Module launch_trace 

2026-09-26: Launch trace: record the kernel launches, async memsets and device-to-device copies a region enqueues, so two executions of it can be diffed op by op (end_and_diff). The verify paths trace their captured region this way to find a host value that differs between two steps. Off until begin is called.

Owner: telemetry. Invariants: record stores an op only between begin and the next end_and_diff.

Structs

Entry

2026-09-26: One enqueued op. kind separates kernels from memsets and copies, so an op that appears or vanishes shows as an op mismatch.

Functions

begin

end_and_diff

2026-09-26: Stop recording and diff this trace against the previous one. Returns None on the first call, else a report of how many ops differ, detailing at most max_report of them.

kernel_name

2026-09-26: The module::func a kernel handle was looked up as, if it was named.

name_kernel

2026-09-26: Remember a kernel handle’s name. The CUDA backend’s GpuBackend::kernel calls it on every successful lookup.

on

record