In this video, SDSC Director Michael Norman show us the IOPSoMeter at SC11. SDSC’s revolutionary Gordon flash-based supercomputer hit an unprecedented 35 Million IOPs recently, demonstrating the advantages of high-bandwidth, low-latency SSD storage.
Industry Perspectives
Stop Building Models, Start Training Data
In this special guest feature, Sanjay Pichaiah, VP of Product Growth at Akridata, highlights why it is time for data scientists to stop building models and start training data. The path to better models and greater model accuracy doesn’t lie exclusively with the model, even though that has been the greatest focus in recent years. To truly accelerate and increase model performance, we need to be focusing more on the training data sets we are supplying the models and stop hoping the data is good enough.
[READ MORE...]White Papers
Presto: Open Source for the Enterprise
Presto addresses a real need for a portable SQL on Hadoop tool. It is architected from the ground up for high performance interactive query processing. Open source is a fount of continual innovation, especially with regard to big data. In addition, there are strong tools that come with specific Hadoop distributions. The fact is that organizations will deploy multiple tools. For organizations moving toward a Unified Data Architecture, the rationale for adopting Presto is even stronger.
Copyright © 2022
Leave a Comment