Multi-FPGA Designs and Scaling of HPC Challenge Benchmarks via MPI and Circuit-switched Inter-FPGA Networks
نویسندگان
چکیده
While FPGA accelerator boards and their respective high-level design tools are maturing, there is still a lack of multi-FPGA applications, libraries, not least, benchmarks reference implementations towards sustained HPC usage these devices. As in the early days GPUs HPC, for workloads that can reasonably be decoupled into loosely coupled working sets, multi-accelerator support achieved by using standard communication interfaces like MPI on host side. However, performance productivity, some applications profit from tighter coupling accelerators. FPGAs offer unique opportunities here when extending dataflow characteristics to interfaces. In this work, we extend HPCC benchmark suite three missing particularly characterize or stress inter-device communication: b_eff, PTRANS, LINPACK. With all implemented current with Intel Xilinx FPGAs, established baseline performance. Additionally, communication-centric benchmarks, explored potential direct FPGA-to-FPGA circuit-switched inter-FPGA network currently only available one boards. The evaluation parallel execution up 26 makes use largest academic installations.
منابع مشابه
A Survey of FPGA Benchmarks
New markets are emerging for the fast growing field-programmable gate array (FPGA) industry. Standard and fair benchmarking practices are necessary to evaluate FPGA systems and determine their potential to support target applications. This paper provides an extensive survey of FPGA benchmarks in both academia and industry.
متن کاملParallelizing FPGA Circuit Placement
Increasing the performance of uniprocessor systems is becoming increasingly difficult. As a result, processor systems are moving towards chip multiprocessor designs. Because of this trend, parallel programming design is becoming increasingly important. This presents new issues for complex software optimized for uniprocessor performance. CAD tools for placing and routing of FPGA designs are an e...
متن کاملImmuno-repairing of FPGA designs
FPGAs can be used for the design of autonomic reliable systems. Advantages are reconfiguration and flexibility in the design. However commercial FPGAs are first prone to errors. Second, the design flow is not yet supported for the use of fault tolerance techniques like Built-In Self-Tests. Fault tolerance can be reached through error detection and fault recovery. Most error detection techniques...
متن کاملEnergy Profiling and Analysis of the HPC Challenge Benchmarks
Future high performance systems must use energy efficiently to achieve PFLOPS computational speeds and beyond. To address this challenge, we must first understand the power and energy characteristics of high performance computing applications. In this paper, we use a power-performance profiling framework called PowerPack to study the power and energy profiles of the HPC Challenge benchmarks. We...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: ACM Transactions on Reconfigurable Technology and Systems
سال: 2023
ISSN: ['1936-7414', '1936-7406']
DOI: https://doi.org/10.1145/3576200