The WHPC Executive Committee, the WHPC SC23 Volunteer Committee and the staff of DoITNow! have announced two Travel Fellows for this year’s Supercomptuing Conference being held in Denver. Application to these fellowships was offered to those who successfully submitted to speak during WHPC’s Early Career session during the organization’s workshop on November 13th.
“The resulting applications were truly spectacular with only narrow margins deciding this year’s recipients,” WHPC said in its announcement.
This year’s Travel Fellows: Babar Khan, representing Technical University Darmstadt, Germany, and Leslie Cook, from Midwestern State University, USA. Both recipients demonstrated exceptional technical expertise and shared compelling personal statements that embody the innovative and inclusive spirit that defines our field.
Babar Khan
Babar Khan
Babar Zaman Khan is a PhD candidate in the Computer Science department at Technical University Darmstadt, Germany. His current PhD research is conducted in collaboration with the industrial partner SAP, Germany, where he is part of the hardware team. Prior to his PhD, he earned his M.Sc. in Electrical Engineering from the University of Bremen, Germany, where he completed his thesis with the highest distinction in a joint partnership with Intel, Germany. The thesis focused on ASIC verification. He obtained his B.Sc. in Electrical Engineering from Lahore, Pakistan. His current PhD research primarily focuses on accelerating distributed storage through hardware accelerators such as FPGAs and SmartNICs. Additionally, he consistently contributes to open-source software and hardware projects. In his free time, he enjoys jogging, hiking, and cycling.
Khan’s talk at SC23 addresses the complexity of the HPC I/O stack, which grapples with the increasing demands of HPC workloads and adds a significant overhead along the entire I/O request path. For instance, our measurements have shown that it takes more than 30,000 instructions to send and receive a single 4KB I/O request. A novel hardware-software framework, DeLiBA (Development of Linux Block I/O Accelerators), introduced in the talk is a proposal to alleviate the complexity of HPC I/O stack for easier research. DeLiBA aims to bridge this gap by facilitating the development of software components within the I/O stack in user space, rather than the kernel space, and leverages a 16 nanometer (nm) FPGA to accelerate the I/Os with the help of FPGA-based I/O accelerators. As a first use-case for DeLiBA, a proof-of-concept of an I/O accelerator for the client side of the distributed storage protocol called Ceph is implemented. For final hardware evaluation, the client-side host in DeLiBA framework uses an AMD EPYC Rome 7302P 16-core CPU with 128GB of memory, attached by 10Gb/s Ethernet to the Ceph server. The FPGA card is attached to the client host by PCIe and uses a system clock of 200MHz. The initial results achieve a 10% increase in throughput and demonstrates up to 2.3 times the I/O operations per second compared to conventional methods. Above all, DeLiBA is an open-source framework, and it is regularly being improved to serve as a common platform for HPC I/O researchers.
Leslie Cook
Leslie Cook
Leslie holds a Bachelor’s degree in Chemistry and is on cusp of completing a Master’s in Computer Science. Her STEM journey was fortified by an internship at the NIH Biomedical Informatics Section, where she delved into the intersections of data science and healthcare. Furthermore, Leslie was awarded a SIGHPC Fellowship that ignited a desire to explore how research in HPC is moving the world forward. This passion for innovation led her to an appointment with Oak Ridge National Laboratory/Department of Energy, where she was immersed in the realm of HPC. Leslie is a tenacious advocate for women in STEM and her journey exemplifies the ability to overcome challenges to carve a path that bridges scientific prowess with unwavering commitment, all while ensuring the best for her family and future.
Leslie will be presenting, “Investigating Linear Solvers for Power Grid Analysis with Exascale Computing: A Journey of Learning and Collaboration”
This work is a contribution to the advancement of linear solvers for the Exascale Computing Project. It focuses on direct sparse linear solvers using High-Performance Computing (HPC) for large-scale power systems, resembling the United States power grids. This paper explores supercomputers at Oak Ridge Leadership Computing Facility, Summit and Frontier, comparing both performance and optimization strategies. The project encompasses a comprehensive test bench for Trilinos Amesos2 CPU-based solvers, KLU2 and ShyLUBasker, and the testing of GPU-based solvers from NVIDIA cuSolver to AMD rocSolver on distinct architecture configurations. The challenges of power flow analysis are addressed through optimization techniques, like matrix symmetry and GPU acceleration, and by evaluating accuracy and stability of linear solvers through residual analysis. Beyond technical gains, this work underscores the significance of collaboration and diverse expertise in HPC for innovative analysis of power grid systems, critical for resilient infrastructure against burgeoning threats like climate change and cyberattacks.
—
As we continue to champion diversity and excellence in high-performance computing, WHPC remains committed to creating opportunities and platforms that empower individuals to thrive and succeed. We invite you to join us in celebrating the accomplishments of our Travel Fellows and look forward to their impactful contributions at the SC23 conference in Denver, Colorado.
How our Travel Fellowships were decided: This fellowship was extended to all those who successfully applied to speak at our Early Career session during the 16th Annual Women in HPC Workshop at SC23, scheduled for November 13, 2023. Speakers required both a stellar talk submission and personal statement in order for consideration. In the interest of fairness all reviews for the fellowship were blind. Each of our award winners will receive up to £3,000 towards travel expenses, which will be reimbursed after the conference comes to a close.
About WHPC Champion, DoITNow!
Do IT, from Italy, HPCNow!, from Spain, and UCit, from France, came together to push the limits of High Performance Computing. Do IT Now has been created to be the market leader and to offer the best solutions to our clients. We share a passion and enthusiasm for facing new challenges of HPC technologies together. Do IT Now deal with the complexity to give simple solutions to scientists and engineers.
The real value we offer is a deep understanding of the most advanced technologies in HPC, along with high quality customer and user support. We offer solutions for different IT sectors like Big Data, artificial intelligence (AI), cloud computing or storage.
