Abstract:This paper presents the implementation and evaluation of the H (hypervisor) extension for the RISC-V instruction set architecture (ISA) on top of the gem5 microarchitectural simulator. The RISC-V ISA, known for its simplicity and modularity, has seen widespread adoption in various computing domains. The H extension aims to enhance RISC-V's capabilities for cloud computing and virtualization. In this paper, we present the architectural integration of the H extension into gem5, an open-source, modular platform for computer system architecture research. We detail the modifications required in gem5's CPU models and virtualization support to accommodate the H extension. We also present evaluation results regarding the performance impact and functional correctness of the extension's implementation on gem5. This study not only provides a pathway for further research and development of RISC-V extensions but also contributes valuable insights into the optimization of the gem5 simulator for advanced architectural features.

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is to enhance the capabilities of RISC - V in cloud computing and virtualization by implementing the H (hypervisor) extension of the RISC - V Instruction Set Architecture (ISA) on the gem5 micro - architecture simulator. Specifically, the paper focuses on the following aspects: 1. **Implementation and Integration of H Extension**: - The paper describes in detail how to implement and integrate the H extension in the latest version (v24.0) of gem5, including the necessary modifications to the CPU model and virtualization support. - These modifications include adding new registers, exception and interrupt handling mechanisms, and support for Two - Stage Address Translation. 2. **Performance Evaluation and Functional Verification**: - The author evaluates the performance impact and functional correctness of the H extension on gem5, especially in the case of running virtual machines (VMs). - Through a series of benchmark tests, the paper shows the performance impact of the H extension on RISC - V in a virtualized environment and provides detailed performance data and analysis. 3. **Contributions to Research and Development**: - The paper not only provides a path for the research and development of RISC - V extensions but also offers valuable insights into the research on optimizing advanced architecture features in the gem5 simulator. - By means of open - source code, the paper encourages more researchers and developers to participate in the related work of RISC - V and gem5, promoting innovation and development in this field. ### Specific Problems and Their Solutions 1. **Virtualization Support**: - The H extension of RISC - V aims to improve the performance and security of virtualization and cloud computing through hardware support. The paper details how to implement this hardware support in gem5, including new registers, interrupt handling mechanisms, and memory management techniques. 2. **Performance Evaluation**: - To evaluate the performance impact of the H extension, the author uses nine workloads in the MiBench Suite for testing. The test results show that although virtualization will bring a certain performance overhead, the H extension can significantly reduce this overhead and improve the overall performance of the system. 3. **Functional Verification**: - The paper verifies the functional correctness of the H extension through a series of specialized tests, including interrupt handling, page faults, virtual instruction handling, etc. These tests ensure the correct implementation and stable operation of the H extension in gem5. ### Conclusions and Future Work - **Conclusions**: - The paper successfully implements the integration of the RISC - V H extension in the gem5 simulator and proves its effectiveness and reliability through detailed performance evaluation and functional verification. - This work provides important technical support for the application of RISC - V in the fields of cloud computing and virtualization and also lays the foundation for future innovation and research. - **Future Work**: - Expand support for all ISA - compatible virtual address sizes and all CPU types in gem5. - Integrate support for KVM (a Type - 2 hypervisor) to further enhance the application range and performance of gem5 in the virtualization field. Through these efforts, the paper not only solves the current problems of RISC - V in virtualization and cloud computing but also points out the direction for future development.

Advancing Cloud Computing Capabilities on gem5 by Implementing the RISC-V Hypervisor Extension

Hypervisor Extension for a RISC-V Processor

Full-stack evaluation of Machine Learning inference workloads for RISC-V systems

A First Look at RISC-V Virtualization from an Embedded Systems Perspective

CVA6 RISC-V Virtualization: Architecture, Microarchitecture, and Design Space Exploration

Gem5-AVX: Extension of the Gem5 Simulator to Support AVX Instruction Sets

HSP-V: Hypervisor-Less Static Partitioning for RISC-V COTS Platforms

A RISC-V Simulator and Benchmark Suite for Designing and Evaluating Vector Architectures

Preparing for HPC on RISC-V: Examining Vectorization and Distributed Performance of an Astrophyiscs Application with HPX and Kokkos

Gem5-X

RISC-V for HPC: Where we are and where we need to go

Simulation Environment with Customized RISC-V Instructions for Logic-in-Memory Architectures

Evaluating HPX and Kokkos on RISC-V using an Astrophysics Application Octo-Tiger

Supporting RISC-V Performance Counters through Performance analysis tools for Linux (Perf)

Validation, Verification, and Testing (VVT) of future RISC-V powered cloud infrastructures: the Vitamin-V Horizon Europe Project perspective

Test-driving RISC-V Vector hardware for HPC

Experiences of running an HPC RISC-V testbed

Co-designing ab initio electronic structure methods on a RISC-V vector architecture

gem5-NVDLA: A Simulation Framework for Compiling, Scheduling and Architecture Evaluation on AI System-on-Chips

A RISC-V-based FPGA Overlay to Simplify Embedded Accelerator Deployment