Tom Birch
Tom Birch
**Describe the bug** In the patch options menu (pressing in patch mode) there's an entry to 'move patch into new commit', and looking at the source code it seems that...
I'm using findNeighbors via PDAL filters.icp, and the general algorithm is (pseudocode): ``` # Given two point clouds A and B iterations = 100 # customizable kdtree = KDTree(A) for...
In my usecase I load .ply (polygons) and .las files in the same view, but the polygons are rendered with meshface.glsl which is special-cased and not supported by the "Shader...
Maybe there's an obvious solution to this. I have multiple build configs, the default one uses clang via bazel-toolchain, the others use emscripten to target js/wasm. When I test c++...
**Description** Running a compute shader with an input of an array of struct of 20 bytes will only see the first 20/32 entries. **Repro steps** 1. Clone https://github.com/froody/wgpu/tree/compute-bug 2. `cd...
Consider the attached program, run it on two machines, e.g. machine1% python hello.py -o machine1 -r 0 -t machine2% python hello.py -o machine1 -r 1 -t Omitting the "-t" (i.e....
I was running some benchmarks with torch-ucc using xccl for collectives, and I noticed very bad performance compared to NCCL. See numbers here: https://gist.github.com/froody/a86a5b2c5d9f46aedba7e930f4b4e08d It's possible this is due to...
`unitwise_norm`, used by `adaptive_grad_clip`, only supports a few values of `ndim`, and raises ValueError when applied to a conv3d kernel since `ndim=5` (HWDIO). Would it be acceptable to add an...
**Describe the bug** blosc2_schunk_open_udio looks like it could be used to read an schunk from any seekable API (e.g. ftp), but it calls stat on the provided uri, requiring that...
I just imported all my runs from mlflow and the timestamps seem to reflect the time to import each run, not the actual times from mlflow