Malte Förster

DBSCAN utilize rbc eps_neighbors

2

This PR enables rbc eps-neighbor computation via raft. The resulting adjacency matrix is sparse and allows to skip the implicit conversion. Notes: * the 'algorithm'-parameter was added to the DBSCAN...

Just a test PR to validate raft fix

[BUG] cagra filter test returns duplicate neighbor indices

1

**Describe the bug** A new consistency check added in #2221 caused some cagra filter tests to fail. The resulting neighborhood does seem to contain duplicate neighboring ids. The effect is...

Several APIs in `pylibraft` implicitly call into `rmm`, e.g. to [create a DeviceBuffer for a ndarray](https://github.com/rapidsai/raft/blob/addb059975478375a422d32e9bec30e1aeb16527/python/pylibraft/pylibraft/common/device_ndarray.py#L63). The standard behavior for `rmm` is to use the default stream. This causes issues...

Remove 'sample' parameter from stats::mean API

2

This PR removes the sample-parameter from the `raft::stats::mean` API to prevent people from using it by accident when for example computing the mean for a sampled variance computation. This also...

**Is your feature request related to a problem? Please describe.** Brought up by https://github.com/rapidsai/raft/issues/2205 the [sum kernel](https://github.com/rapidsai/raft/blob/b863f1810fdb0679f7777b656486dbfc96470d2b/cpp/include/raft/stats/detail/sum.cuh#L51) has been adapted to account for underflow issues when adding large amounts of...

[BUG] IVF-PQ index creation crashes on aarch64 for wiki_all_1M benchmark

1

**Describe the bug** IVF-PQ build of wiki_all_1M fails on Grace/H200 with `CUDA Exception: Warp Illegal Address Thread 1 "RAFT_IVF_PQ_ANN" received signal CUDA_EXCEPTION_14, Warp Illegal Address. [Switching focus to CUDA kernel...

refactor memory ownership svm

3

As discussed in #6057 this is a PR to refactor the raw memory allocations in the svm model which are passed back to the caller. I have removed all raw...

[WIP] Add disk2disk serialization foe ACE Algorithm

2

This PR adds a serialization routine that allows combination of dataset, graph, and mapping as per step (3) of #1404. The data will be combined on-the-fly while streamed from disk...