi”
VEGA: Towards an End-to-End Configurable AutoML Pipeline (2020, methodology): “VEGA supports multiple hardware platforms for model training or searching, such as CPU, GPU, Ascend (Liao et al., 2019) and multiple deep learning frameworks as back-ends, including Pytorch (Paszke et al., 2019), Tensorﬂow (Abadi et al., 2016) and MindSpore 1 .”
The Deep Learning Compiler: A Comprehensive Survey (2020, background): “Take the edge device for example, its computation units may consist of ARM CPU, Mail GPU, DSP, and probably NPU.”
Deep Learning Training in Facebook Data Centers: Design of Scale-up and Scale-out Systems (2020, methodology): “Each node processes the input data independently with replicas of parameters on each node, and aggregating the local parameter updates into a global update on all the nodes.”
A Survey on Vision Transformer (2020, background): “The performance determines whether the model can be applied on real-world applications, while the resource cost influences the deployment on devices [30], [136].”