NOWLAB :: Publications

Journals
Conferences &
Workshops

Journals (4)
1	T. Tran, B. Ramesh, B. Michalowicz, M. Abduljabbar, H. Subramoni, A. Shafi, and DK Panda, Accelerating Communication with Multi-HCA Aware Collectives in MPI, Concurrency and Computation: Practice and Experience (CCPE), July 2023,
2	K. Suresh, K. Khorassani, C. Chen, B. Ramesh, M. Abduljabbar, A. Shafi, H. Subramoni, and DK Panda, Network Assisted Non-Contiguous Transfers for GPU-Aware MPI Libraries, IEEE Micro, Jan 2023.
3	K. Khorassani, C. Chen, B. Ramesh, A. Shafi, H. Subramoni, and DK Panda, High Performance MPI over the Slingshot Interconnect, Special Issue of Journal of Computer Science and Technology (JCST), Feb 2023.
4	A. Jain, N. Alnaasan, A. Shafi, H. Subramoni, and DK Panda, Optimizing Distributed DNN Training using CPUs and BlueField-2 DPUs, IEEE Micro, doi: 10.1109/MM.2021.3139027,

Conferences & Workshops (39)
1	PML-MPI: A Pre-Trained ML Framework for Efficient Collective Algorithm Selection in MPI M. Han, G. Kuncham, B. Michalowicz, R. Vaidya, M. Abduljabbar, A. Shafi, H. Subramoni, and DK Panda, iWAPT '24 (IPDPSW), May 2024 [Bib - Plain]
2	HINT: Designing Cache-Efficient MPI_Alltoall using Hybrid Memory Copy Ordering and Non-Temporal Instructions B. Ramesh, N. Contini, N. Alnaasan, K. Suresh, M. Abduljabbar, A. Shafi, H. Subramoni, and DK Panda, 38th IEEE International Parallel & Distributed Processing Symposium, May 2024 [Bib - Plain]
3	Exploiting Inter-Layer Expert Affinity for Accelerating Mixture-of-Experts Model Inference J. Yao, Q. Anthony, A. Shafi, H. Subramoni, and DK Panda, 38th IEEE International Parallel & Distributed Processing Symposium, May 2024 [Bib - Plain]
4	Accelerating MPI AllReduce Communication with Efficient GPU-Based Compression Schemes on Modern GPU Clusters Q. Zhou, B. Ramesh, A. Shafi, M. Abduljabbar, H. Subramoni, and DK Panda, ISC HIGH PERFORMANCE 2024, May 2024 [Bib - Plain]
5	Accelerating Large Language Model Training with Hybrid GPU-based Compression L. Xu, Q. Anthony, Q. Zhou, N. Alnaasan, R. Gulhane, A. Shafi, H. Subramoni, and DK Panda, IEEE/ACM International Symposium on Cluster, Cloud, and Internet Computing 2024, May 2024 [Bib - Plain]
6	Flover: A Temporal Fusion Framework for Efficient Autoregressive Model Parallel Inference J. Yao, N. Alnaasan, T. Chen, A. Shafi, H. Subramoni, and DK Panda, 30th IEEE INTERNATIONAL CONFERENCE ON HIGH PERFORMANCE COMPUTING, DATA, & ANALYTICS, Dec 2023 [Bib - Plain]
7	HARVEST: High-Performance Artificial Vision Framework for Expert Labeling using Semi-Supervised Training N. Alnaasan, M. Lieber, A. Shafi, H. Subramoni, S. Shearer, and DK Panda, 2023 IEEE International Conference on Big Data, Dec 2023 [Bib - Plain]
8	MPI4Spark Meets YARN: Enhancing MPI4Spark through YARN support for HPC K. Al Attar, A. Shafi, H. Subramoni, and DK Panda, 11th International Workshop on Distributed Storage and Blockchain Technologies for Big Data (IEEE Big Data '23), Dec 2023 [Bib - Plain]
9	Designing In-network Computing Aware Reduction Collectives in MPI B. Ramesh, G. Kuncham, K. Suresh, R. Vaidya, N. Alnaasan, M. Abduljabbar, A. Shafi, and DK Panda, Hot Interconnects 2023, Aug 2023 [Bib - Plain]
10	A Novel Framework for Efficient Offloading of Communication Operations to Bluefield SmartNICs K. Suresh, B. Michalowicz, B. Ramesh, N. Contini, J. Yao, S. Xu, A. Shafi, H. Subramoni, and DK Panda, 37th IEEE International Parallel & Distributed Processing Symposium (IPDPS '23), May 2023 [Bib - Plain]
11	MCR-DL: Mix-and-Match Communication Runtime for Deep Learning Q. Anthony, Ammar Awan, J. Rasley, Y. He, A. Shafi, M. Abduljabbar, H. Subramoni, and DK Panda, 37th IEEE International Parallel & Distributed Processing Symposium (IPDPS '23), May 2023 [Bib - Plain]
