profiling

A benchmark is an experiment, and most are badly designed enough to produce confident wrong answers. This lesson covers what a measurement must control to mean anything, the ways microbenchmarks lie including code the compiler deletes, how to catch regressions in continuous integration despite noisy machines, and how to recognise the point where optimising stops paying.

Slow software is slow for a short list of reasons, and each one has a signature you can recognise before you find the code. This lesson covers the six recurring bottleneck classes, waiting on I/O, chatty queries, allocation pressure, lock contention, memory access patterns and serialisation, with the symptom that identifies each and the fix that actually works.

A profiler is not a neutral observer: sampling and instrumentation see different things, distort the program in different ways, and answer different questions. This lesson covers how each works, why CPU time and wall-clock time give opposite answers, and how to read a flame graph correctly, including the axis that means nothing and is misread constantly.

Most optimisation effort is spent on code that was never the problem, and the reason is that intuition about where time goes is reliably wrong. This lesson covers why guessing fails, the arithmetic that caps what any optimisation can buy, the difference between latency and throughput, and how to set a target that tells you when to stop.

The pattern that made attention slow recurs across the stack, and once you know what to look for it is easy to find. This lesson applies the diagnosis to normalisation layers, optimizer steps, loss functions and inference decoding, covers why fused attention silently stops applying when a model deviates slightly from standard, and gives the profiling routine that decides where to look first.

A working CUDA kernel is the start, not the finish. How to measure occupancy, spot uncoalesced loads and warp divergence, and read the three numbers in Nsight Compute that actually matter.