Telamalloc: Efficient on-chip memory allocation for production machine learning accelerators
Memory buffer allocation for on-chip memories is a major challenge in modern machine
learning systems that target ML accelerators. In interactive systems such as mobile phones …
learning systems that target ML accelerators. In interactive systems such as mobile phones …
Archgym: An open-source gymnasium for machine learning assisted architecture design
Machine learning (ML) has become a prevalent approach to tame the complexity of design
space exploration for domain-specific architectures. While appealing, using ML for design …
space exploration for domain-specific architectures. While appealing, using ML for design …
GANDSE: Generative adversarial network-based design space exploration for neural network accelerator design
With the popularity of deep learning, the hardware implementation platform of deep learning
has received increasing interest. Unlike the general purpose devices, eg, CPU or GPU …
has received increasing interest. Unlike the general purpose devices, eg, CPU or GPU …
Tackling Resource Utilization in Deep Neural Network Accelerators
I Uwizeyimana - 2022 - search.proquest.com
Good resource utilization plays an important role in maximizing the performance of DNN
accelerators. One method to maximize resource utilization is to divide accelerator resources …
accelerators. One method to maximize resource utilization is to divide accelerator resources …
LayerDAG: A Layerwise Autoregressive Diffusion Model of Directed Acyclic Graphs for System
Directed acyclic graphs (DAGs) are ubiquitous in the design and optimizations of systems.
For example, neural networks have become a key computational workload for system …
For example, neural networks have become a key computational workload for system …
DOSA: One-Loop DSE for DNN Accelerators Using Differentiable Models
The process of hardware design space exploration requires both hardware parameters and
mappings from the algorithm onto the target hardware to be discovered and optimized …
mappings from the algorithm onto the target hardware to be discovered and optimized …