CannyLab / tsne-cuda

GPU Accelerated t-SNE for CUDA with Python bindings
BSD 3-Clause "New" or "Revised" License
1.81k stars 130 forks source link

Fail with CUDA error 9. #123

Open Mi-Peng opened 1 year ago

Mi-Peng commented 1 year ago

Thanks for your working, this is a great implementation. I'm using tsne-cuda on Linux OS with NVIDIA GPU. Running tsnecuda.test() is fine but when I run for my code got

Faiss assertion 'err__ == cudaSuccess' failed in void faiss::gpu::ivfInterleavedScanImpl_32_(faiss::gpu::Tensor<float, 2, true>&, faiss::gpu::Tensor<int, 2, true>&, thrust::device_vector<void*>&, thrust::device_vector<void*>&, faiss::gpu::IndicesOptions, thrust::device_vector<int>&, int, faiss::MetricType, bool, faiss::gpu::Tensor<float, 3, true>&, faiss::gpu::GpuScalarQuantizer*, faiss::gpu::Tensor<float, 2, true>&, faiss::gpu::Tensor<long int, 2, true>&, faiss::gpu::GpuResources*) at /home/conda/feedstock_root/build_artifacts/faiss-split_1636459943780/work/faiss/gpu/impl/scan/IVFInterleaved32.cu:13; details: CUDA error 9 invalid configuration argument

Here is how I run:

from tsnecuda import TSNE

tsne_embedding = TSNE(
            n_components=2,
            perplexity=perplexity,
            early_exaggeration=early_exaggeration,
            learning_rate=learning_rate,
            n_iter=n_iter,
            metric=metric,
            init='random',
        ).fit_transform(embedding)

and I install tsnecuda by conda command:

conda install tsnecuda -c cannylab

and

conda install tsnecuda -c conda-forge

but none of them work.

artemZholus commented 1 year ago

same issue

boycehbz commented 1 year ago

Is the problem being addressed?

boycehbz commented 1 year ago

I found it may fail on a large dataset