Require numba-cuda>=0.16.0

conda/environments/all_cuda-129_arch-aarch64.yaml

@@ -54,7 +54,7 @@ dependencies:
 - nbsphinx
 - ninja
 - notebook
-- numba-cuda>=0.14.0,<0.15.0a0
+- numba-cuda>=0.16.0,<0.17.0a0
 - numba>=0.59.1,<0.62.0a0
 - numpy>=1.23,<3.0a0
 - numpydoc
@@ -69,7 +69,6 @@ dependencies:
 - pre-commit
 - pyarrow>=14.0.0,<20.0.0a0
 - pydata-sphinx-theme>=0.15.4
-- pynvjitlink>=0.0.0a0
 - pynvml>=12.0.0,<13.0.0a0
 - pytest
 - pytest-benchmark

conda/environments/all_cuda-129_arch-x86_64.yaml

@@ -55,7 +55,7 @@ dependencies:
 - nbsphinx
 - ninja
 - notebook
-- numba-cuda>=0.14.0,<0.15.0a0
+- numba-cuda>=0.16.0,<0.17.0a0
 - numba>=0.59.1,<0.62.0a0
 - numpy>=1.23,<3.0a0
 - numpydoc
@@ -70,7 +70,6 @@ dependencies:
 - pre-commit
 - pyarrow>=14.0.0,<20.0.0a0
 - pydata-sphinx-theme>=0.15.4
-- pynvjitlink>=0.0.0a0
 - pynvml>=12.0.0,<13.0.0a0
 - pytest
 - pytest-benchmark

conda/recipes/cudf/recipe.yaml

@@ -69,7 +69,7 @@ requirements:
     - typing_extensions >=4.0.0
     - pandas >=2.0,<2.4.0dev0
     - cupy >=12.0.0
-    - numba-cuda >=0.14.0,<0.15.0a0
+    - numba-cuda >=0.16.0,<0.17.0a0
     - numba >=0.59.1,<0.62.0a0
     - numpy >=1.23,<3.0a0
     - pyarrow>=14.0.0,<20.0.0a0
@@ -78,13 +78,7 @@ requirements:
     - ${{ pin_compatible("rmm", upper_bound="x.x") }}
     - fsspec >=0.6.0
     - cuda-cudart
-    # Needed by Numba for CUDA support
-    - cuda-nvcc-impl
-    # TODO: Add nvjitlink here
-    # xref: https://github.com/rapidsai/cudf/issues/12822
-    - cuda-nvrtc
     - cuda-python >=12.6.2,<13.0a0
-    - pynvjitlink
     - if: linux and x86_64
       then:
         - libcufile

dependencies.yaml

@@ -666,12 +666,14 @@ dependencies:
       - output_types: [conda, requirements, pyproject]
         packages:
           - cachetools
-          - &numba-cuda-dep numba-cuda>=0.14.0,<0.15.0a0
           - &numba-dep numba>=0.59.1,<0.62.0a0
           - nvtx>=0.2.1
           - packaging
           - rich
           - typing_extensions>=4.0.0
+      - output_types: [conda]
+        packages:
+          - &numba-cuda-dep numba-cuda>=0.16.0,<0.17.0a0
       - output_types: requirements
         packages:
           # pip recognizes the index as a global option for the requirements.txt file
@@ -684,26 +686,23 @@ dependencies:
             packages: &run_cudf_packages_all_cu12
               - cuda-python>=12.6.2,<13.0a0
           - {matrix: null, packages: *run_cudf_packages_all_cu12}
-      - output_types: conda
+      - output_types: [requirements, pyproject]
         matrices:
           - matrix: {cuda: "12.*"}
             packages:
-              - &pynvjitlink_unsuffixed pynvjitlink>=0.0.0a0
+              - &numba-cuda-cu12-dep numba-cuda[cu12]>=0.16.0,<0.17.0a0
+          - matrix: # Fallback for no matrix
+            packages:
+              - *numba-cuda-cu12-dep
       - output_types: [requirements, pyproject]
         matrices:
           - matrix:
               cuda: "12.*"
               cuda_suffixed: "true"
             packages:
-              - pynvjitlink-cu12>=0.0.0a0
               - nvidia-cuda-nvcc-cu12
               - nvidia-cuda-nvrtc-cu12
-          - matrix:
-              cuda: "12.*"
-              cuda_suffixed: "false"
-            packages: &run_cudf_cu12_unsuffixed
-              - *pynvjitlink_unsuffixed
-          - {matrix: null, packages: *run_cudf_cu12_unsuffixed}
+          - {matrix: null, packages: []}
   run_cudf_polars:
     common:
       - output_types: [conda, requirements, pyproject]
@@ -750,7 +749,7 @@ dependencies:
               cuda: "12.*"
             packages:
               - cuda-sanitizer-api
-          - matrix: # Fallback for CUDA 12 or no matrix
+          - matrix: # Fallback for no matrix
             packages:
   # packages we want in the 'test_cpp' group in 'files', for CI, but which
   # shouldn't be added to 'all' for building a development environment
@@ -780,16 +779,15 @@ dependencies:
   test_python_cudf_common:
     specific:
       # Define additional constraints for testing with oldest dependencies.
-      - output_types: [conda, requirements]
+      - output_types: [conda, requirements, pyproject]
         matrices:
           - matrix: {dependencies: "oldest"}
             packages:
-              - numba-cuda==0.14.0
               - numba==0.59.1
+              - numba-cuda==0.16.0
               - pandas==2.0.*
           - matrix: {dependencies: "latest"}
             packages:
-              - *numba-cuda-dep
               - *numba-dep
               - pandas==2.3.1
           - matrix:
@@ -877,8 +875,6 @@ dependencies:
       - output_types: [conda, requirements, pyproject]
         packages:
           - dask-cuda==25.8.*,>=0.0.0a0
-          - *numba-cuda-dep
-          - *numba-dep
     specific:
       - output_types: [conda, requirements]
         matrices:

