Sam Van Kooten

Yeah, a quick test shows I can do what I originally wanted with this PR. Thanks for pushing this all along!

I've been seeing similar sporadic deadlocks, but with prefect 3.2.15. Just now it occurred with a very light load on the server (

I'm very excited to see this! Just took it for a spin with PUNCH test data and it's working great. Thanks, @ayshih and all!

I don't know anything about using GPUs (and I wish I did), but I'm very interested in this, as are a few of my colleagues. I'd be really interested in...

Thanks for posting your code! I tried out your function on my computer and I'm afraid I'm not seeing any speedup. [Here's how I explored this](https://gist.github.com/svank/94196433ea3711d8163b38bfb91f2a53), on a computer with...

I'm also getting a big speedup with your code: ``` Running CPU version... CPU execution time: 4.105450 seconds Running GPU version... GPU execution time: 2.498859 seconds GPU speedup: 39.13% Results...

Yeah, if we drop the idea of individual files being split into chunks, I think this PR becomes easy to finish. I think the only remaining large task is figuring...

I think at some point this might warrant PUNCH hours, but it's too far down the priority list right now. For `reproject_adaptive`, the inner-most loop is just summing weighted input...