From d36a5bbde8b26d486e9c9f0adb275ded0beca4e2 Mon Sep 17 00:00:00 2001 From: Jason Pascucci Date: Tue, 17 Nov 2026 13:37:34 -0800 Subject: [PATCH] test: document tests skipped in every CI configuration refs #2077 Add CI coverage comments for all 148 skipped test cases reported in issue #2077. Identify affected parameter values where other cases run in CI, and annotate applicable CUDA 12 counterparts. Comments only; test behavior is unchanged. Validation: verified all reported cases are annotated, executable code is unchanged, and git diff --check passes. --- cuda_bindings_12/tests/nvml/test_device.py | 10 ++++++++++ cuda_bindings_12/tests/nvml/test_pynvml.py | 4 ++++ cuda_bindings_12/tests/test_graphics_apis.py | 1 + cuda_bindings_13/tests/legacy_api/test_legacy_cuda.py | 1 + cuda_bindings_13/tests/nvml/test_device.py | 11 +++++++++++ cuda_bindings_13/tests/nvml/test_pci.py | 1 + cuda_bindings_13/tests/nvml/test_pynvml.py | 3 +++ cuda_bindings_13/tests/nvml/test_unit.py | 1 + cuda_bindings_13/tests/test_cuda.py | 2 ++ cuda_bindings_13/tests/test_graphics_apis.py | 1 + cuda_bindings_13/tests/test_utils.py | 1 + .../tests/example_tests/test_sample_environment.py | 1 + cuda_core/tests/example_tests/test_samples.py | 7 +++++++ cuda_core/tests/graph/test_graph_definition.py | 5 +++++ cuda_core/tests/graph/test_graph_node_update.py | 2 ++ cuda_core/tests/memory/test_managed_ops.py | 1 + cuda_core/tests/memory_ipc/test_errors.py | 6 ++++++ cuda_core/tests/memory_ipc/test_leaks.py | 4 ++++ cuda_core/tests/memory_ipc/test_send_buffers.py | 2 ++ cuda_core/tests/memory_ipc/test_vmm_fabric.py | 9 +++++++++ cuda_core/tests/memory_ipc/test_workerpool.py | 2 ++ cuda_core/tests/system/test_system_device.py | 3 +++ cuda_core/tests/test_checkpoint.py | 2 ++ cuda_core/tests/test_linker.py | 3 +++ cuda_core/tests/test_managed_memory_warning.py | 2 ++ cuda_core/tests/test_memory.py | 2 ++ cuda_core/tests/test_memory_peer_access.py | 1 + cuda_core/tests/test_module.py | 1 + cuda_core/tests/test_object_protocols.py | 3 +++ cuda_core/tests/test_program.py | 2 ++ cuda_core/tests/test_tensor_map.py | 6 ++++++ cuda_core/tests/test_utils.py | 10 ++++++++++ cuda_pathfinder/tests/test_driver_lib_loading.py | 2 ++ cuda_pathfinder/tests/test_load_nvidia_dynamic_lib.py | 2 ++ 34 files changed, 114 insertions(+) diff --git a/cuda_bindings_12/tests/nvml/test_device.py b/cuda_bindings_12/tests/nvml/test_device.py index e2d761a79e9..3b43efeb7a7 100644 --- a/cuda_bindings_12/tests/nvml/test_device.py +++ b/cuda_bindings_12/tests/nvml/test_device.py @@ -70,6 +70,7 @@ def test_grid_licensable_features(all_devices): nvml.GridLicenseExpiry(feature.license_expiry) +# This test is not currently run in any CI job def test_get_nv_link_supported_bw_modes(all_devices, subtests): for device in all_devices: with subtests.test(device_index=nvml.device_get_index(device)): @@ -101,6 +102,7 @@ def test_device_get_unrepairable_memory_flag(all_devices, subtests): assert isinstance(status, int) +# This test is not currently run in any CI job def test_device_vgpu_get_heterogeneous_mode(all_devices, subtests): for device in all_devices: with subtests.test(device_index=nvml.device_get_index(device)): @@ -127,6 +129,7 @@ def test_get_power_management_limit(all_devices, subtests): nvml.device_get_power_management_limit(device) +# This test is not currently run in any CI job def test_set_power_management_limit(all_devices, subtests): for device in all_devices: with ( @@ -203,6 +206,7 @@ def _check_vgpu_type_id_list(type_ids): assert ids == [] +# This test is not currently run in any CI job @pytest.mark.agent_authored(model="claude-sonnet-5") def test_device_get_supported_vgpus_no_namerror(all_devices, subtests): """device_get_supported_vgpus must not raise a NameError, whether or not @@ -215,6 +219,7 @@ def test_device_get_supported_vgpus_no_namerror(all_devices, subtests): _check_vgpu_type_id_list(vgpu_type_ids) +# This test is not currently run in any CI job @pytest.mark.agent_authored(model="claude-sonnet-5") def test_device_get_creatable_vgpus_no_namerror(all_devices, subtests): """device_get_creatable_vgpus must not raise a NameError, whether or not @@ -227,6 +232,7 @@ def test_device_get_creatable_vgpus_no_namerror(all_devices, subtests): _check_vgpu_type_id_list(vgpu_type_ids) +# This test is not currently run in any CI job @pytest.mark.agent_authored(model="claude-sonnet-5") def test_device_get_active_vgpus_no_namerror(all_devices, subtests): """device_get_active_vgpus must not raise a NameError, whether or not @@ -260,6 +266,7 @@ def _require_mig_enabled(device): pytest.skip(f"MIG is not enabled on device {device}") +# This test is not currently run in any CI job @pytest.mark.agent_authored(model="claude-sonnet-5") def test_device_get_gpu_instances_empty_result(all_devices, subtests): """device_get_gpu_instances must not raise ValueError: Invalid shape in @@ -293,6 +300,7 @@ def test_device_get_gpu_instances_empty_result(all_devices, subtests): assert len(set(handles)) == len(handles) +# This test is not currently run in any CI job @pytest.mark.agent_authored(model="claude-sonnet-5") def test_gpu_instance_get_compute_instances_empty_result(all_devices, subtests): """gpu_instance_get_compute_instances must not raise ValueError: Invalid @@ -340,6 +348,7 @@ def test_gpu_instance_get_compute_instances_empty_result(all_devices, subtests): assert len(set(handles)) == len(handles) +# This test is not currently run in any CI job @pytest.mark.agent_authored(model="claude-sonnet-5") def test_device_get_vgpu_utilization_sized_array(all_devices, subtests): """device_get_vgpu_utilization must return an array sized to the real @@ -375,6 +384,7 @@ def test_device_get_vgpu_utilization_sized_array(all_devices, subtests): assert len(set(vgpu_instances)) == len(vgpu_instances) +# This test is not currently run in any CI job @pytest.mark.agent_authored(model="claude-sonnet-5") def test_device_get_vgpu_process_utilization_returns_array(all_devices, subtests): """device_get_vgpu_process_utilization must return a correctly-sized array diff --git a/cuda_bindings_12/tests/nvml/test_pynvml.py b/cuda_bindings_12/tests/nvml/test_pynvml.py index e3ea533dd83..2912605e811 100644 --- a/cuda_bindings_12/tests/nvml/test_pynvml.py +++ b/cuda_bindings_12/tests/nvml/test_pynvml.py @@ -40,6 +40,7 @@ def test_system_get_driver_version(nvml_init): assert tuple(int(x) for x in vsn.split(".")[:2]) > (0, 0) +# This test is not currently run in any CI job def test_device_get_attributes(mig_handles): # nvmlDeviceGetAttributes requires MIG device handle @@ -61,6 +62,7 @@ def test_device_get_handle_by_pci_bus_id(ngpus, pci_info): assert len(handles) == ngpus +# This test is not currently run in any CI job @pytest.mark.parametrize("scope", [nvml.AffinityScope.NODE, nvml.AffinityScope.SOCKET]) @pytest.mark.skipif(util.is_wsl() or util.is_windows(), reason="Not supported on WSL or Windows") def test_device_get_memory_affinity(handles, scope): @@ -72,6 +74,7 @@ def test_device_get_memory_affinity(handles, scope): assert len(node_set) == size +# This test is not currently run in any CI job @pytest.mark.parametrize("scope", [nvml.AffinityScope.NODE, nvml.AffinityScope.SOCKET]) @pytest.mark.skipif(util.is_wsl() or util.is_windows(), reason="Not supported on WSL or Windows") def test_device_get_cpu_affinity_within_scope(handles, scope): @@ -245,6 +248,7 @@ def test_device_get_pcie_throughput(ngpus, handles): # Test pynvml.nvmlDeviceGetNvLinkRemotePciInfo +# This test is not currently run in any CI job @pytest.mark.parametrize( "cap_type", [ diff --git a/cuda_bindings_12/tests/test_graphics_apis.py b/cuda_bindings_12/tests/test_graphics_apis.py index 365bbe1fb2c..82e65f77d8a 100644 --- a/cuda_bindings_12/tests/test_graphics_apis.py +++ b/cuda_bindings_12/tests/test_graphics_apis.py @@ -9,6 +9,7 @@ from cuda.bindings import runtime as cudart +# This test is not currently run in any CI job def test_graphics_api_smoketest(): if sys.platform.startswith("linux") and not (os.environ.get("DISPLAY") or os.environ.get("WAYLAND_DISPLAY")): pytest.skip("No display available for the OpenGL smoke test") diff --git a/cuda_bindings_13/tests/legacy_api/test_legacy_cuda.py b/cuda_bindings_13/tests/legacy_api/test_legacy_cuda.py index 47043cf74f6..45781a7ce23 100644 --- a/cuda_bindings_13/tests/legacy_api/test_legacy_cuda.py +++ b/cuda_bindings_13/tests/legacy_api/test_legacy_cuda.py @@ -31,6 +31,7 @@ def callableBinary(name): return shutil.which(name) is not None +# This test is not currently run in any CI job @pytest.mark.skipif(True, reason="Always skip!") def test_always_skip(): pass diff --git a/cuda_bindings_13/tests/nvml/test_device.py b/cuda_bindings_13/tests/nvml/test_device.py index 41c9e47a420..a1522d72f7c 100644 --- a/cuda_bindings_13/tests/nvml/test_device.py +++ b/cuda_bindings_13/tests/nvml/test_device.py @@ -81,6 +81,7 @@ def test_get_handle_by_uuidv(all_devices, subtests): assert new_handle == device +# This test is not currently run in any CI job def test_get_nv_link_supported_bw_modes(all_devices, subtests): for device in all_devices: with subtests.test(device_index=nvml.device_get_index(device)): @@ -119,6 +120,7 @@ def test_device_get_unrepairable_memory_flag(all_devices, subtests): assert isinstance(status, int) +# This test is not currently run in any CI job def test_device_vgpu_get_heterogeneous_mode(all_devices, subtests): for device in all_devices: with subtests.test(device_index=nvml.device_get_index(device)): @@ -138,6 +140,7 @@ def test_read_prm_counters(all_devices, subtests): assert len(read_counters) == 5 +# This test is not currently run in any CI job @pytest.mark.thread_unsafe(reason="API appears to be thread-unsafe (2026-06)") def test_read_write_prm(all_devices, subtests): for device in all_devices: @@ -160,6 +163,7 @@ def test_get_power_management_limit(all_devices, subtests): nvml.device_get_power_management_limit(device) +# This test is not currently run in any CI job def test_set_power_management_limit(all_devices, subtests): for device in all_devices: with ( @@ -241,6 +245,7 @@ def _check_vgpu_type_id_list(type_ids): assert ids == [] +# This test is not currently run in any CI job @pytest.mark.agent_authored(model="claude-sonnet-5") def test_device_get_supported_vgpus_no_namerror(all_devices, subtests): """device_get_supported_vgpus must not raise a NameError, whether or not @@ -253,6 +258,7 @@ def test_device_get_supported_vgpus_no_namerror(all_devices, subtests): _check_vgpu_type_id_list(vgpu_type_ids) +# This test is not currently run in any CI job @pytest.mark.agent_authored(model="claude-sonnet-5") def test_device_get_creatable_vgpus_no_namerror(all_devices, subtests): """device_get_creatable_vgpus must not raise a NameError, whether or not @@ -265,6 +271,7 @@ def test_device_get_creatable_vgpus_no_namerror(all_devices, subtests): _check_vgpu_type_id_list(vgpu_type_ids) +# This test is not currently run in any CI job @pytest.mark.agent_authored(model="claude-sonnet-5") def test_device_get_active_vgpus_no_namerror(all_devices, subtests): """device_get_active_vgpus must not raise a NameError, whether or not @@ -298,6 +305,7 @@ def _require_mig_enabled(device): pytest.skip(f"MIG is not enabled on device {device}") +# This test is not currently run in any CI job @pytest.mark.agent_authored(model="claude-sonnet-5") def test_device_get_gpu_instances_empty_result(all_devices, subtests): """device_get_gpu_instances must not raise ValueError: Invalid shape in @@ -331,6 +339,7 @@ def test_device_get_gpu_instances_empty_result(all_devices, subtests): assert len(set(handles)) == len(handles) +# This test is not currently run in any CI job @pytest.mark.agent_authored(model="claude-sonnet-5") def test_gpu_instance_get_compute_instances_empty_result(all_devices, subtests): """gpu_instance_get_compute_instances must not raise ValueError: Invalid @@ -378,6 +387,7 @@ def test_gpu_instance_get_compute_instances_empty_result(all_devices, subtests): assert len(set(handles)) == len(handles) +# This test is not currently run in any CI job @pytest.mark.agent_authored(model="claude-sonnet-5") def test_device_get_vgpu_utilization_sized_array(all_devices, subtests): """device_get_vgpu_utilization must return an array sized to the real @@ -413,6 +423,7 @@ def test_device_get_vgpu_utilization_sized_array(all_devices, subtests): assert len(set(vgpu_instances)) == len(vgpu_instances) +# This test is not currently run in any CI job @pytest.mark.agent_authored(model="claude-sonnet-5") def test_device_get_vgpu_process_utilization_returns_array(all_devices, subtests): """device_get_vgpu_process_utilization must return a correctly-sized array diff --git a/cuda_bindings_13/tests/nvml/test_pci.py b/cuda_bindings_13/tests/nvml/test_pci.py index 7abff4bc45b..a340bf9bf1d 100644 --- a/cuda_bindings_13/tests/nvml/test_pci.py +++ b/cuda_bindings_13/tests/nvml/test_pci.py @@ -36,6 +36,7 @@ def test_bridge_chip_hierarchy_t(): assert isinstance(hierarchy.bridge_chip_info, nvml.BridgeChipInfo) +# This test is not currently run in any CI job def test_bridge_chip_info(all_devices, subtests): for device in all_devices: with subtests.test(device_index=nvml.device_get_index(device)): diff --git a/cuda_bindings_13/tests/nvml/test_pynvml.py b/cuda_bindings_13/tests/nvml/test_pynvml.py index 57b0b8c0f1c..733daf8c5e6 100644 --- a/cuda_bindings_13/tests/nvml/test_pynvml.py +++ b/cuda_bindings_13/tests/nvml/test_pynvml.py @@ -41,6 +41,7 @@ def test_system_get_driver_version(nvml_init): assert tuple(int(x) for x in vsn.split(".")[:2]) > (0, 0) +# This test is not currently run in any CI job def test_device_get_attributes(mig_handles): # nvmlDeviceGetAttributes requires MIG device handle @@ -73,6 +74,7 @@ def test_device_get_handle_by_pci_bus_id(ngpus): assert len(handles) == ngpus +# This test is not currently run in any CI job @pytest.mark.parametrize("scope", [nvml.AffinityScope.NODE, nvml.AffinityScope.SOCKET]) @pytest.mark.skipif(IS_WSL or IS_WINDOWS, reason="Not supported on WSL or Windows") def test_device_get_memory_affinity(handles, scope, subtests): @@ -85,6 +87,7 @@ def test_device_get_memory_affinity(handles, scope, subtests): assert len(node_set) == size +# This test is not currently run in any CI job @pytest.mark.parametrize("scope", [nvml.AffinityScope.NODE, nvml.AffinityScope.SOCKET]) @pytest.mark.skipif(IS_WSL or IS_WINDOWS, reason="Not supported on WSL or Windows") def test_device_get_cpu_affinity_within_scope(handles, scope, subtests): diff --git a/cuda_bindings_13/tests/nvml/test_unit.py b/cuda_bindings_13/tests/nvml/test_unit.py index fb8c2552839..65dc40d7998 100644 --- a/cuda_bindings_13/tests/nvml/test_unit.py +++ b/cuda_bindings_13/tests/nvml/test_unit.py @@ -15,6 +15,7 @@ def unit_handles(nvml_init): return [nvml.unit_get_handle_by_index(i) for i in range(count)] +# This test is not currently run in any CI job @pytest.mark.agent_authored(model="claude-sonnet-4-6") def test_unit_get_devices_returns_array(unit_handles): """Regression test: unit_get_devices must not crash due to double-pointer cast. diff --git a/cuda_bindings_13/tests/test_cuda.py b/cuda_bindings_13/tests/test_cuda.py index a2ad692b4d3..c26d6837dba 100644 --- a/cuda_bindings_13/tests/test_cuda.py +++ b/cuda_bindings_13/tests/test_cuda.py @@ -39,6 +39,7 @@ def node_array(nodes): return (ctypes.c_void_p * len(nodes))(*nodes) +# This test is not currently run in any CI job @pytest.mark.skipif(True, reason="Always skip!") def test_always_skip(): pass @@ -88,6 +89,7 @@ def test_cuda_memcpy(): driver.mem_free_v2(dptr) +# This test is not currently run in any CI job @pytest.mark.agent_authored(model="claude-sonnet-5-5") @pytest.mark.skipif(driver_version_less_than(13041), reason="cuGraphAddNode_v3 was added in CUDA 13.4.1") def test_graph_add_node_v3_returns_the_node(): diff --git a/cuda_bindings_13/tests/test_graphics_apis.py b/cuda_bindings_13/tests/test_graphics_apis.py index 9b357620db8..092ab1a32ae 100644 --- a/cuda_bindings_13/tests/test_graphics_apis.py +++ b/cuda_bindings_13/tests/test_graphics_apis.py @@ -95,6 +95,7 @@ def _gl_context(): win.close() +# This test is not currently run in any CI job @pytest.mark.parametrize( "flags", [ diff --git a/cuda_bindings_13/tests/test_utils.py b/cuda_bindings_13/tests/test_utils.py index 84f3b7ac1d0..adddfe764c6 100644 --- a/cuda_bindings_13/tests/test_utils.py +++ b/cuda_bindings_13/tests/test_utils.py @@ -197,6 +197,7 @@ def test_check_nvvm_compiler_options_arch_detection(): assert check_nvvm_compiler_options(["-arch=compute_99999"]) is False +# This test is not currently run in any CI job def test_check_nvvm_compiler_options_no_libnvvm(): if _libnvvm_available: pytest.skip("libNVVM is available; this test targets the fallback path") diff --git a/cuda_core/tests/example_tests/test_sample_environment.py b/cuda_core/tests/example_tests/test_sample_environment.py index 8db28f41fb5..3d5fc92b8f4 100644 --- a/cuda_core/tests/example_tests/test_sample_environment.py +++ b/cuda_core/tests/example_tests/test_sample_environment.py @@ -14,6 +14,7 @@ import pytest +# This test is not currently run in any CI job @pytest.mark.parametrize( ("distribution_name", "expected_source", "expected_version"), ( diff --git a/cuda_core/tests/example_tests/test_samples.py b/cuda_core/tests/example_tests/test_samples.py index 8ccb6e1861d..2ae42b4b6d7 100644 --- a/cuda_core/tests/example_tests/test_samples.py +++ b/cuda_core/tests/example_tests/test_samples.py @@ -28,6 +28,13 @@ # Samples launch full GPU workloads in their own subprocess, so keep them # serialized when the suite is run under pytest-run-parallel. +# For sample_key values: cuda_core/binarySearch, cuda_core/cudaComputeLambdas, cuda_core/customPyTorchKernel, +# cuda_core/customTensorFlowKernel, cuda_core/fftSignalAnalysis, cuda_core/glInteropFluidNumbaCudaMlir, +# cuda_core/glInteropFluid, cuda_core/glInteropMipmapLod, cuda_core/glInteropPlasma, cuda_core/greenContext, +# cuda_core/matrixMulSharedMem, cuda_core/multiGPUGradientAverage, cuda_core/numpyVsCupy, cuda_core/pageRank, +# cuda_core/parallelReduction, cuda_core/prefixSum, cuda_core/processCheckpoint, cuda_core/reductionMultiBlockCG, +# cuda_core/simplePrint, cuda_core/tmaTensorMap: +# This test is not currently run in any CI job @pytest.mark.parallel_threads_limit(1) @pytest.mark.parametrize("sample_key", _SAMPLES) @pytest.mark.samples diff --git a/cuda_core/tests/graph/test_graph_definition.py b/cuda_core/tests/graph/test_graph_definition.py index 0d4f2c8f843..3919d7b377e 100644 --- a/cuda_core/tests/graph/test_graph_definition.py +++ b/cuda_core/tests/graph/test_graph_definition.py @@ -689,6 +689,8 @@ def test_node_type_preserved_by_pred_succ(node_spec): assert matched[0] is node +# For node_spec values empty and host_callback_cfunc: +# This test is not currently run in any CI job def test_node_attrs(node_spec): """Type-specific attributes have expected values after construction.""" spec, g, node, expected_attrs = node_spec @@ -702,6 +704,8 @@ def test_node_attrs(node_spec): assert actual == expected, f"{spec.name}.{attr}: expected {expected}, got {actual}" +# For node_spec values empty and host_callback_cfunc: +# This test is not currently run in any CI job def test_node_attrs_preserved_by_nodes(node_spec): """Type-specific attributes survive round-trip through graph_definition.nodes().""" spec, g, node, expected_attrs = node_spec @@ -922,6 +926,7 @@ def test_alloc_peer_access(mempool_device_x2): assert d1.device_id in node.peer_access +# This test is not currently run in any CI job @pytest.mark.agent_authored(model="gpt-5.6-sol") def test_alloc_memory_type_host(init_cuda): """HOST graph alloc nodes reconstruct memory_type from the driver, not the Python argument.""" diff --git a/cuda_core/tests/graph/test_graph_node_update.py b/cuda_core/tests/graph/test_graph_node_update.py index e8e977771d9..cfb05a0481b 100644 --- a/cuda_core/tests/graph/test_graph_node_update.py +++ b/cuda_core/tests/graph/test_graph_node_update.py @@ -993,6 +993,8 @@ def test_failed_definition_node_update_preserves_state( case.assert_exec_uses(graph, case.original) +# For the host-callback-ctypes parameter case: +# This test is not currently run in any CI job @pytest.mark.agent_authored(model="gpt-5.6") def test_definition_node_update_rejects_wrong_type( definition_update_case, diff --git a/cuda_core/tests/memory/test_managed_ops.py b/cuda_core/tests/memory/test_managed_ops.py index 0500ad57df8..bf6130691d7 100644 --- a/cuda_core/tests/memory/test_managed_ops.py +++ b/cuda_core/tests/memory/test_managed_ops.py @@ -392,6 +392,7 @@ def test_preferred_location_roundtrip(self, location_ops_device, external_manage buf.preferred_location = None assert buf.preferred_location is None + # This test is not currently run in any CI job @pytest.mark.thread_unsafe(reason="external_managed_buffer is shared between threads") def test_preferred_location_roundtrip_host_numa(self, location_ops_device): """Host(numa_id=N) round-trips correctly on CUDA 13 builds.""" diff --git a/cuda_core/tests/memory_ipc/test_errors.py b/cuda_core/tests/memory_ipc/test_errors.py index a036c87363b..e7a00327576 100644 --- a/cuda_core/tests/memory_ipc/test_errors.py +++ b/cuda_core/tests/memory_ipc/test_errors.py @@ -90,6 +90,8 @@ def test_register_rejects_non_ipc_memory_resource(mempool_device): DeviceMemoryResource.from_registry(key) +# For the VirtualMR parameter case: +# This test is not currently run in any CI job @pytest.mark.human_authored def test_register_rejects_mismatched_uuid(ipc_memory_resource): """register() rejects a UUID that does not match the resource's own. @@ -138,6 +140,8 @@ def test_ipc_allocation_handle_state_tracks_close(): os.close(write_fd) +# For the VirtualMR parameter case: +# This test is not currently run in any CI job @pytest.mark.agent_authored(model="gpt-5.6") def test_closed_ipc_allocation_handle_rejected_before_registry_hit(ipc_device, ipc_memory_resource): mr = ipc_memory_resource @@ -330,6 +334,8 @@ def ASSERT(self, exc_type, exc_msg): assert exc_msg.startswith("Argument 'ipc_descriptor' has incorrect type") +# For the inherited test_main[VirtualMR] case: +# This test is not currently run in any CI job class TestDanglingBuffer(ChildErrorHarness): """ Error when importing a buffer object without registering its memory diff --git a/cuda_core/tests/memory_ipc/test_leaks.py b/cuda_core/tests/memory_ipc/test_leaks.py index 50240f57731..92023b8dc19 100644 --- a/cuda_core/tests/memory_ipc/test_leaks.py +++ b/cuda_core/tests/memory_ipc/test_leaks.py @@ -26,6 +26,8 @@ pytestmark = pytest.mark.thread_unsafe(reason="Tests number of fds which is shared.") +# For the VirtualMR parameter case: +# This test is not currently run in any CI job @pytest.mark.flaky(reruns=2) @skip_if_unrunnable def test_alloc_handle(ipc_memory_resource): @@ -97,6 +99,8 @@ def _get_buffer_desc(mr, stream): return mr.allocate(NBYTES, stream=stream).ipc_descriptor +# For VirtualMR with getobject=alloc_handle (all launchers): +# This test is not currently run in any CI job @pytest.mark.flaky(reruns=2) @skip_if_unrunnable @pytest.mark.parametrize( diff --git a/cuda_core/tests/memory_ipc/test_send_buffers.py b/cuda_core/tests/memory_ipc/test_send_buffers.py index 55df245d577..89efd2d0af9 100644 --- a/cuda_core/tests/memory_ipc/test_send_buffers.py +++ b/cuda_core/tests/memory_ipc/test_send_buffers.py @@ -85,6 +85,8 @@ class TestIpcReexport: re-exported from B to C. """ + # For the VirtualMR parameter case: + # This test is not currently run in any CI job @pytest.mark.flaky(reruns=2) def test_main(self, ipc_device, ipc_memory_resource): # Set up the device. diff --git a/cuda_core/tests/memory_ipc/test_vmm_fabric.py b/cuda_core/tests/memory_ipc/test_vmm_fabric.py index bb0491c4dcb..f3b444c0955 100644 --- a/cuda_core/tests/memory_ipc/test_vmm_fabric.py +++ b/cuda_core/tests/memory_ipc/test_vmm_fabric.py @@ -113,6 +113,7 @@ def _round_trip(mode, owner_uuid, payload): return result +# This test is not currently run in any CI job @pytest.mark.agent_authored(model="gpt-6.1-sol") def test_export_shape(vmm_fabric_device): device = vmm_fabric_device @@ -173,6 +174,7 @@ def test_descriptor_pickle_and_validation(): descriptor.from_handles(handles, sizes, size=size) +# This test is not currently run in any CI job @pytest.mark.parametrize("mode", ["queue", "pipe", "argument", "pool"]) @pytest.mark.parametrize("send_buffer", [False, True], ids=["descriptor", "buffer"]) @pytest.mark.thread_unsafe(reason="counts process-global file descriptors") @@ -192,6 +194,7 @@ def test_cross_process(vmm_fabric_device, mode, send_buffer): assert len(os.listdir("/proc/self/fd")) == before +# This test is not currently run in any CI job @pytest.mark.agent_authored(model="gpt-6.1-sol") def test_second_gpu_and_wrong_owner(vmm_fabric_device_x2): owner, peer = vmm_fabric_device_x2 @@ -236,6 +239,7 @@ def test_handle_type_mismatch(vmm_ipc_device): # Fixed patterns: a random first byte of 0x01 would read as the driver's own handle type and take another error path. +# This test is not currently run in any CI job @pytest.mark.parametrize("handle", [bytes(64), b"\xff" * 64], ids=["zero", "ones"]) @pytest.mark.agent_authored(model="gpt-6.1-sol") def test_garbage(vmm_fabric_device, handle): @@ -248,6 +252,7 @@ def test_garbage(vmm_fabric_device, handle): _check_import_error(info.value) +# This