2.基于RISC-V生态系统的计算机体系结构与AI加速器设计教学 .pdf

1、Teaching Computer Architecture and AI Accelerator Design through the RISC-V EcosystemSpeaker:Siting LiuSchool of Information Science and Technology(SIST)ShanghaiTech UniversityOutlineIntroductionBackground&challenges MotivationWhy RISC-V for teaching?Course material&the use of RISC-VComputer archite

2、ctureAI accelerator designTakeaways2BackgroundOverview:ICeducationatShanghaiTechPhysicsofSemiconductor;Intro.toNanoelectronics;OptoelectronicDevices;Micro/NanoProcessingTechnology;MicroelectronicDevices;etc.DevicesCircuitsDigital/Analog/RFIntegratedCircuits;DigitalVLSIDesignFlow;OptoelectronicDevice

3、s;Micro/NanoProcessingTechnology;etc.SystemsVLSIDesignAutomation;FPGA-basedHardwareSystemDesign;ChipTesting:FundamentalsandApplications;ComputerAidedVerification;ComputerArchitecture;AIComputingSystems;3ChallengesTraditional courses rely on proprietary ISAs(e.g.,x86/ARM),which fundamentally restrict

4、s architectural exploration and hands-on design experiences.It limits the chance that students learn from practice.RISC-V is open,simple yet elegant.4Teaching Process&Course Content5Knowledge transfer(lecture)Learning from practise(labs&projects)Computer architecture I and its projectBasic understan

5、ding of how a computer works;Memory hierarchy&memory management;Optimizations through parallelism;AI accelerator design with RISC-V extension(backward design)ISA extension and hardware implementation;Advanced hardware/software technologies for optimization;Memory hierarchy considerations;Basic under

6、standing on domain specific architecture(DSA);Basic Course Info.6Computer architecture I and its projectOne of the most important required course of an undergraduate SISTors;Consists of the theoretical part(4 credits)and the hands-on part(2 credits);Involves 200 students each year;Developed based on