Masoud Hemmatpour

I am Masoud Hemmatpour, a computer science researcher specializing in high-performance computing (HPC) interconnects. My research primarily focuses on emerging hardware technologies and optimizing memory access in distributed architectures to enhance computational efficiency.

You can find all available, ongoing, and completed theses and internships. Feel free to contact me if you’re interested in any of the topics.

Available thesis/internship

1. PCI Express support for BeeGFS

  • Description: The main goal of this master thesis is to add SuperSocket™️ functionality to ThinkParQs BeeGFS filesystem when using Dolphin’s PCIe NTB Fabric and to compare the performance with InfiniBand.
  • Link

Ongoing thesis

1. In-network Acceleration of Distributed In-Memory Machine Learning

  • Description: In this thesis, we are investigating SmartNIC which is equipped with an Arm processor to enhance the performance of the learning algorithms and get more insight of offloading task to SmartNIC.
  • Location: University of Oslo (UiO)
  • Link

2. In-network Acceleration of Big Data Processing

  • Description: In this thesis, we are investigating SmartNIC which is equipped with an Arm processor to enhance the performance of Bigdata processing/accessing.
  • Location: University of Oslo (UiO)
  • Link

3. Experimenting PCIe interconnect

  • Description: In this thesis we are experimenting PCIe networks and evaluating in specific applications.
  • Location: University of Oslo (UiO)

  • 4. Energy Efficiency of TinyML inference on tiny devices

  • Description: In this thesis, we are implementing ML models into Tiny devices with limited resources. The main goal of this thesis is to optimize ML model and reduce energy consumption of ML models on devices with resource constraint.
  • Location: University of Tromso

5. Path Planning ML Models for Autonomous UAV

  • Description: In this thesis, we are modeling the path of a real-world problem to reduce the energy consumption of a UAV under wind.
  • Location: University of Tromso

6. Path Planning ML Models for Autonomous UAVs to detect hot spot

  • Description: In this thesis, we are modeling the path of a UAV to detect hot spot.
  • Location: University of Tromso

Terminated

1. Architecting In-network memory access with FPGA-based SmartNIC for programmable data centers

  • Description: In this thesis, we evaluate the perfomance various memory designs on FPGA-based SmartNICs.
  • Location: University of Oslo

2. Evaluating BlueField-2 SmartNIC

  • Description: During this internship, we assessed the performance of the BlueField SmartNIC.
  • Location: Simula Research Laboratory

3. In-network monitoring of RDMA traffic

  • Description: During this internship, we investigated in-network monitoring strategies for tracking RDMA traffic in HPC clusters.
  • Location: Simula Research Laboratory
  • Link

4. Accelerating ResilientDB Fabric via RDMAs & SGX

  • Description: In this thesis, we investigated the use of RDMA to enhance the performance of an existing distributed transaction system.
  • Location: University of California Davis
  • Link

5. Effect of CPU scheduling on power consumption

  • Description: During this internship, we assessed the impact of the Linux scheduler on CPU power consumption.
  • Location: Politecnico di Torino
  • Link

6. Change scheduling priority according to cache miss rate

  • Description: This project involved modifying the Linux scheduler to incorporate cache miss rate as a parameter, allowing the system to penalize processes with high cache miss rates.
  • Location: Politecnico di Torino
  • Link

7. Microsoft Azure InfiniBand Configuration

  • Description: During this internship, we evaluated the InfiniBand infrastructure and RDMA communication on Microsoft Azure.
  • Location: Politecnico di Torino
  • Link

8. Fault injection on scheduler using FreeRTOS

  • Description: In this study, we injected faults into a real-time operating system and assessed their impact on the system’s behavior.
  • Location: Politecnico di Torino
  • Link