AI-449-paul.wilmoth-2026-08-20-01-14-38
Below is a syllabus template that includes WSU's required syllabus elements. Please complete all items highlighted in yellow.
Title of Course [AI Hardware Accelerators and Edge AI]
Prefix and Number [AI 449]
Semester and Year [tbd]
Number of Credit Hours [3 credit hours]
Prerequisites [AI 260 or CPTS 260]
Course Details
Day and Time: [tbd]
Meeting Location: [tbd]
Instructor Contact Information
Instructor Name: [tbd]
Instructor Contact Information: [office location, phone, email] [tbd]
Instructor Office Hours: [click here for best practices] [tbd]
TA Name: [tbd]
TA Contact Information: [office location, phone, email]: [tbd]
TA Office Hours: [click here for best practices] [tbd]
Course Description
This course examines the architecture, design, and optimization of computing systems for the efficient execution and deployment of modern AI models. Students study AI workloads and deep neural-network computation, accelerator architectures, memory hierarchies, compute organization, dataflow, interconnects, and parallel processing across GPUs, TPUs/NPUs, FPGAs, and specialized AI accelerators.
Students analyze how architectural choices affect latency, throughput, area, energy efficiency, and overall system performance. They study and apply techniques for efficient AI inference, including quantization, pruning, sparsity, model compression, and hardware-software co-design.
Building on these foundations, students deploy and optimize representative AI models on FPGA, embedded, and resource-constrained edge platforms using contemporary software and hardware development tools. Students evaluate alternative accelerator and deployment strategies using performance, energy, accuracy, and resource-utilization measures and conduct systematic bottleneck and design-tradeoff analyses.
The course also addresses reliability, data privacy and security, sustainability, and the responsible deployment of edge-AI systems. A culminating project requires students to design, implement, optimize, evaluate, and communicate an accelerated or edge-AI application.
Course Materials
Books: No required textbook. Required readings, lecture notes, tutorials, and research papers will be provided through Canvas or made available through the WSU Libraries at no additional cost.
Other Materials: Students must have regular access to a computer capable of running course-provided AI-development, simulation, modeling, and performance-analysis tools. Required software will be free, open source, or provided through WSU. Access to any required FPGA, embedded, edge-computing, or remote-computing platforms will be provided through the course. Estimated additional cost: $0.
Fees: No additional course fees are anticipated.
|
Course Learning Outcomes (students will be able to:) |
Activities Supporting the Learning Outcomes | Assessment of the Learning Outcomes |
|---|---|---|
| Analyze the computational structure of AI workloads to determine their compute, dataflow, memory-bandwidth, and energy requirements (1b). | Lectures and readings; workload-profiling laboratory; dataflow exercises. | Problem set; profiling-laboratory report; examination questions. |
| Apply performance and energy models to identify bottlenecks and optimize data movement, numerical precision, and hardware-resource utilization (1e, 6c). | Roofline exercises; memory-hierarchy and mapping laboratory; optimization case study. | Modeling problem set; laboratory report; examination questions. |
| Evaluate CPUs, GPUs, NPUs/TPUs, FPGAs, and edge accelerators using latency, throughput, energy, accuracy, memory, and cost metrics (2e, 6b). |
Architecture case studies; accelerator-simulation laboratory; benchmark comparisons. |
Benchmark report; design critique; examination questions. |
|
Design and optimize an edge-AI inference pipeline using quantization, sparsity, mapping, and hardware-software co-design (1e, 2a, 2g, 6a). |
FPGA-based simulations, literature review of different methods. |
Working demonstration of different co-design methods, class presentations. |
|
Explore emerging accelerator paradigms, including compute-in-memory, neuromorphic, chiplet-based, and photonic systems, and assess their opportunities and limitations (6e, 7d, 7e). |
Research-paper discussions; student-led technical brief; emerging- technology comparison. |
Literature review; technical presentation; future-accelerator design memo. |
| Dates | Lesson Topic | Assignment | Assessment |
|---|---|---|---|
|
Week 1 |
Introduction to AI hardware accelerators and edge AI; review of computing systems for AI. | Development-environment setup and introductory AI-workload exercise. | Environment-setup verification and introductory exercise. |
| Week 2 [dates] |
AI workloads and deep neural-network computation; computational characteristics of fully connected, convolutional, and attention-based models. | AI-workload analysis. | Homework 1. |
| Week 3 [dates] |
Accelerator evaluation metrics, including latency, throughput, area, energy efficiency, memory use, accuracy, and resource utilization. | Workload-profiling laboratory and Roofline-analysis exercises. | Laboratory 1 report. |
| Week 4 [dates] |
Compute organization, processing-element arrays, spatial and temporal dataflows, and data reuse. | dataflow and workload-mapping analysis. | Homework 2. |
| Week 5 [dates] |
Accelerator memory hierarchies, on-chip storage, external memory, bandwidth limitations, and data-movement costs. | hierarchy and mapping laboratory. | Laboratory 2 report. |
| Week 6 [dates] |
Interconnect architectures, parallel processing, communication overhead, and scalability in AI accelerators. | Interconnect and parallel-processing design exercises. | Homework 3. |
| Week 7 [dates] |
GPU architectures and parallel execution of AI workloads. | GPU workload-mapping and performance-analysis exercise. | GPU analysis report. |
| Week 8 [dates] |
TPU, NPU, systolic-array, and other specialized AI-accelerator architectures. | Comparative accelerator-architecture case study. | Architecture comparison and examination review. |
| Week 9 [dates] |
FPGA-based AI acceleration, configurable computing, hardware-resource mapping, and design tradeoffs. |
