Datasetpublic

willychan21/ParallelKernelBench_Problems

ParallelKernelBench (benchmark) Reference problems for ParallelKernelBench: a benchmark for LLM-generated multi-GPU CUDA kernels. This dataset contains 87 reference implementations in reference/ and the input tensor specification in utils/input_output_tensors.py. Files Path Description data/problems.parquet One row per problem (tabular access) reference/*.py Reference solution() implementations utils/input_output_tensors.py Input/output tensor… See the full description on the dataset page: https://huggingface.co/datasets/willychan21/ParallelKernelBench_Problems.

sourceHugging Faceapache-2.0updated 4mo agoView on Hugging Face

0likes266downloads

rootcodereference2_allgather.py

2_allgather.py11 linesDownload Raw Back to reference

1import torch2import torch.distributed as dist3 4 5@torch.no_grad()6def solution(tensor: torch.Tensor) -> torch.Tensor:7    world_size = dist.get_world_size()8    out = tensor.new_empty((world_size,) + tensor.shape)9    dist.all_gather_into_tensor(out, tensor)10    return out11