High-performance CUDA kernel execution on FPGAs

In this work, we propose a new FPGA design flow that combines the CUDA programming model from Nvidia with the state of the art high-level synthesis tool AutoPilot from AutoESL, to efficiently map the exposed parallelism in CUDA kernels onto reconfigurable devices. The use of the CUDA programming model offers the advantage of a common programming interface for exploiting parallelism on two very different types of accelerators – FPGAs and GPUs. Moreover, by leveraging the advanced synthesis capabilities of AutoPilot we enable efficient exploitation of the FPGA configurability for application specific acceleration. Our flow is based on a compilation process that transforms the SPMD CUDA thread blocks into high-concurrency AutoPilot-C code. We provide an overview of our CUDA-toFPGA flow and demonstrate the highly competitive performance of the generated multi-core accelerators. Categories and Subject Descriptors D.3.3 [Computer Systems Organization]: Performance of Systems– design stu...

• Type: Conference
• Year: 2009
• Where: ICS
• Authors: Alexandros Papakonstantinou, Karthik Gururaj, John A. Stratton, Deming Chen, Jason Cong, Wen-mei W. Hwu
• Keywords: CUDA Kernels, CUDA Programming Model, CUDA Thread Blocks, ICS 2009, Theoretical Computer Science