GPU Roofline Model Study for a Computer Vision Inference Workload
Overview
What this challenge is about.
Profile ResNet-50 on A100 and H100 GPUs, build roofline plots, and identify compute- vs. memory-bound kernels to earn your verifiable certificate.
The scenario
The capacity team allocates around USD 4M/quarter in GPU spend — a 15-percent placement improvement justifies the analysis on its own.
The Brief
What you'll do, and what you'll demonstrate.
Build empirical roofline models for two GPU generations, profile a ResNet-50 inference workload, and recommend batch-size-and-SKU placement.
Earning criteria — what you'll demonstrate
- Construct an empirical roofline model from measured peak FLOPs and bandwidth
- Profile a real GPU workload with Nsight Compute
- Map workload regimes to GPU SKUs based on arithmetic intensity
- Communicate placement recommendations to a capacity team
Program Fit
Where this fits in your program.
Sharpens the same skills your degree expects you to demonstrate.
Aligned coursework coming soon.
Skills
Skills you'll demonstrate.
Each one shows up on your verified credential.
- Gpu Architecture
Apply gpu architecture to solve real industry problems and demonstrate production-level capability.
- Roofline Model
Apply roofline model to solve real industry problems and demonstrate production-level capability.
- Performance Modeling
Apply performance modeling to solve real industry problems and demonstrate production-level capability.
- Benchmarking
Apply benchmarking to solve real industry problems and demonstrate production-level capability.
- Memory Subsystems
Apply memory subsystems to solve real industry problems and demonstrate production-level capability.
- Stakeholder Communication
Apply stakeholder communication to solve real industry problems and demonstrate production-level capability.
Careers
Career paths this challenge builds toward
Completing this challenge demonstrates skills that transfer directly to these roles: