I/O patterns in AI workflows and their impact on the performance of HPC simulations
The Dagstuhl Seminar on Integrating HPC, AI, and Workflows for Scientific Data Analysis held between Aug 27 and Sep 1st 2023 focused on understanding of AI+H...
Talks by Topics
AI workflows
How to efficiently abuse your HPC system in the age of AI workflows
I had the pleasure of being invited to give a talk to the Scientific Machine Learning Research Group at the Rutherford Appleton Laboratory. My talk focuses o...
A data-centric view on workflows that couple HPC with large-scale models
The first edition of the Workshop on Advancing Neural Network Training (WANT) held as part of thre NeurIPS workshop presents talk relevant to computationally...
Schema for the interactions between HPC, AI frameworks and visalization
The Open Data Schema Workshop held in conjunction with the Data-driven Plasma Science and Engineering Workshop in Oxford, UK takes a first step at establishi...
Efficient Digital Twin Training using Uncertainty-Guided Data Generation
I had the pleasure of giving an invited talk at the SIAM CSE25 workshop: Neural Acceleration, Surrogate Models, and Learning Techniques for HPC Kernels. The ...
Data management
ADIOS2 Kokkos Interoperability
My talk at the 2023 Kokkos User Group meeting presented early performance results on integrating Kokkos into the ADIOS I/O library.
Strategies for querying large-scale scientific data
My presentation focuses on efficient strategies for querying large datasets, specifically addressing quantities of interest and derived data. Although not d...
To Derive or Not to Derive: I/O Libraries Take Charge of Derived Quantities Computation
Online presentation for the IEEE 36th International Symposium on Computer Architecture and High Performance Computing (SBAC-PAD) conference. ORNL scientists ...
Optimizing Derived Data Computation with Kokkos for I/O-Bound Workloads
Scientific data analysis often involves complex queries across distributed datasets, requiring manipulation of multiple variables and generating derived data...
Schema for the interactions between HPC, AI frameworks and visalization
The Open Data Schema Workshop held in conjunction with the Data-driven Plasma Science and Engineering Workshop in Oxford, UK takes a first step at establishi...
ADIOS GPU backend and defining application QoIs using derived variables
The ADIOS User Group Meeting was a one day event attended by users from the US and Europe and has included talks about success stories using ADIOS in extreme...
Real time performance monitoring with ADIOS for Kokkos applications
Modern performance analysis tools are increasingly capable of capturing a high volume of metrics at ever-finer granularity. This abundance of information pre...
Priority-BF: a Task Manager for Priority-Based Scheduling
The increasing demand for computational resources, particularly in High-Performance Computing environments, necessitates to rethink how we handle job schedul...
HPC Applications
Scheduling the I/O of AI applications with a focus on medical imaging
My talk focuses on ways to build and automate AI workflows by separating the data and computational planes and attaching tasks to data offloading the I/O man...
To Derive or Not to Derive: I/O Libraries Take Charge of Derived Quantities Computation
Online presentation for the IEEE 36th International Symposium on Computer Architecture and High Performance Computing (SBAC-PAD) conference. ORNL scientists ...
Efficient Digital Twin Training using Uncertainty-Guided Data Generation
I had the pleasure of giving an invited talk at the SIAM CSE25 workshop: Neural Acceleration, Surrogate Models, and Learning Techniques for HPC Kernels. The ...
Performance
I/O patterns in AI workflows and their impact on the performance of HPC simulations
The Dagstuhl Seminar on Integrating HPC, AI, and Workflows for Scientific Data Analysis held between Aug 27 and Sep 1st 2023 focused on understanding of AI+H...
How to efficiently abuse your HPC system in the age of AI workflows
I had the pleasure of being invited to give a talk to the Scientific Machine Learning Research Group at the Rutherford Appleton Laboratory. My talk focuses o...
ADIOS2 Kokkos Interoperability
My talk at the 2023 Kokkos User Group meeting presented early performance results on integrating Kokkos into the ADIOS I/O library.
A data-centric view on workflows that couple HPC with large-scale models
The first edition of the Workshop on Advancing Neural Network Training (WANT) held as part of thre NeurIPS workshop presents talk relevant to computationally...
To Derive or Not to Derive: I/O Libraries Take Charge of Derived Quantities Computation
Online presentation for the IEEE 36th International Symposium on Computer Architecture and High Performance Computing (SBAC-PAD) conference. ORNL scientists ...
Optimizing Derived Data Computation with Kokkos for I/O-Bound Workloads
Scientific data analysis often involves complex queries across distributed datasets, requiring manipulation of multiple variables and generating derived data...
Real time performance monitoring with ADIOS for Kokkos applications
Modern performance analysis tools are increasingly capable of capturing a high volume of metrics at ever-finer granularity. This abundance of information pre...
Scheduling
Scheduling the I/O of AI applications with a focus on medical imaging
My talk focuses on ways to build and automate AI workflows by separating the data and computational planes and attaching tasks to data offloading the I/O man...
Strategies for querying large-scale scientific data
My presentation focuses on efficient strategies for querying large datasets, specifically addressing quantities of interest and derived data. Although not d...
On demand scheduling for large-scale workflows with analysis requirements
My talk for the 18th scheduling workshop workshop focuses on efficiently executing the analysis codes attached to large-scale simulations. These tasks requir...
Priority-BF: a Task Manager for Priority-Based Scheduling
The increasing demand for computational resources, particularly in High-Performance Computing environments, necessitates to rethink how we handle job schedul...
Tutorial
ADIOS-2: A Framework to Enable HPC Tools for Extreme Scale I/O, In Situ Visualization, and Performance Analysis
Full-day tutorial at SC23 focuses on building in situ and file-based data processing workflows for extreme scale systems, including interactive, on-demand, i...
Using the ADIOS library for handling data at extreme scales
Half-day tutorial as part of the CEA event Parallel IO and in situ analytics: High-performance data handling @ Exascale that showcases the must-know tools ...
ADIOS: An adaptable and scalable I/O framework for storage, in situ data processing and wide-area data transfer
This half-day tutorial provides a comprehensive introduction to building blocks of complex scientific workflows, including I/O, remote data access, and visua...