All manual fixes in one commit.

.github/actions/workflow-build/build-workflow.py

@@ -157,7 +157,6 @@ def canonicalize_host_compiler_name(cxx_string):
         )
 
     hc_def = matrix_yaml["host_compilers"][id]
-    hc_versions = hc_def["versions"]
 
     if not version:
         version = max(
@@ -852,10 +851,10 @@ def apply_matrix_job_exclusion(matrix_job, exclusion):
         # Some tags are left unexploded (e.g. 'jobs') to optimize scheduling,
         # so the values can be either a list or a single value.
         # Standardize to a list for comparison:
-        if type(excluded_values) != list:
+        if not isinstance(excluded_values, list):
             excluded_values = [excluded_values]
         matrix_values = matrix_job[tag]
-        if type(matrix_values) != list:
+        if not isinstance(matrix_values, list):
             matrix_values = [matrix_values]
 
         # Identify excluded values that are present in the matrix job for this tag:
@@ -1075,7 +1074,7 @@ def parse_workflow_matrix_jobs(args, workflow_name):
 
     if args:
         if (
-            args.dirty_projects != None
+            args.dirty_projects is not None
         ):  # Explicitly check for None, as an empty list is valid:
             matrix_jobs = [
                 job for job in matrix_jobs if job["project"] in args.dirty_projects

.github/actions/workflow-results/prepare-execution-summary.py

@@ -237,9 +237,6 @@ def get_summary_stats(summary):
 
 
 def get_summary_heading(summary, walltime):
-    passed = summary["passed"]
-    failed = summary["failed"]
-
     if summary["passed"] == 0:
         flag = "🟥"
     elif summary["failed"] > 0:
@@ -298,7 +295,6 @@ def get_tag_line(tag, tag_summary):
 def get_value_line(value, value_summary, tag_summary):
     passed = value_summary["passed"]
     failed = value_summary["failed"]
-    total = passed + failed
 
     parent_size = len(tag_summary["values"])
     parent_failed = tag_summary["failed"]
@@ -372,7 +368,7 @@ def main():
     # The timing file is not required.
     try:
         job_times = json.load(args.job_times)
-    except:
+    except Exception:
         job_times = None
 
     write_workflow_summary(workflow, job_times)

benchmarks/scripts/analyze.py

@@ -293,7 +293,10 @@ def case_top(alpha, N, algname, ct_point_name, case_dfs):
     print("{}[{}]:".format(algname, ct_point_name))
 
     if alpha < 1.0:
-        case_df = remove_matching_distributions(alpha, case_df)
+        for subbench in case_dfs:
+            case_dfs[subbench] = remove_matching_distributions(
+                alpha, case_dfs[subbench]
+            )
 
     for subbench in case_dfs:
         case_dfs[subbench] = extract_complete_variants(case_dfs[subbench])

benchmarks/scripts/cccl/__init__.py

@@ -1 +1 @@
-from . import bench
+from . import bench as bench

benchmarks/scripts/cccl/bench/__init__.py

@@ -1,6 +1,6 @@
-from .config import *
-from .storage import *
-from .bench import Bench
-from .cmake import CMake
-from .score import *
-from .search import *
+from .config import *  # noqa: F403
+from .storage import *  # noqa: F403
+from .bench import Bench as Bench
+from .cmake import CMake as CMake
+from .score import *  # noqa: F403
+from .search import *  # noqa: F403

benchmarks/scripts/cccl/bench/bench.py

@@ -8,10 +8,10 @@
 import numpy as np
 
 from .cmake import CMake
-from .config import *
+from .config import BasePoint, Config
 from .storage import Storage, get_bench_table_name
-from .score import *
-from .logger import *
+from .score import compute_axes_ids, compute_weight_matrices, get_workload_weight
+from .logger import Logger
 
 
 def first_val(my_dict):

benchmarks/scripts/cccl/bench/cmake.py

@@ -6,7 +6,7 @@
 from .build import Build
 from .config import Config
 from .storage import Storage
-from .logger import *
+from .logger import Logger
 
 
 def create_builds_table(conn):

benchmarks/scripts/cccl/bench/config.py

@@ -5,8 +5,8 @@
 
 def randomized_cartesian_product(list_of_lists):
     length = 1
-    for l in list_of_lists:
-        length *= len(l)
+    for lst in list_of_lists:
+        length *= len(lst)
 
     visited = set()
     while len(visited) < length:

cub/benchmarks/docker/recipe.py

@@ -4,7 +4,7 @@
 
 hpccm.config.set_container_format("docker")
 