FPGA-based simulation laboratory. |
Midterm examination |
| Week 10 [dates] |
Numerical precision, quantization, model compression, and accuracy-efficiency tradeoffs. | Quantization and model-compression laboratory. | Laboratory 3 report. |
| Week 11 [dates] |
Pruning, structured and unstructured sparsity, sparse execution, and efficient inference. | Homework 4 on sparse and compressed AI inference. | Homework 4. |
| Week 12 [dates] |
Hardware-software co-design, operator mapping, scheduling, and systematic bottleneck analysis. | Final-project proposal and hardware-software co-design exercise. |
Final-project proposal and design review. |
| Week 13 [dates] |
Deployment of AI models on embedded and resource-constrained edge platforms; performance, energy, accuracy, and resource-utilization analysis. | Edge-AI deployment and optimization laboratory. | Laboratory 4 report and Project Milestone 1. |
| Week 14 [dates] |
Emerging accelerator paradigms, including compute-in-memory, neuromorphic, chiplet-based, and photonic systems. | Research-paper review and emerging-accelerator technical brief. | Technical presentation, future-accelerator design memo, and Project Milestone 2. |
| Week 15 [dates] |
Reliability, data privacy and security, sustainability, responsible edge-AI deployment, and final-project presentations. | Final project demonstration, presentation, and report. | Working project demonstration, class presentation, and final report. |
Expectations for Student Effort
[Describe how much time students should expect to invest in the course each week. Graduate courses should state: "For each hour of lecture equivalent, students should expect to have a minimum of two hours of work outside of class." Note that Global campus courses will automatically include credit hour equivalents in the syllabus.]
AI 449 is a three-credit course. Students should expect approximately three hours of scheduled class activities and at least six hours of work outside class each week. Outside work includes reading assigned material, reviewing lecture content, completing problem sets, conducting laboratory experiments, using accelerator-analysis and development tools, preparing technical reports, and developing the final project.
Project-intensive weeks may require additional effort. Students should plan for an average total commitment of approximately nine hours per week.
Grading [add more lines if necessary]
| Type of Assignment (tests, papers, etc) | Points | Percent of Overall Grade |
|---|---|---|
| Homework and laboratories | 400 | 40% |
| Midterm examination | 250 | 25% |
| Final project, including the proposal, milestones, working demonstration, presentation, and final report | 350 | 35% |
| Grade | Percent | Grade | Percent |
|---|---|---|---|
| A |
93.0–100% |
C | 73.0–76.9% |
| A- | 90.0–92.9% | C- | 70.0–72.9% |
| B+ | 87.0–89.9% | D+ | 67.0–69.9% |
| B | 83.0–86.9% | D | 60.0–66.9% |
| B- |
80.0–82.9% |
F | Below 60.0% |
| C+ | 77.0–79.9% |
[Provide information about how grades will be rounded (eg, if 89% earns a B+ and 90% earns an A-, what grade is given to a student with an 89.5?]
Final course percentages will be rounded to the nearest tenth of a percentage point before the letter-grade thresholds are applied. For example, 89.95% rounds to 90.0% and earns an A−, whereas 89.5% remains 89.5% and earns a B+. Grades will not otherwise be curved or rounded to the next letter-grade category.
Attendance and Make-Up Policy
[Provide details on how attendance affects final course grades. Indicate whether and how missed exams, laboratory sessions, etc. can be made up. Sample attendance statement: “Students should make all reasonable efforts to attend all class meetings. However, in the event a student is unable to attend a class, it is the responsibility of the student to inform the instructor as soon as possible, explain the reason for the absence (and provide documentation, if appropriate), and make up class work missed within a reasonable amount of time, if allowed. Missing class meetings may result in reducing the overall grade in the class.” ]
Students should make all reasonable efforts to attend class meetings and participate in laboratories, technical discussions, project reviews, and presentations. Attendance is not assigned a separate percentage of the final grade. However, students remain responsible for all material, announcements, graded activities, and deadlines associated with missed classes.
If a student cannot attend class because of illness, an emergency, a university-authorized activity, a religious observance, or another legitimate circumstance, the student should contact the instructor as soon as reasonably possible. When an absence is foreseeable, notice should be provided before the affected class or deadline. Documentation may be requested when appropriate.
Make-up examinations or alternative arrangements will be provided only for approved absences. Students should contact the instructor before the examination whenever possible or within 48 hours following an emergency. Missed laboratories, presentations, demonstrations, or project reviews may be made up only when the activity can reasonably be replicated. Approved make-up work must be completed within the timeframe established by the instructor.
Academic Integrity Statement
You are responsible for reading WSU's Academic Integrity Policy, which is based on Washington State law. If you cheat in your work in this class you will:
Receive a grade of zero on the affected assignment, laboratory, project component, or examination. Serious or repeated violations may result in failure of the course, in addition to being reported to the Center for Community Standards.
-Be reported to the Center for Community Standards
-Have the right to appeal my decision
-Not be able to drop the course of withdraw from the course until the appeals process is finished
If you have any questions about what you can and cannot do in this course, ask me.
If you want to ask for a change in my decision about academic integrity, use the form at the Center for Community Standards website. You must submit this request within 21 calendar days of the decision.