Rangan Sukumar

Just a curious data scientist...

[HPE Labs @HPE_labs]

In next week’s episode of @HPE_labs podcast “From Research to Reality,” @dejanm hosts five colleagues across @HPE: @paolofrb @Rangan_Sukumar Nic Dube, Arti Garg & Ken Leach. Together they present a distinctive point of view on heterogeneous computing. https://t.co/zTEEHZNG4G

[Rangan Sukumar @Rangan_Sukumar]

Quite excited to make an Exascale start on Instagram :).

[Argonne National Lab @argonne]

Set your reminders for 2:00 Central!

On Instagram (@argonne) Live we explore how Polaris, a new @HPE-built / Argonne-hosted supercomputer, is advancing the science of biological systems, climate change, and the secrets of the universe all while on the path to #exascale.

[Argonne National Lab @argonne]

Set your reminders for 2:00 Central!

On Instagram (@argonne) Live we explore how Polaris, a new @HPE-built / Argonne-hosted supercomputer, is advancing the science of biological systems, climate change, and the secrets of the universe all while on the path to #exascale.

[Rangan Sukumar @Rangan_Sukumar]

@argonne A11: A single #Exascale machine allows the scientist to imagine at scale - what can you find if you had the computational power of Google in your hands for a few hours? - better resolution, higher fidelity, newer insights, impactful discoveries

[Rangan Sukumar @Rangan_Sukumar]

@thebron @argonne That is #exascale getting into philosophy @thebron !

[Rangan Sukumar @Rangan_Sukumar]

@argonne A6: The AI models in science are orders of magnitude more compute hungry today. Imagine #exascale creating an AI model to analyze genomes, mapping genotypes and phenotypes of a billion people, creating personalized medicine based on the analysis...

[Rangan Sukumar @Rangan_Sukumar]

@argonne #Exascale is about the search for the what-is, what-if, what-else and the what-could-be. I need an #exascale computer to think all possible combinations of 100+ years of physics and 50+ years of real data coming back with answers in 15 minutes 😉

[Rangan Sukumar @Rangan_Sukumar]

@argonne A2: The use cases keep coming - @mike_woodacre and I wrote a recent blog on this topic about What is #exascale and what it could be ! We talk about use-cases from cosmology, microcopy, drug discovery, fintech here - https://t.co/l4ODInT6tD

[Rangan Sukumar @Rangan_Sukumar]

@argonne A1: #Exascale computers are now a Swiss-army knife for scientific workflows with high-performance solvers, data science and artificial intelligence tools.

[Rangan Sukumar @Rangan_Sukumar]

@argonne @Rangan_Sukumar from @HPE_Cray ! Excited to be here.

[Rangan Sukumar @Rangan_Sukumar]

@Carlos_I_Rojas @Dion2730 @HPE_Cray @NERSC #AskPerlmutter - Novel AI models are helping design innovative new materials/structures with desirable properties of strength, conductivity etc.

[Rangan Sukumar @Rangan_Sukumar]

@dch3njw @jan_fredin @HPE_Cray 8000 @doescience scientists with access to 6000 @nvidia GPUs ! That is bringing AI for science...! #AskPerlmutter

[Rangan Sukumar @Rangan_Sukumar]

@HPE_Cray @Rangan_Sukumar from the Chief Technology Office @HPE_Cray #AskPerlmutter

[Rangan Sukumar @Rangan_Sukumar]

@argonne The frameworks like #TensorFlow #PyTorch should be available to run and scale on #exascale computers, the challenge is going to be figuring out how to integrate them with HPC workflows with mixed precision and supporting the data pre-processing and I/O.

[Rangan Sukumar @Rangan_Sukumar]

@mike_woodacre @argonne @Google Exactly ! It is time to ask - What science can one learn at #Exascale by running PageRank/Deep-learning etc. on WWW size dataset if one is able to run a 50-1000x faster than possible.

[Rangan Sukumar @Rangan_Sukumar]

@argonne AI in biology and health is "augmented intelligence" - creating new molecules, capturing the knowledge of a million simulation from the past, searching through literature of 30+ million documents, etc. HPC offers "explainable intelligence" to the blind spots of AI.

[Rangan Sukumar @Rangan_Sukumar]

@argonne I can imagine #Exascale being proactive before a pandemic strikes - able to predict the genetic evolution of zoonotic viruses of today into the future, prioritize wet-lab research based on molecular-dynamics simulations, get prepared with a matrix of re-purposable drugs etc.

[Rangan Sukumar @Rangan_Sukumar]

@GhateVirendra @argonne @HPE we hear from scientists @doescience wanting to explore, integrate methods of HPC, data science and AI. The possibilities of simulations, streaming analysis and unstructured search over grids is what is driving different architectures from the edge-to-supercomputers.

[Rangan Sukumar @Rangan_Sukumar]

@argonne This is @Rangan_Sukumar from @HPE_Cray !

[Argonne National Lab @argonne]

Chemists ⚗️
Physicists ⚛️
Geologists 🌎
Weather and Climate scientists 🌀
Biologists 🐙
Neuroscientists 🧠
Molecular dynamics-ists 💃
Quantum scientists ▪️
even if you work in ML/AI 💻

You have one hour. Get ready to ask us your exascale-themed science questions.