Past Talks

GPU Technology Conference 2021 - Virtual, times in EDT (US)
(Apr 12 - 16, 2021)

Time Location Event Speaker(s)

Monday, April 12

1:00PM - 1:10PM

High Performance and Scalable All-to-All and Allgather for GPU Systems

[Poster Presentation]

K. Khorassani
1:00PM - 1:10PM

Scaling Distributed PyTorch and DeepSpeed with MVAPICH2-GDR

[Poster Presentation]

Q. Anthony
1:00PM - 1:10PM

Distributed DNN Training Using GPU-Enabled Model-Parallelism System (GEMS) with MVAPICH2-GDR

[Poster Presentation]

A. Jain
1:00PM - 1:10PM

Benefits of On-the-Fly Compression for Modern GPU Clusters with MVAPICH2-GDR

[Poster Presentation]

Q. Zhou
1:00PM - 1:20PM

Accelerating GPU-Enabled HPC and Data Science Applications with On-the-Fly Compression

[Talk]

DK Panda

Tuesday, April 13

10:00AM - 10:40AM

Introducing Cloud-Native Supercomputing: Bare-Metal, Secured Supercomputing Architecture

[Talk]

G. Shainer (NVIDIA)
P. Calleja (Univ. of Cambridge)
DK Panda
12:00PM - 12:40PM

Optimizing Communication on GPU-Based HPC Systems for Dask and cuML Using MVAPICH2-GDR

[Talk]

DK Panda
A. Shafi
2:00PM - 2:40PM

High-Performance Scalable Distributed Deep Learning with MVAPICH2-GDR

[Talk]

DK Panda
H. Subramoni

OpenFabrics Alliance Workshop 2021 - Virtual, times in PDT (US)
(Mar 15 - 19, 2021)

Time Location Event Speaker(s)

Monday, March 15

9:00AM - 9:30AM

Designing a ROCm-aware MPI Library for AMD GPUs over High-speed Networks

[Talk]

H. Subramoni

Tuesday, March 16

8:00AM - 8:30AM

RDMA Spark Meets OSU INAM: Performance Engineering Big Data Applications on HPC Clusters

[Talk]

A. Shafi

Thursday, March 18

9:00AM - 9:30AM

Efficient MPI Offloading Designs on Modern BlueField Smart NICs

[Talk]

DK Panda
10:00AM - 10:30AM

Designing a High-performance MPI library using In-Network Computing

[Talk]

H. Subramoni

SC 2020 - Atlanta, GA, USA
(Nov 15 - 20, 2020)

Time Location Event Speaker(s)

Monday, November 16

10:00AM - 5:00PM NVIDIA Virtual Theater

MVAPICH2 and In-Network Computing for HPC and Deep Learning

[Talk]

DK Panda

Tuesday, November 17

3:00PM - 3:30PM Track 5

Technical Paper: A Hierarchical and Load-Aware Design for Large Message Neighborhood Collectives

[Paper Presentation]

M. Ghazimirsaeed

Wednesday, November 18

11:00AM - 11:30AM Track 3

Technical Paper: GEMS: GPU-Enabled Memory-Aware Model-Parallelism System for Distributed DNN Training

[Paper Presentation]

A. Jain