ECP is ending

E in ECP refers to “Exascale”, exascale computing refers to the ability to perform one quintillion (10^18) calculations per second. This level of computing power is necessary for many scientific and industrial applications, such as simulating complex systems in fields like astrophysics, CFD, weather forecasting, energy, and medicine.

“zetta” scale computing seems too costly and unclear at this point, Moore’s is sort of not working after a long time tech review article, can GPUs come to rescue (maybe:)

ECP Webpage

Argonne is setting up 2 exaflops system AURORA

My pic with Aurora :O

Software plays a critical role in achieving exascale computing, as it is necessary to efficiently manage and utilize the massive amounts of data and resources involved. In the USA, the Department of Energy’s Exascale Computing Project (ECP) is working to develop the software, hardware, and infrastructure needed to achieve exascale computing capabilities.

Over the past few years, I am/was a part of the

1. ExaStar project: link

2. CANDLE project: Several researchers around the world cite our work on (Supervisor) hyper-parameter optimization of cancer deep learning as “state-of-art” in the field of machine learning for medical applications. link

3. Urban couple multiphysics (seed): Paper: “Representation and evolution of urban weather boundary conditions in downtown Chicago” link

This is the final annual meeting as the project ends this year. Detailed agenda is here

We Don’t Need Science Communication (And Other Lies We Tell Ourselves)

Paul Sutter presented a compelling case of why we need to write blogs and do more to communicate our science. Here is a picture from the talk:

Paul has a very interesting background, more about Paul: Podcast Bio

The future is somewhere in between: Applications specific to hardware (more research in hardware architectures is required) or Software tailored to a hardware (AI is driving new hardware changes)