Merge branch 'main' into clearer_error_messages

Author: rwgk

cuda_bindings/README.md

@@ -95,10 +95,10 @@ To run these samples:
 * `python -m pytest tests/cython/` against editable installations
 * `pytest tests/cython/` against installed packages
 
-### Benchmark (WIP)
+### Benchmark
 
-Benchmarks were used for performance analysis during initial release of CUDA Python. Today they need to be updated the 12.x toolkit and are work in progress.
+Allows for analyzing binding performance using plugin [pytest-benchmark](https://github.com/ionelmc/pytest-benchmark).
 
-The intended way to run these benchmarks was:
-* `python -m pytest --benchmark-only benchmark/` against editable installations
-* `pytest --benchmark-only benchmark/` against installed packages
+To run these benchmarks:
+* `python -m pytest --benchmark-only benchmarks/` against editable installations
+* `pytest --benchmark-only benchmarks/` against installed packages

cuda_bindings/benchmarks/__init__.py

@@ -1,4 +1,4 @@
-# Copyright 2021-2024 NVIDIA Corporation.  All rights reserved.
+# Copyright 2021-2025 NVIDIA Corporation.  All rights reserved.
 #
 # Please refer to the NVIDIA end user license agreement (EULA) associated
 # with this source code for terms and conditions that govern your use of
@@ -25,7 +25,7 @@ def ASSERT_DRV(err):
         raise RuntimeError(f"Unknown error type: {err}")
 
 
-@pytest.fixture
+@pytest.fixture(scope="function")
 def init_cuda():
     # Initialize
     (err,) = cuda.cuInit(0)
@@ -47,7 +47,7 @@ def init_cuda():
     ASSERT_DRV(err)
 
 
-@pytest.fixture
+@pytest.fixture(scope="function")
 def load_module():
     module = None
 

cuda_bindings/benchmarks/perf_test_utils.py renamed to cuda_bindings/benchmarks/conftest.py

@@ -1,4 +1,4 @@
-# Copyright 2021-2024 NVIDIA Corporation.  All rights reserved.
+# Copyright 2021-2025 NVIDIA Corporation.  All rights reserved.
 #
 # Please refer to the NVIDIA end user license agreement (EULA) associated
 # with this source code for terms and conditions that govern your use of

cuda_bindings/benchmarks/kernels.py

@@ -1,4 +1,4 @@
-# Copyright 2021-2024 NVIDIA Corporation.  All rights reserved.
+# Copyright 2021-2025 NVIDIA Corporation.  All rights reserved.
 #
 # Please refer to the NVIDIA end user license agreement (EULA) associated
 # with this source code for terms and conditions that govern your use of
@@ -9,17 +9,14 @@
 
 import pytest
 
-# Always skip since cupy is not CTK 12.x yet
-skip_tests = True
-if not skip_tests:
-    try:
-        import cupy
+try:
+    import cupy
 
-        skip_tests = False
-    except ImportError:
-        skip_tests = True
+    skip_tests = False
+except ImportError:
+    skip_tests = True
 
-from .kernels import kernel_string
+from kernels import kernel_string
 
 
 def launch(kernel, args=()):

cuda_bindings/benchmarks/test_cupy.py

@@ -1,4 +1,4 @@
-# Copyright 2021-2024 NVIDIA Corporation.  All rights reserved.
+# Copyright 2021-2025 NVIDIA Corporation.  All rights reserved.
 #
 # Please refer to the NVIDIA end user license agreement (EULA) associated
 # with this source code for terms and conditions that govern your use of
@@ -8,12 +8,11 @@
 import ctypes
 
 import pytest
+from conftest import ASSERT_DRV
+from kernels import kernel_string
 
 from cuda import cuda
 
-from .kernels import kernel_string
-from .perf_test_utils import ASSERT_DRV
-
 
 def launch(kernel, stream, args=(), arg_types=()):
     cuda.cuLaunchKernel(

cuda_bindings/benchmarks/test_launch_latency.py

@@ -1,4 +1,4 @@
-# Copyright 2021-2024 NVIDIA Corporation.  All rights reserved.
+# Copyright 2021-2025 NVIDIA Corporation.  All rights reserved.
 #
 # Please refer to the NVIDIA end user license agreement (EULA) associated
 # with this source code for terms and conditions that govern your use of

cuda_bindings/benchmarks/test_numba.py

@@ -1,4 +1,4 @@
-# Copyright 2021-2024 NVIDIA Corporation.  All rights reserved.
+# Copyright 2021-2025 NVIDIA Corporation.  All rights reserved.
 #
 # Please refer to the NVIDIA end user license agreement (EULA) associated
 # with this source code for terms and conditions that govern your use of
@@ -8,11 +8,10 @@
 import random
 
 import pytest
+from conftest import ASSERT_DRV
 
 from cuda import cuda
 
-from .perf_test_utils import ASSERT_DRV
-
 random.seed(0)
 
 idx = 0