Challenges of Virtualization FPGA in a Cloud context

Rym Skhiri (1) , Virginie Fresse (2) , Jean-Paul Jamont (3) , Benoit Suffran (4)

1 EμE - Laboratoire d'Electronique et de Microélectronique [Monastir]
2 LHC - Laboratoire Hubert Curien
3 LCIS - Laboratoire de Conception et d'Intégration des Systèmes
4 ST-GRENOBLE - STMicroelectronics [Grenoble]

Field Programmable Gate Array (FPGAs) have been known as high efficient devices providing architectural flexibility, reconfigurability and high timing performances for embedded real time systems. For highly critical applications, FPGA provides low power consumption, significant resource optimization and can meet the real time constraints. Data acquisition systems and many measurement systems for industrial applications are based on standalone platforms integrating FPGA. Now the FPGA are considered in cloud computing for which the concept of “cloud FPGA” has been emerging over the last few years. Most of the time cloud FPGA is getting hold of the cloud computing infrastructure. Cloud FPGA leads to a set of challenges to be solved and clarifying what is the Cloud FPGA and which types of cloud FPGA exist should be detailed. In this context, this paper summarizes the principle of FPGA computing and the associated design flow on mono and multi-FPGA is presented. All challenges existing for each step of the design flow are detailed. The objective of this paper is to review all challenges in the cloud FPGA context that must be considered when proposing a cloud FPGA infrastructure.

Mots clés: FPGA design; challenges; cloud FPGA; mono FPGA; multi FPGA

Domaines: Systèmes embarqués

Rym Skhiri, Virginie Fresse, Jean-Paul Jamont, Benoit Suffran. Challenges of Virtualization FPGA in a Cloud context. international Conference on Computational Intelligence and Virtual Environments for Measurement Systems and Applications, Jun 2017, Annecy, France. ⟨hal-01563835⟩