Ahmad Lashgar, PhD

Home | Resume | Software | X | Blog | Tools

Follow @lashgar

Short bio: I received my PhD in Computer Engineering at University of Victoria, BC, Canada in August 2017. For completing my PhD, I worked on developing hardware/software optimizations for accelerators. I developed IPMACC framework for translating OpenACC applications to CUDA, OpenCL, and ISPC backends. The framework is publically available at GitHub. Currently, I am working as Senior Software Engineering at Zoox. We are building the world's first robotaxi designed from ground-up for riders.

Contact: lashgar [at] uvic.ca

Latest from me:

Education: PhD in Electrical and Computer Engineering January 2014 to August 2017 Electrical and Computer Engineering Department University of Victoria Victoria, BC, Canada Supervisor: Amirali Baniasadi Thesis: Hardware/Software Optimizations for Accelerators Supervisory Committee Members: Nikitas Dimopoulos, Mihai SIMA, Alex Thomo, and Xipeng Shen [Dissertation, Slides] M.S. in Computer Architecture September 2010 to November 2012 School of Electrical and Computer Engineering University of Tehran Tehran, Iran Thesis: Improving GPU Performance with Improved SIMD Efficiency Supervisors: Ahmad Khonsari and Amirali Baniasadi Approvals: Hamid Sarbazi-Azad, Sied Mehdi Fakhraie, and Saeed Safari [Report (in persian), Slides] more on education B.S. in Computer Engineering - Hardware September 2006 to June 2010 Jundi Shapor University of Technology Dezful, Iran Thesis: GPU architecture and comparision with CPU Architecture (بررسي معماري هاي پردازنده گرافيكي و مقايسه آنها با پردازنده مركزي) [Report (in persian), Slides] Supervisor: Mojtaba Soltani Diploma in Math and Science Alborz High School Tehran, Iran

Publications:

1. Ahmad Lashgar, Ehsan Atoofian, and Amirali Baniasadi
   Loop Perforation in OpenACC
   16th IEEE International Symposium on Parallel and Distributed Processing with Applications (ISPA2018)
   Melbourne, Australia, 2018, pp. 163-170.
   [Paper/Final version]
   
2. Ahmad Lashgar and Amirali Baniasadi
   Efficient Implementation of OpenACC cache Directive on NVIDIA GPUs
   International Journal of High Performance Computing and Networking (IJHPCN), Special Issue on High-level Programming Approaches for Accelerators, Vol. 13, No. 1, 2019
   [Paper/Code/Final version]
   
3. Ahmad Lashgar, Ehsan Atoofian, and Amirali Baniasadi
   TELEPORT: Hardware/Software Alternative To CUDA Shared Memory Programming
   Microprocessors and Microsystems: Embedded Hardware Design (MICPRO), Volume 63, 2018, Pages 169-181, ISSN 0141-9331
   [Paper/Final version]
   
4. Ahmad Lashgar and Amirali Baniasadi
   OpenACC cache Directive: Opportunities and Optimizations
   In proceedings of Third Workshop on Accelerator Programming Using Directives (WACCPD2016), (in conjunction with SC2016)
   Salt Lake City, Utah, USA, November 14, 2016.
   [Paper/Slides/Code/Final version]
   
5. Ebad Salehi, Ahmad Lashgar, and Amirali Baniasadi
   Employing Compression Solutions under OpenACC
   In proceedings of 21st International Workshop on High-Level Parallel Programming Models and Supportive Environments (HIPC2016), (in conjunction with IPDPS2016)
   Chicago, Illinois, USA, May 23-27, 2016.
   [Accepted version/Final version/Slides/Code]
   
6. Ahmad Lashgar, Ebad Salehi, and Amirali Baniasadi
   A Case Study in Reverse Engineering GPGPUs: Outstanding Memory Handling Resources
   ACM SIGARCH Computer Architecture News - HEART '15, Volume 43 Issue 4
   [Accepted version/Final version/Code]
7. Ahmad Lashgar and Amirali Baniasadi
   Employing Software-Managed Caches in OpenACC: Opportunities and Benefits
   ACM Transactions on Modeling and Performance Evaluation of Computing Systems (ToMPECS), Volume 1 Issue 1, March 2016
   [Accepted version/Final version]
8. Ahmad Lashgar and Amirali Baniasadi
   Rethinking Prefetching in GPGPUs: Exploiting Unique Opportunities
   In proceedings of 17th IEEE International Conference on High Performance Computing and Communications (HPCC)
   New York, NY, USA, August 24-26, 2015.
   [Paper/Slides/Final version]
9. Ahmad Lashgar, Ebad Salehi, and Amirali Baniasadi
   Understanding Outstanding Memory Request Handling Resources in GPGPUs
   In proceedings of The Sixth International Symposium on Highly Efficient Accelerators and Reconfigurable Technologies (HEART)
   Best Paper Candidate
   Boston MA, USA, June 1-2, 2015.
   [Paper/Slides/Code]
10. Ahmad Lashgar, Alireza Majidi, and Amirali Baniasadi
    IPMACC: Translating OpenACC API to OpenCL
    In poster session of the 3rd International Workshop on OpenCL (IWOCL)
    Stanford University, California, USA, May 11-13, 2015.
    [Accepted version/Paper/Poster]
11. Ahmad Lashgar and Amirali Baniasadi
    A Case Against Small Data Types on GPGPUs
    The 25th IEEE International Conference on Application-specific Systems, Architectures and Processors (ASAP)
    IBM Research, Zurich, Switzerland, June 18-20, 2014.
    [Paper/Slides/Final version]
12. HARP: Harnessing Inactive Threads in Many-Core Processors
    
