LazyLabES

Run GPU clusters
like you built them.

Learn GPU cluster architecture, operations, optimization, monitoring, performance and security by doing. Read a section, play with the live model, clear the checkpoint, then take on the boss: a broken cluster with a real terminal.

Rank
Intern
0 XP · 150 to Junior SRE
Sections
0/44
Bosses
0/8
Streak
0 days
Start reading: GPU →

Contents

Or skip to free play in the Lab →
Chapter 1

Anatomy of a GPU Cluster

From one streaming multiprocessor to a thousand-GPU fabric

8 sections · 1 bossnot started
Chapter 2

Running Jobs

Drivers, schedulers and the plumbing that gets work onto GPUs

6 sections · 1 bossnot started
Chapter 3

Watching the Fleet

Telemetry, error codes and finding the one broken thing

6 sections · 2 bossesnot started
Chapter 4

Making Training Fast

Kernels, precision and the parallelism zoo

11 sections · 2 bossesnot started
Chapter 5

Measuring and Scaling

Rooflines, MFU, collectives and the price of a joule

7 sections · 1 bossnot started
Chapter 6

Securing Shared GPUs

Isolation, exposed services and incident response

6 sections · 1 bossnot started