-Stage0 += hpccm.primitives.baseimage(image="nvidia/cuda:12.2.0-devel-ubuntu22.04")
+Stage0 = hpccm.primitives.baseimage(image="nvidia/cuda:12.2.0-devel-ubuntu22.04")
 Stage0 += hpccm.building_blocks.apt_get(
     ospackages=[
         "git",

python/cuda_cooperative/cuda/cooperative/__init__.py

@@ -2,5 +2,4 @@
 #
 # SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 
-import cuda.cooperative.experimental
-from cuda.cooperative._version import __version__
+from cuda.cooperative._version import __version__ as __version__

python/cuda_cooperative/cuda/cooperative/experimental/__init__.py

@@ -2,5 +2,6 @@
 #
 # SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 
-from cuda.cooperative.experimental import block, warp
-from cuda.cooperative.experimental._types import StatefulFunction
+from cuda.cooperative.experimental import block as block
+from cuda.cooperative.experimental import warp as warp
+from cuda.cooperative.experimental._types import StatefulFunction as StatefulFunction

python/cuda_cooperative/cuda/cooperative/experimental/block/__init__.py

@@ -2,11 +2,19 @@
 #
 # SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 
-from cuda.cooperative.experimental.block._block_merge_sort import merge_sort_keys
-from cuda.cooperative.experimental.block._block_reduce import reduce, sum
-from cuda.cooperative.experimental.block._block_scan import exclusive_sum
+from cuda.cooperative.experimental.block._block_merge_sort import (
+    merge_sort_keys as merge_sort_keys,
+)
+from cuda.cooperative.experimental.block._block_reduce import reduce as reduce
+from cuda.cooperative.experimental.block._block_reduce import sum as sum
+from cuda.cooperative.experimental.block._block_scan import (
+    exclusive_sum as exclusive_sum,
+)
+from cuda.cooperative.experimental.block._block_radix_sort import (
+    radix_sort_keys as radix_sort_keys,
+)
 from cuda.cooperative.experimental.block._block_radix_sort import (
-    radix_sort_keys,
-    radix_sort_keys_descending,
+    radix_sort_keys_descending as radix_sort_keys_descending,
 )
-from cuda.cooperative.experimental.block._block_load_store import load, store
+from cuda.cooperative.experimental.block._block_load_store import load as load
+from cuda.cooperative.experimental.block._block_load_store import store as store

python/cuda_cooperative/cuda/cooperative/experimental/block/_block_load_store.py

@@ -3,7 +3,17 @@
 # SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 
 
-from cuda.cooperative.experimental._types import *
+import numba
+
+from cuda.cooperative.experimental._types import (
+    Algorithm,
+    Dependency,
+    DependentArray,
+    DependentPointer,
+    Invocable,
+    Pointer,
+    TemplateParameter,
+)
 from cuda.cooperative.experimental._common import make_binary_tempfile
 
 CUB_BLOCK_LOAD_ALGOS = {

python/cuda_cooperative/cuda/cooperative/experimental/block/_block_merge_sort.py

@@ -3,7 +3,18 @@
 # SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 
 import numba
-from cuda.cooperative.experimental._types import *
+
+from cuda.cooperative.experimental._types import (
+    Algorithm,
+    Constant,
+    Dependency,
+    DependentArray,
+    DependentOperator,
+    Invocable,
+    Pointer,
+    TemplateParameter,
+    numba_type_to_wrapper,
+)
 from cuda.cooperative.experimental._common import make_binary_tempfile
 
 

python/cuda_cooperative/cuda/cooperative/experimental/block/_block_radix_sort.py

@@ -3,7 +3,16 @@
 # SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 
 import numba
-from cuda.cooperative.experimental._types import *
+
+from cuda.cooperative.experimental._types import (
+    Algorithm,
+    Dependency,
+    DependentArray,
+    Invocable,
+    Pointer,
+    TemplateParameter,
+    Value,
+)
 from cuda.cooperative.experimental._common import make_binary_tempfile
 
 

python/cuda_cooperative/cuda/cooperative/experimental/block/_block_reduce.py

@@ -2,7 +2,19 @@
 #
 # SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 
-from cuda.cooperative.experimental._types import *
+import numba
+
+from cuda.cooperative.experimental._types import (
+    Algorithm,
+    Dependency,
+    DependentOperator,
+    DependentReference,
+    Invocable,
+    Pointer,
+    TemplateParameter,
+    Value,
+    numba_type_to_wrapper,
+)
 from cuda.cooperative.experimental._common import make_binary_tempfile
 
 

python/cuda_cooperative/cuda/cooperative/experimental/block/_block_scan.py

@@ -2,8 +2,17 @@
 #
 # SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 
+import numba
 
-from cuda.cooperative.experimental._types import *
+from cuda.cooperative.experimental._types import (
+    Algorithm,
+    Invocable,
+    Dependency,
+    DependentArray,
+    DependentOperator,
+    Pointer,
+    TemplateParameter,
+)
 from cuda.cooperative.experimental._common import make_binary_tempfile
 
