Pascal Mütschard

I am making these observations from the perspective of the API the server exposes, and not from the implementation: - The currently returned value of `[0]-[6]` is certainly incorrect, as...

> in terms of framebuffer request, the subcommand for [1, 0, 7, 0, 0, 0] (RenderPass) should be [371, 0, 0, 6] This will, of course, all depend on how...

Sorry to be spammy, I want to make sure we also consider multiple render passes and render pass parents: Yours: ``` ... - 371: vkQueueSubmit - 371.0: [371 0] -...

SGTM. Thanks for the clarification. I agree, it makes more sense to prioritize the grouping at the buffers.

Looks like `dumpsys package` is returning some bad data. What device is this? and what version of Android? Can you please also provide the output of running `adb shell dumpsys...

Thanks for the suggestion. It sounds reasonable and we'll prioritize. As for getting the data using CLI, you can do so via `gapit perfetto`, but it'll require you to write...

AGI does not yet support `VK_EXT_descriptor_indexing`. It is core in 1.2, so we will be adding support in the near future. Make sure you have the `Hide Unknown Extensions` checkbox...

Thank you for the feedback. For VSync: in the trace dialog, ensure you have the "Frame Lifecycle" option enabled. This will cause SurfaceFlinger to issue events from the buffer queue....

I think this has been fixed in 3.0.1 release. Can you please confirm the version of AGI you are using and try 3.0.1?

Oooh, I like the idea behind #17! It "automates" some of the work. Sorry, I filed this bug with a solution in mind rather than just asking for what I...