基于高密度FPGA服务器的实验平台建设

Construction of an Experimental Platform Based on High-Density FPGA Servers

  • 摘要: 针对采用传统现场可编程门阵列(field programmable gate array, FPGA)实验设备的硬件实验教学中存在的设备资源受限、维护成本高及实验模式僵化等问题,该文提出基于高密度FPGA服务器集群的在线实践教学平台建设方案。该平台通过将数据采集模块集成到FPGA芯片内部,实现了单机架式服务器28块开发板的高密度集成,构建了支持多用户并发访问的云端实验环境。基于此平台,该文采用自顶向下的实验案例设计方法对计算机组成原理课程的实验案例进行了重新设计,并建立了基于黑盒测试的自动化评测体系。实验教学应用表明,该平台可支撑239名学生同步开展计算机组成原理全系列实验,设备复用率可提升至4:1。该方案形成了“硬件资源虚拟化、实验操作云端化、评测反馈实时化”三位一体的新型教学模式,打破了传统硬件实验对物理时空的刚性约束,实现FPGA实验的泛在化自主实践。

     

    Abstract: To address the challenges of limited equipment resources, high maintenance costs, and rigid experimental modes in hardware experimental teaching using traditional field programmable gate array (FPGA) experimental equipment, this paper proposes a construction scheme for an online practical teaching platform based on a high-density FPGA server cluster. By integrating the data acquisition module into the FPGA chip, this platform achieves high-density integration of 28 development boards in a single rack-mounted server, constructing a cloud-based experimental environment that supports concurrent access for multiple users. Building upon this platform, this paper adopts a top-down experimental case design method to redesign experimental cases for the Computer Organization Principles course, and establishes an automated evaluation system based on black-box testing. Practical teaching application demonstrates that the platform can support 239 students in simultaneously conducting the full series of Computer Organization Principles experiments, with the equipment reuse ratio increased to 4:1. This scheme forms a new teaching model integrating “hardware resource virtualization, cloud-based experimental operation, and real-time evaluation feedback,” thereby breaking the rigid constraints of physical time and space in traditional hardware experiments, and enabling ubiquitous autonomous practice in FPGA experiments.

     

/

返回文章
返回