12	Accelerating Distributed Deep Learning Training with Compression Assisted Allgather and Reduce-Scatter Communication Q. Zhou, Q. Anthony, L. Xu, A. Shafi, M. Abduljabbar, H. Subramoni, and DK Panda, 37th IEEE International Parallel & Distributed Processing Symposium (IPDPS '23), May 2023 [Bib - Plain]
13	In-Depth Evaluation of a Lower-Level Direct-Verbs API on InfiniBand-based Clusters: Early Experiences B. Michalowicz, K. Suresh, B. Ramesh, A. Shafi, H. Subramoni, M. Abduljabbar, and DK Panda, 25th Workshop on Advances in Parallel and Distributed Computational Models, May 2023 [Held in conjunction with IPDPS 2023] [Bib - Plain]
14	ScaMP: Scalable Meta-Parallelism for Deep Learning Search Q. Anthony, L. Xu, A. Shafi, H. Subramoni, and DK Panda, THE 23RD IEEE/ACM International Symposium on Cluster, Cloud and Internet Computing, May 2023 [Bib - Plain]
15	Implementing and Optimizing a GPU-aware MPI Library for Intel GPUs: Early Experiences C. Chen, K. Khorassani, G. Kuncham, R. Vaidya, M. Abduljabbar, A. Shafi, H. Subramoni, and DK Panda, THE 23RD IEEE/ACM International Symposium on Cluster, Cloud and Internet Computing, May 2023 [Bib - Plain]
16	Performance Characterization of using Quantization for DNN Inference on Edge Devices H. Ahn, T. Chen, N. Alnaasan, A. Shafi, M. Abduljabbar, H. Subramoni, and DK Panda, 7TH IEEE INTERNATIONAL CONFERENCE ON FOG AND EDGE COMPUTING, May 2023 [Bib - Plain]
17	Designing Efficient Pipelined Communication Schemes using Compression in MPI Libraries B. Ramesh, Q. Zhou, A. Shafi, M. Abduljabbar, H. Subramoni, and DK Panda, 29th IEEE International Conference on High Performance Computing, Data, and Analytics, Dec 2022 [Bib - Plain]
18	Accelerating Broadcast Communication with GPU Compression for Deep Learning Workloads Q. Zhou, Q. Anthony, A. Shafi, H. Subramoni, and DK Panda, 29th IEEE International Conference on High Performance Computing, Data, and Analytics, Dec 2022 [Bib - Plain]
19	AccDP: Accelerated Data-Parallel Distributed DNN Training for Modern GPU-Based HPC Clusters N. Alnaasan, A. Jain, A. Shafi, H. Subramoni, and DK Panda, 29th IEEE International Conference on High Performance Computing, Data, and Analytics, Dec 2022 [Bib - Plain]
20	Efficient Personalized and Non-Personalized Alltoall Communication for Modern Multi-HCA GPU-Based Clusters K. Suresh, A. Paniraja Guptha, B. Michalowicz, B. Ramesh, M. Abduljabbar, A. Shafi, H. Subramoni, and DK Panda, 29th IEEE International Conference on High Performance Computing, Data, and Analytics, Dec 2022 [Bib - Plain]
21	Spark Meets MPI: Towards High-Performance Communication Framework for Spark using MPI K. Al Attar, A. Shafi, M. Abduljabbar, H. Subramoni, and DK Panda, IEEE Cluster '22, Sep 2022 [Bib - Plain]
22	Designing Hierarchical Multi-HCA Aware Allgather in MPI T. Tran, B. Michalowicz, B. Ramesh, H. Subramoni, A. Shafi, and DK Panda, Fifteenth International Workshop on Parallel Programming Models and Systems Software for High-End Computing (P2S2), 2022, Aug 2022 [To be held in conjunction with ICPP 2022: The 51st International Conference on Parallel Processing August 29th to Sept 1st, 2022 in Bordeaux, France] [Bib - Plain]
23	Network-Assisted Non-Contiguous Transfers for GPU-Aware MPI Libraries K. Suresh, K. Khorassani, C. Chen, B. Ramesh, M. Abduljabbar, A. Shafi, and DK Panda, Hot Interconnects 29, Aug 2022 [Bib - Plain]
24	High Performance MPI over the Slingshot Interconnect: Early Experiences K. Khorassani, C. Chen, B. Ramesh, A. Shafi, H. Subramoni, and DK Panda, Practice and Experience in Advanced Research Computing, Jul 2022 [Best Student Paper Award] [Bib - Plain]