python/cudf/cudf_pandas_tests/third_party_integration_tests/tests/test_stumpy.py

@@ -1,4 +1,4 @@
-# Copyright (c) 2023-2024, NVIDIA CORPORATION.
+# Copyright (c) 2023-2025, NVIDIA CORPORATION.
 
 import numpy as np
 import pandas as pd
@@ -36,7 +36,7 @@ def test_1d_gpu():
         50  # Approximately, how many data points might be found in a pattern
     )
     all_gpu_devices = [
-        device.id for device in cuda.list_devices()
+        int(device.id) for device in cuda.list_devices()
     ]  # Get a list of all available GPU devices
 
     return stumpy.gpu_stump(

python/cudf/pyproject.toml

@@ -23,7 +23,7 @@ dependencies = [
     "cupy-cuda12x>=12.0.0",
     "fsspec>=0.6.0",
     "libcudf==25.8.*,>=0.0.0a0",
-    "numba-cuda>=0.14.0,<0.15.0a0",
+    "numba-cuda[cu12]>=0.16.0,<0.17.0a0",
     "numba>=0.59.1,<0.62.0a0",
     "numpy>=1.23,<3.0a0",
     "nvtx>=0.2.1",
@@ -32,7 +32,6 @@ dependencies = [
     "pyarrow>=14.0.0,<20.0.0a0,!=17.0.0; platform_machine=='aarch64'",
     "pyarrow>=14.0.0,<20.0.0a0; platform_machine=='x86_64'",
     "pylibcudf==25.8.*,>=0.0.0a0",
-    "pynvjitlink>=0.0.0a0",
     "rich",
     "rmm==25.8.*,>=0.0.0a0",
     "typing_extensions>=4.0.0",

python/dask_cudf/dask_cudf/tests/test_distributed.py

@@ -1,6 +1,5 @@
-# Copyright (c) 2020-2024, NVIDIA CORPORATION.
+# Copyright (c) 2020-2025, NVIDIA CORPORATION.
 
-import numba.cuda
 import pytest
 
 import dask
@@ -9,6 +8,7 @@
 from distributed.utils_test import cleanup, loop, loop_in_thread  # noqa: F401
 
 import cudf
+import rmm
 from cudf.testing import assert_eq
 
 import dask_cudf
@@ -17,7 +17,7 @@
 
 
 def at_least_n_gpus(n):
-    ngpus = len(numba.cuda.gpus)
+    ngpus = rmm._cuda.gpu.getDeviceCount()
     return ngpus >= n
 
 

python/dask_cudf/pyproject.toml

@@ -48,8 +48,6 @@ cudf = "dask_cudf.backends:CudfBackendEntrypoint"
 [project.optional-dependencies]
 test = [
     "dask-cuda==25.8.*,>=0.0.0a0",
-    "numba-cuda>=0.14.0,<0.15.0a0",
-    "numba>=0.59.1,<0.62.0a0",
     "pytest",
     "pytest-cov",
     "pytest-xdist",