AXI HyperConnect: A Predictable, Hypervisor-level Interconnect for Hardware Accelerators in FPGA SoC

FPGA-based system-on-chips (SoC) are powerful computing platforms to implement mixed-criticality systems that require both multiprocessing and hardware acceleration. Virtualization via hypervisor technologies is, de-facto, an effective technique to allow the co-existence of multiple execution domains with different criticality levels in isolation upon the same platform. Implementing such technologies on FPGA-based SoC poses new challenges: one of such is the isolation of hardware accelerators deployed on the FPGA fabric that belong to different domains but share common resources such as a memory bus. This paper proposes AXI HyperConnect, a hypervisor-level hardware component that allows interconnecting hardware accelerators to the same bus while ensuring isolation and predictability features. AXI HyperConnect has been implemented on modern FPGA-SoC by Xilinx and tested with real-world accelerators, including one for Deep Neural Network inference.

[1]  Marco Pagani,et al.  Is Your Bus Arbiter Really Fair? Restoring Fairness in AXI Interconnects for FPGA SoCs , 2019, ACM Trans. Embed. Comput. Syst..

[2]  Giorgio Buttazzo,et al.  Safely Preventing Unbounded Delays During Bus Transactions in FPGA-based SoC , 2020, 2020 IEEE 28th Annual International Symposium on Field-Programmable Custom Computing Machines (FCCM).

[3]  Chita R. Das,et al.  A hybrid SoC interconnect with dynamic TDMA-based transaction-less buses and on-chip networks , 2006, 19th International Conference on VLSI Design held jointly with 5th International Conference on Embedded Systems Design (VLSID'06).

[4]  Ganesh Lakshminarayana,et al.  The LOTTERYBUS on-chip communication architecture , 2006, IEEE Transactions on Very Large Scale Integration (VLSI) Systems.

[5]  Yu-Jung Huang,et al.  A reconfigurable arbiter for SOC applications , 2008, APCCAS 2008 - 2008 IEEE Asia Pacific Conference on Circuits and Systems.

[6]  M. Masmano,et al.  XtratuM: a Hypervisor for Safety Critical Embedded Systems , 2012 .

[7]  Luca Benini,et al.  Adaptive TDMA bus allocation and elastic scheduling: A unified approach for enhancing robustness in multi-core RT systems , 2010, 2010 IEEE International Conference on Computer Design.

[8]  Douglas L. Maskell,et al.  Virtualized Execution and Management of Hardware Tasks on a Hybrid ARM-FPGA Platform , 2014, J. Signal Process. Syst..

[9]  Giorgio C. Buttazzo,et al.  Modeling and Analysis of Bus Contention for Hardware Accelerators in FPGA SoCs , 2020, ECRTS.

[10]  Giuseppe Lipari,et al.  A Bandwidth Reservation Mechanism for AXI-Based Hardware Accelerators on FPGAs , 2019, ECRTS.

[11]  Jürgen Teich,et al.  Runtime Reconfigurable Bus Arbitration for Concurrent Applications on Heterogeneous MPSoC Architectures , 2014, 2014 17th Euromicro Conference on Digital System Design.

[12]  Tian Xia,et al.  Ker-ONE: A new hypervisor managing FPGA reconfigurable accelerators , 2019, J. Syst. Archit..

[13]  Alois Knoll,et al.  Priority division: A high-speed shared-memory bus arbitration with bounded latency , 2011, 2011 Design, Automation & Test in Europe.