25	Arm meets Cloud: A Case Study of MPI Library Performance on AWS Arm-based HPC Cloud with Elastic Fabric Adapter S. Xu, A. Shafi, H. Subramoni, and DK Panda, 24th Workshop on Advances in Parallel and Distributed Computational Models, May 2022 [Bib - Plain]
26	Towards Java-based HPC using the MVAPICH2 Library: Early Experiences K. Al Attar, A. Shafi, H. Subramoni, and DK Panda, HIPS '22 (IPDPSW), May 2022 [Bib - Plain]
27	Highly Efficient Alltoall and Alltoallv Communication Algorithms for GPU Systems C. Chen, K. Khorassani, Q. Anthony, A. Shafi, H. Subramoni, and DK Panda, Heterogeneity in Computing Workshop (HCW 2022), May 2022 [held in conjunction with IPDPS'22] [Bib - Plain]
28	OMB-Py: Python Micro-Benchmarks for Evaluating Performance of MPI Libraries on HPC Systems N. Alnaasan, A. Jain, A. Shafi, H. Subramoni, and DK Panda, 23rd Parallel and Distributed Scientific and Engineering Computing Workshop (PDSEC) at IPDPS22, May 2022 [Bib - Plain]
29	"Hey CAI" - Enhancing User Productivity through a Conversational AI Enabled User Interface for HPC Tools P. Kousha, A. Jain, A. Kolli, S. Prasanna, S. Miriyala, H. Subramoni, A. Shafi, and DK Panda, ISC HIGH PERFORMANCE 2022, May 2022 [Bib - Plain]
30	OMB-Py: Python Micro-Benchmarks for Evaluating Performance of MPI Libraries and Machine Learning Applications on HPC Systems N. Alnaasan, A. Jain, A. Shafi, H. Subramoni, and DK Panda, ISC HIGH PERFORMANCE 2022, May 2022 [Research Poster] [Best Poster Award] [Bib - Plain]
31	Accelerating MPI All-to-All Communication with Online Compression on Modern GPU Clusters Q. Zhou, P. Kousha, Q. Anthony, K. Khorassani, A. Shafi, H. Subramoni, and DK Panda, ISC HIGH PERFORMANCE 2022, May 2022 [Bib - Plain]
32	Hy-Fi: Hybrid Five-Dimensional Parallel DNN Training on High-Performance GPU Clusters A. Jain, A. Shafi, Q. Anthony, P. Kousha, H. Subramoni, and DK Panda, ISC HIGH PERFORMANCE 2022, May 2022 [Bib - Plain]
33	Layout aware Hardware assisted Designs for Derived Data Types in MPI K. Suresh, B. Ramesh, C. Chen, M. Ghazimirsaeed, M. Bayatpour, A. Shafi, H. Subramoni, and DK Panda, 28th IEEE International Conference on High Performance Computing, Data, Analytics, and Data Science, Dec 2021 [Bib - Plain]
34	Towards Architecture-aware Hierarchical Communication Trees on Modern HPC Systems B. Ramesh, J. Hashmi, S. Xu, A. Shafi, M. Ghazimirsaeed, M. Bayatpour, H. Subramoni, and DK Panda, 28th IEEE International Conference on High Performance Computing, Data, Analytics, and Data Science, Dec 2021 [Best Paper Finalist] [Bib - Plain]
35	Accelerating CPU-based Distributed DNN Training on Modern HPC Clusters using BlueField-2 DPUs A. Jain, N. Alnaasan, A. Shafi, H. Subramoni, and DK Panda, 28th IEEE Hot Interconnects, Aug 2021 [Bib - Plain]
36	INAM: Cross-stack Profiling and Analysis of Communication in MPI-based Applications P. Kousha, Kamal Raj Sankarapandian Dayala Ganesh Ram, M. Kedia, H. Subramoni, A. Jain, A. Shafi, DK Panda, Trey Dockendorf, Heechang Na, and K. Tomko, Practice and Experience in Advanced Research Computing 2021, Jul 2021 [Bib - Plain]
37	Efficient MPI-based Communication for GPU-Accelerated Dask Applications A. Shafi, J. Hashmi, H. Subramoni, and DK Panda, The 21st IEEE/ACM International Symposium on Cluster, Cloud and Internet Computing, May 2021 [Bib - Plain]
38	Blink: Towards Efficient RDMA-based Communication Coroutines for Parallel Python Applications A. Shafi, J. Hashmi, H. Subramoni, and DK Panda, 27TH IEEE INTERNATIONAL CONFERENCE ON HIGH PERFORMANCE COMPUTING, DATA, AND ANALYTICS, Dec 2020 [Bib - Plain]
39	Accelerating GPU-based Machine Learning in Python using MPI Library: A Case Study with MVAPICH2-GDR M. Ghazimirsaeed, Q. Anthony, A. Shafi, H. Subramoni, and DK Panda, 6th Workshop on Machine Learning in HPC Environments, Nov 2020 [Bib - Plain]

