High Level Synthesis as a service

Emerging cloud computing has improved computing performance and optimized usage of resources. One of the other elements that can improve computing performance is using acceleration resources. FPGA is a computing resource that is attractive attention due to its functionalities in many processes. High level programs which are designed for execution on CPU are not able to execute on FPGA and should be converted to HDL representation that is executable with FPGA. Many programmers do not have programming skill of HDL languages but the trend shows to using FPGA advantages or comparison and analysis their program execution on different computing resources. High Level Synthesis (HLS) is a tool to translate high level language representation to HDL language. In this paper, high level synthesis is proposed as a service in cloud computing infrastructure. User can order his requested service by OpenStack platform and interconnect to HLS as a Service (HLSaaS) via transparent and direct modes. The strong point of this paper is executing the High Level Synthesis as a Service (HLSaaS) for the first time in real environment of cloud computing infrastructure of the Razi Campus. It is useful in providing the services to users. The experiments are evaluated by CHStone benchmark. The evaluation of results shows that VMs' resources do not have the significant effect in HLS elapsed time. This issue shows weakness of current HLS tools.