Jason Taylor, VP Infrastructure @Facebook, delivered a keynote describing their plan to enable a hardware ecosystem for machine learning based on their new Glow compiler initiative, along with the next steps, through partner support.
Read more about Glow here.
Watch the video here.
Glow is a community driven, open source framework (or compiler) allowing partners such as Esperanto Technologies to rapidly develop and optimize silicon products for #AI and #ML.
Esperanto fully supports the Glow initiative, which will allow us to provide optimized support for ML/DL frameworks such as PyTorch on our family of products.
Esperanto’s massively parallel, energy efficient products, with thousands of domain-specific RISC-V based acceleration cores, will benefit in performance and efficiency from the Glow compiler.
The entire Facebook ML stack running on Esperanto’s flexible, energy-efficient products will allow innovation from the research level as new ML/DL approaches and algorithms are created to production deployment in the data center and at the edge.
Also, the open source nature of the Glow initiative is a perfect fit given Esperanto’s DNA. We look forward to working with Facebook and the entire AI/ML community towards the goals outlined by Facebook for Glow.