Dante Gama Dessavre

Add capability to use different allocators in cuML Python benchmarks

1

PR adds a new CLI option to run_benchmarks.py to set up different RMM allocators.

First pass of removing tests, codepaths and other bits related to CUDA 11

2

Related to #cuda11 Version 25.08 drops CUDA 11.x support, this PR removes codepaths, tests and other small bits related to it.

[FEA] Add automatic dispatching for cuml/sklearn RandomForest models to FIL load function

5

**Is your feature request related to a problem? Please describe.** A clear and concise description of what the problem is. Ex. I wish I could use cuML to do [...]...

Move cumlprims_mg into cuML

1

This PR removes the dependency on the separate cumlprims_mg library by moving all of its code directly into cuML. The cumlprims_mg library provided multi-node multi-GPU (MNMG) primitives used by several...

Add xgboost with native API and QuantileDMatrix to benchmark utilities

2

Follow up to #7350 that added simple xgboost to the benchmark utility, this PR allows to use the native XGB API as well as using the QuantileDMatrix optionally.