The document discusses Nimbix, a company that provides cloud computing services for high-performance computing (HPC) and artificial intelligence (AI) workloads. It describes Nimbix's history and infrastructure, including partnerships with IBM to provide IBM Power systems and GPUs. The document then explains concepts around AI, different types of AI, and how Nimbix's cloud is well-suited for AI tasks like research, analysis, algorithm development and training.
I’m technologist, not a scientist, but I design and enable technology for human beings, including scientists
I’m from Texas – we do actually have technology there
No platform out there to address this problem in an efficient way
Moore’s law as it applies to scale up
Data processing and analytics (big data) was en vogue at the time, but this gradually morphed into data science and AI
- 3 principles
- Self-service, on-demand Cloud to democratize
Mention 10% or so of the use cases on Summit will be AI
So just in case I make it to the end and someone asks “what’s AI”…
Admit this deep learning focused
Of course self driving cars never kill drivers nor hit pedestrians, that must just be a sick joke
A lot of cultural and socioeconomic anxiety about this
Self driving cars killing people aside, we’ll see examples of AI used to identify safety violations in work sites, predict equipment failures, etc.
- Choice, in a nutshell (can’t democratize without offering choice)
All these rival system architectures coexist and are user selectable at runtime for just about any workflow, or part of a broader workflow
E.g. you wouldn’t use FPGA’s for training neural networks, but they sure can help with low latency inference on real-time data
Edge may be self driving car, mobile device, etc.
In 2018 nothing interesting really happens without software, so…
If you can write a Dockerfile, you can customize any workflow or build your own from scratch
You can even deploy your custom workflows in the marketplace and make money off of them if you like (assuming someone finds them interesting enough to pay for)
Talk about 50% utilization on GPUs
Talk about GPU direct RDMA as well
Species identification indicates salinity and temperature and is used to study effects of climate change