Saagar Jha
Saagar Jha
Autotuning takes a while and for us most of that time is actually spent compiling the JIT kernel for each configuration rather than running the code. Since this process happens...
**Describe the bug** Some files are missing the headers that they rely on, which means they cannot be included by themselves. This is "hidden" in most of the examples because...
(I'm actually seeing this in the PyTorch profiler, fwiw: I just assume that it's using Kineto under the hood.) If I create and launch a CUDA graph that contains a...
**Describe the bug** This is half-bug half-feature request (or maybe even a question if this already exists). I am have a bunch of ephemeral runners I spin up in response...
### Describe the bug I haven't looked into why this is happening yet but I've been able to reduce it. If you run the following code it will hang: ```python3...
This is a feature request but it would be nice in some scenarios to have support for tracing asynchronous operations in Proton. Many things you'd want to measure have a...
Maybe there is a way to do this but I couldn't find it. In my head I would expect there to be a grab handle somewhere on the process/thread list...
When I'm navigating a trace, I will usually click around a few times before I actually have a need to see any details on anything. Perfetto likes to open the...
You can do this for threads but I couldn't see a way to do it for processes