 

python/cuda_cooperative/cuda/cooperative/experimental/warp/__init__.py

@@ -2,6 +2,9 @@
 #
 # SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 
-from cuda.cooperative.experimental.warp._warp_scan import exclusive_sum
-from cuda.cooperative.experimental.warp._warp_reduce import reduce, sum
-from cuda.cooperative.experimental.warp._warp_merge_sort import merge_sort_keys
+from cuda.cooperative.experimental.warp._warp_scan import exclusive_sum as exclusive_sum
+from cuda.cooperative.experimental.warp._warp_reduce import reduce as reduce
+from cuda.cooperative.experimental.warp._warp_reduce import sum as sum
+from cuda.cooperative.experimental.warp._warp_merge_sort import (
+    merge_sort_keys as merge_sort_keys,
+)

python/cuda_cooperative/cuda/cooperative/experimental/warp/_warp_merge_sort.py

@@ -3,7 +3,18 @@
 # SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 
 import numba
-from cuda.cooperative.experimental._types import *
+
+from cuda.cooperative.experimental._types import (
+    Algorithm,
+    Constant,
+    Dependency,
+    DependentArray,
+    DependentOperator,
+    Invocable,
+    Pointer,
+    TemplateParameter,
+    numba_type_to_wrapper,
+)
 from cuda.cooperative.experimental._common import make_binary_tempfile
 
 

python/cuda_cooperative/cuda/cooperative/experimental/warp/_warp_reduce.py

@@ -2,8 +2,18 @@
 #
 # SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 
-
-from cuda.cooperative.experimental._types import *
+import numba
+
+from cuda.cooperative.experimental._types import (
+    Algorithm,
+    Dependency,
+    DependentOperator,
+    DependentReference,
+    Invocable,
+    Pointer,
+    TemplateParameter,
+    numba_type_to_wrapper,
+)
 from cuda.cooperative.experimental._common import make_binary_tempfile
 
 

python/cuda_cooperative/cuda/cooperative/experimental/warp/_warp_scan.py

@@ -3,7 +3,16 @@
 # SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 
 
-from cuda.cooperative.experimental._types import *
+import numba
+
+from cuda.cooperative.experimental._types import (
+    Algorithm,
+    Dependency,
+    DependentReference,
+    Invocable,
+    Pointer,
+    TemplateParameter,
+)
 from cuda.cooperative.experimental._common import make_binary_tempfile
 
 

python/cuda_cooperative/setup.py

@@ -13,8 +13,10 @@
 project_path = os.path.abspath(os.path.dirname(__file__))
 cccl_path = os.path.abspath(os.path.join(project_path, "..", ".."))
 cccl_headers = [["cub", "cub"], ["libcudacxx", "include"], ["thrust", "thrust"]]
+__version__ = None
 with open(os.path.join(project_path, "cuda", "cooperative", "_version.py")) as f:
     exec(f.read())
+assert __version__ is not None
 ver = __version__
 del __version__
 

python/cuda_cooperative/tests/test_block_load.py

@@ -44,7 +44,6 @@ def output_index(i):
 
     @cuda.jit(link=block_load.files)
     def kernel(d_input, d_output):
-        tid = cuda.threadIdx.x
         temp_storage = cuda.shared.array(shape=temp_storage_bytes, dtype="uint8")
         thread_data = cuda.local.array(shape=items_per_thread, dtype=dtype)
         block_load(temp_storage, d_input, thread_data)

python/cuda_cooperative/tests/test_block_store.py

@@ -44,7 +44,6 @@ def input_index(i):
 
     @cuda.jit(link=block_store.files)
     def kernel(d_input, d_output):
-        tid = cuda.threadIdx.x
         temp_storage = cuda.shared.array(shape=temp_storage_bytes, dtype="uint8")
         thread_data = cuda.local.array(shape=items_per_thread, dtype=dtype)
         for i in range(items_per_thread):

python/cuda_parallel/cuda/parallel/__init__.py

@@ -2,5 +2,4 @@
 #
 # SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 
-import cuda.parallel.experimental
-from cuda.parallel._version import __version__
+from cuda.parallel._version import __version__ as __version__

python/cuda_parallel/cuda/parallel/experimental/iterators.py

@@ -19,13 +19,11 @@ def CacheModifiedInputIterator(device_array, modifier):
 
 def ConstantIterator(value):
     """Python facade (similar to itertools.repeat) for C++ Random Access ConstantIterator."""
-    value_type = value.dtype
     return _iterators.ConstantIterator(value)
 
 
 def CountingIterator(offset):
     """Python facade (similar to itertools.count) for C++ Random Access CountingIterator."""
-    value_type = offset.dtype
     return _iterators.CountingIterator(offset)
 
 

python/cuda_parallel/setup.py

@@ -15,8 +15,10 @@
 project_path = os.path.abspath(os.path.dirname(__file__))
 cccl_path = os.path.abspath(os.path.join(project_path, "..", ".."))
 cccl_headers = [["cub", "cub"], ["libcudacxx", "include"], ["thrust", "thrust"]]
+__version__ = None
 with open(os.path.join(project_path, "cuda", "parallel", "_version.py")) as f:
     exec(f.read())
+assert __version__ is not None
 ver = __version__
 del __version__