Scott Todd

cc @KoolJBlack @dcaballe (sounded like there was some interest in improving Tracy workflows on Android, this could help and would be 1-3 days of work I think)

This source snapshotting also helps connect locations with sampled assembly: With `--iree-hal-dump-executable-sources-to={path}`:  Without: 

Yeah. With embedded (`--iree-llvmcpu-link-embedded=true`), I don't see any sampling or assembly for executable code:  I do at least see the instrumented source though.

To clarify (since I just confused myself about this): * If you compile on machine A and run on machine A, the .tracy file _will_ be portable to machine B...

Fixed by https://github.com/openxla/iree/pull/16757

What version of TensorFlow are you using?

> This just tests that it compiles, or does it test anything else? Just testing for compilation, yeah. Could add a lit test too 🤔

Tried adding a lit test, but that ended up being tricky: https://github.com/iree-org/iree/actions/runs/9273552641/job/25513953669?pr=16454 * The tracing job just builds the runtime, but to run a lit test we need the compiler...

We could separate "unit tests" from "integration/e2e tests" in the CMake project. Unit tests should be able to run right after the build step, while integration tests should use a...

I'm deciding which of these job sequences to aim for: * `build_dist_package` --> `compile_test_deps` --> `test_gpu` * `build_dist_package` --> `test_gpu` Portable targets like Android _require_ running the compiler on a...