NOWLAB: Network Based Computing Lab

https://people.engineering.osu.edu/people/shafi.16

This page lists the publications by Aamir Shafi

Journals (4)

Conferences & Workshops (39)

PML-MPI: A Pre-Trained ML Framework for Efficient Collective Algorithm Selection in MPI

HINT: Designing Cache-Efficient MPI_Alltoall using Hybrid Memory Copy Ordering and Non-Temporal Instructions

Exploiting Inter-Layer Expert Affinity for Accelerating Mixture-of-Experts Model Inference

Accelerating MPI AllReduce Communication with Efficient GPU-Based Compression Schemes on Modern GPU Clusters

Accelerating Large Language Model Training with Hybrid GPU-based Compression

Flover: A Temporal Fusion Framework for Efficient Autoregressive Model Parallel Inference

HARVEST: High-Performance Artificial Vision Framework for Expert Labeling using Semi-Supervised Training

MPI4Spark Meets YARN: Enhancing MPI4Spark through YARN support for HPC

Designing In-network Computing Aware Reduction Collectives in MPI

A Novel Framework for Efficient Offloading of Communication Operations to Bluefield SmartNICs

MCR-DL: Mix-and-Match Communication Runtime for Deep Learning

Accelerating Distributed Deep Learning Training with Compression Assisted Allgather and Reduce-Scatter Communication

In-Depth Evaluation of a Lower-Level Direct-Verbs API on InfiniBand-based Clusters: Early Experiences

ScaMP: Scalable Meta-Parallelism for Deep Learning Search

Implementing and Optimizing a GPU-aware MPI Library for Intel GPUs: Early Experiences

Performance Characterization of using Quantization for DNN Inference on Edge Devices

Designing Efficient Pipelined Communication Schemes using Compression in MPI Libraries

Accelerating Broadcast Communication with GPU Compression for Deep Learning Workloads

AccDP: Accelerated Data-Parallel Distributed DNN Training for Modern GPU-Based HPC Clusters

Efficient Personalized and Non-Personalized Alltoall Communication for Modern Multi-HCA GPU-Based Clusters

Spark Meets MPI: Towards High-Performance Communication Framework for Spark using MPI

Designing Hierarchical Multi-HCA Aware Allgather in MPI

Network-Assisted Non-Contiguous Transfers for GPU-Aware MPI Libraries

High Performance MPI over the Slingshot Interconnect: Early Experiences

Arm meets Cloud: A Case Study of MPI Library Performance on AWS Arm-based HPC Cloud with Elastic Fabric Adapter

Towards Java-based HPC using the MVAPICH2 Library: Early Experiences

Highly Efficient Alltoall and Alltoallv Communication Algorithms for GPU Systems

OMB-Py: Python Micro-Benchmarks for Evaluating Performance of MPI Libraries on HPC Systems

"Hey CAI" - Enhancing User Productivity through a Conversational AI Enabled User Interface for HPC Tools

OMB-Py: Python Micro-Benchmarks for Evaluating Performance of MPI Libraries and Machine Learning Applications on HPC Systems

Accelerating MPI All-to-All Communication with Online Compression on Modern GPU Clusters

Hy-Fi: Hybrid Five-Dimensional Parallel DNN Training on High-Performance GPU Clusters

Layout aware Hardware assisted Designs for Derived Data Types in MPI

Towards Architecture-aware Hierarchical Communication Trees on Modern HPC Systems

Accelerating CPU-based Distributed DNN Training on Modern HPC Clusters using BlueField-2 DPUs

INAM: Cross-stack Profiling and Analysis of Communication in MPI-based Applications

Efficient MPI-based Communication for GPU-Accelerated Dask Applications

Blink: Towards Efficient RDMA-based Communication Coroutines for Parallel Python Applications

Accelerating GPU-based Machine Learning in Python using MPI Library: A Case Study with MVAPICH2-GDR