Performance Guides

Nightly measurements

1. Benchmarks — a nightly run on Ascend 910B1 (Atlas A2), reporting device time per op per workload against the fastest other implementation of the same op. How the numbers are taken and how to read the ratio is set out in "How these numbers are taken" in that section.

Tools for locating a problem

1. In-Kernel Timeline Trace — annotate a kernel body with markers and read back a per-CTA timeline: gaps, stalls, and how far the producer and consumer overlap. None of that is visible to a per-kernel profiler. The API is Trace.

Tuning practice for TileLang

1. Tuning Memory-Bound Kernels — upstream NVIDIA / H200 reference measurements, not Ascend results: what memory-bound means on the roofline, and measured guidance on the two places an access pattern decides the bandwidth a kernel reaches: global memory and shared memory.

The work one call costs comes from op.eval_roofline(), derived from the manifest's roofline field — the ceiling a measurement is read against. The model and the field specification are in Roofline.