    Ahmad Lashgar, Ahmad Khonsari, and Amirali Baniasadi
    ACM Transactions on Embedded Computing Systems (TECS) - Special Issue on Design Challenges for Many-Core Processors (DCMP), Special Section on ESTIMedia'13 and Regular Papers, Volume 13 Issue 3s
    
    [Accepted version/Final version]
13. Towards Green GPUs: Warp Size Impact Analysis
    
    Ahmad Lashgar, Amirali Baniasadi, and Ahmad Khonsari
    4th International Green Computing Conference (IGCC 2013) WorkInProgress Session
    Arlington, VA, USA, June 28, 2013.
    
    [Paper/Slides]
14. Warp Size Impact in GPUs: Large or Small?
    
    Ahmad Lashgar, Amirali Baniasadi, and Ahmad Khonsari
    Sixth Workshop on General Purpose Processing Using GPUs (GPGPU6), (Held with ASPLOS XVIII)
    Houston, TX, USA, March 16, 2013.
    
    [Paper/Slides/Previous Technical Report]
15. Inter-Warp Instruction Temporal Locality in Deep-Multithreaded GPUs
    
    Ahmad Lashgar, Amirali Baniasadi, and Ahmad Khonsari
    26th International Conference on Architecture of Computing Systems (ARCS 2013)
    Prague, Czech Republic, February 19, 2013.
    
    [Paper/Slides/Code]
16. Dynamic Warp Resizing: Analysis and Benefits in High-Performance SIMT
    
    Ahmad Lashgar, Amirali Baniasadi, and Ahmad Khonsari
    The 30th IEEE International Conference on Computer Design (ICCD 2012) Poster Session
    Montreal, Quebec, Canada, October 1, 2012.
    
    [Paper/Extended version/Poster]
17. Performance in GPU Architectures: Potentials and Distances
    
    Ahmad Lashgar and Amirali Baniasadi
    9th Annual Workshop on Duplicating, Deconstructing, and Debunking (WDDD 2011), (in conjunction with ISCA 2011)
    San Jose, CA, USA, June 5, 2011.
    
    [Paper/Slides]

Software:

• IPMACC: OpenACC Compiler for CUDA/OpenCL backends
  C/C++/OpenCL/CUDA/Python/Bash
  December 2014, Code
• Globe Now: Find local events
  Java/Python/SQL/Json
  December 2017, Android App
• Accelerating InSAR Partner Pixel Coherence Algorithm using OpenACC/CUDA
  C/C++/OpenACC/CUDA
  November 2016, Source Code/Test Dataset (294MB)/Mitacs
• Bufferless CHIPPER-like router simulator for 2D-mesh
  C++
  February 2012, Code
• DevEff: Estimating the development effort of a C code
  Python
  March 2016, Code
• Accelerating Image Processing Kernels
  CUDA/C/C++
  January 2012, Noise Reduction (Median Filter)/Edge Detection (Sobel Filter)
  

Side Projects:

• GlobeNow! Find What Is Happening In A City Right Now!
  [Paper/Slides/Code]
  Latest: service shutdown as of April 6, 2020 for not being profitable.
  Idea: Have you ever been bored and needed an app to tell you what events are happening around you right now? Ambitiously, this is the goal of this project! I started with the towns I'm living in, e.g. Victoria and Vancouver, and expanded the project to 16 major cities in north America so far, e.g. New York and Los Angeles. Latest services and more information are available at http://www.twitter.com/glbnow
  
• PromoteBud! Commercial Service for Advertising Certain Product/Service
  Idea: PromoteBud! is a collection of tools to promote certain product/service via social media. Check the website to see how it works. For now, it supports Twitter and Telegram. Instagram is coming up next.
  

Honors & Awards:

• 36th recipient of CUDA Bronze Badge at StackOverflow
  January 2017
  
• Selected as the recipient of Alexander and Helen Stafford MacCarthy Muir Graduate Scholarship donor award by University of Victoria Graduate Awards Committee
  November 2015
  
• Selected as the recipient of Jarmila Vasta Von Drak Thouvenelle Graduate Scholarships donor award by University of Victoria Graduate Awards Committee
  November 2015
  
• Recipient of UVic Graduate Award
  January 2015
  
• Recipient of UVic Graduate Award
  January 2014
  
• Ranked 6 in the international MEMOCODE 2012 Design Contest (Sequence Alignment). [Contest Report]