Summary
Overview
Work History
Education
Skills
Websites
Teaching
Papers
Personal Information
Awards
Timeline
Generic

Pedram Alizadeh

5605-12 York St, Toronto,Canada

Summary

My main research area is high-performance communication for HPC and AI/ML/DL. Currently, I work as a Sr. software development engineer in AMD's ROCm Collective Communication Library (RCCL).

Overview

7
7
years of professional experience

Work History

Sr. SW Dev Eng., GPU Communication

AMD
09.2022 - Current

Contributions:

  • representing AMD at MPI forum and contributed to memory kinds document.
  • RCCL performance tuning (algo/proto) for mi300x
  • MSCCL performance tuning for various collectives for mi300x
  • Adding new test cases to RCCL Unittests and debugging CI failures
  • Various performance and functionality debugging for main costumers
  • Adding various Hip-mpi test cases
  • Delivering company-wide training sessions for RCCL and advanced memory optimization concepts and techniques
  • Delivering training sessions for multi-gpu, multi-node, rccl + mpi programming and profiling

Adjunct Lecturer

Queen's University
01.2022 - 04.2022
  • ELEC470 Computer System Architecture

Research Assistant

Queen's University
01.2018 - 01.2021
  • Efficient Process Arrival Pattern Aware Collective Communication for HPC and Deep Learning

Education

M.A.Sc. in Computer Engineering -

Queen's University
Kingston, ON
01.2021

B.Sc. in Electrical Engineering - Electronics

K.N.Toosi University of Technology
01.2017

Skills

  • C/C
  • MPI/UCC/UCX
  • CUDA/Hip
  • NCCL/RCCL
  • FPGAs/GPUs
  • Pthread/OpenMP
  • Profiling tools
  • AI/ML/DL
  • Linux

Teaching

  • Computer System Architecture- ELEC470, Queen's University, Kingston, ON, Canada, Adjunct Lecturer, Winter 2022
  • Digital Systems Engineering - ELEC374, Queen's University, Kingston, ON, Canada, Teaching Assistant, Winter 2019, Winter 2020

Papers

  • Yiltan H. Temucin, AmirHossein Sojoodi, Pedram Alizadeh, and Ahmad Afsahi. "Efficient Multi-Path NVLink/PCIe-Aware UCX based Collective Communication for Deep Learning." In 2021 IEEE Symposium on High-Performance Interconnects (HOTI), pp. 25-34. IEEE, 2021.
  • Yiltan H. Temucin, AmirHossein Sojoodi, Pedram Alizadeh, Benjamin W. Kitor, and Ahmad Afsahi, "Accelerating Deep Learning using Interconnect-Aware UCX Communication for MPI Collectives", IEEE Micro, 42(2):68-76, Mar-Apr 1, 2022.
  • Pedram Alizadeh, AmirHossein Sojoodi, Yiltan H. Temucin, and Ahmad Afsahi, "Efficient Process Arrival Pattern Aware Collective Communication for Deep Learning", 29th EuroMPI /USA Conference, University of Tennessee, Chattanooga, USA, Sept. 26-28, 2022.

Personal Information

Name: Pedram Surname: Mohammad Alizadeh Bakhtevari

Date of Birth: 04/18/94

Awards

  • Recipient of the Vector Scholarship in Artificial Intelligence
  • Recipient of the VP spotlight award for MI-300 project at AMD

Timeline

Sr. SW Dev Eng., GPU Communication

AMD
09.2022 - Current

Adjunct Lecturer

Queen's University
01.2022 - 04.2022

Research Assistant

Queen's University
01.2018 - 01.2021

M.A.Sc. in Computer Engineering -

Queen's University

B.Sc. in Electrical Engineering - Electronics

K.N.Toosi University of Technology
Pedram Alizadeh