test is not currently run in any CI job @pytest.mark.agent_authored(model="gpt-6.1-sol") def test_stale_descriptor(vmm_fabric_device): device = vmm_fabric_device @@ -259,6 +264,7 @@ def test_stale_descriptor(vmm_fabric_device): _check_import_error(info.value) +# This test is not currently run in any CI job @pytest.mark.agent_authored(model="gpt-6.1-sol") def test_reexport_imported_derivatives(vmm_fabric_device): device = vmm_fabric_device @@ -276,6 +282,7 @@ def test_reexport_imported_derivatives(vmm_fabric_device): pickle.dumps(imported) +# This test is not currently run in any CI job @pytest.mark.agent_authored(model="gpt-6.1-sol") def test_buffer_pickle(vmm_fabric_device): device = vmm_fabric_device @@ -303,6 +310,7 @@ def _child_stale_queue(queue, ready, outgoing): outgoing.close() +# This test is not currently run in any CI job @pytest.mark.agent_authored(model="gpt-6.1-sol") def test_queued_buffer_does_not_pin(vmm_fabric_device): device = vmm_fabric_device @@ -342,6 +350,7 @@ def test_queued_buffer_does_not_pin(vmm_fabric_device): assert _read(fresh) == b"\x58" * 16 +# This test is not currently run in any CI job @pytest.mark.skipif(IS_WINDOWS or IS_WSL, reason="Fabric IPC requires Linux") @pytest.mark.agent_authored(model="gpt-6.1-sol") def test_unsupported_fabric_construction(init_cuda): diff --git a/cuda_core/tests/memory_ipc/test_workerpool.py b/cuda_core/tests/memory_ipc/test_workerpool.py index 7eeb2415d09..ef7916ef2ae 100644 --- a/cuda_core/tests/memory_ipc/test_workerpool.py +++ b/cuda_core/tests/memory_ipc/test_workerpool.py @@ -129,6 +129,8 @@ def init_worker(mrs): # Passing mrs implicitly registers them. pass + # For VirtualMR with nmrs=1 or nmrs=3: + # This test is not currently run in any CI job @pytest.mark.flaky(reruns=2) @pytest.mark.parametrize("nmrs", (1, NMRS)) def test_main(self, ipc_device, ipc_memory_resource_factory, nmrs): diff --git a/cuda_core/tests/system/test_system_device.py b/cuda_core/tests/system/test_system_device.py index 80be43e75c7..295beee4206 100644 --- a/cuda_core/tests/system/test_system_device.py +++ b/cuda_core/tests/system/test_system_device.py @@ -129,6 +129,7 @@ def test_affinity(subtests): assert isinstance(affinity, list) +# This test is not currently run in any CI job def test_numa_node_id(subtests): for device in system.Device.get_all_devices(): with subtests.test(device_index=device.index): @@ -339,6 +340,7 @@ def test_device_pci_bus_id(): assert new_device.index == device.index +# This test is not currently run in any CI job @pytest.mark.skipif(helpers.IS_WSL or helpers.IS_WINDOWS, reason="Device attributes not supported on WSL or Windows") def test_device_attributes(subtests): for device in system.Device.get_all_devices(): @@ -635,6 +637,7 @@ def test_get_inforom_version(subtests): inforom.validate() +# This test is not currently run in any CI job def test_auto_boosted_clocks_enabled(subtests): for device in system.Device.get_all_devices(): with subtests.test(device_index=device.index): diff --git a/cuda_core/tests/test_checkpoint.py b/cuda_core/tests/test_checkpoint.py index a52d1b7dabf..a9f64e5aa9a 100644 --- a/cuda_core/tests/test_checkpoint.py +++ b/cuda_core/tests/test_checkpoint.py @@ -579,6 +579,7 @@ class TestCheckpointGpuMigration: NVBug 5437334). """ + # This test is not currently run in any CI job def test_rotation_migrates_context(self): """Rotate context through all GPUs and back to the origin. @@ -589,6 +590,7 @@ def test_rotation_migrates_context(self): """ _run_checkpoint_scenario_or_skip("rotation_migrates_context", timeout=180) + # This test is not currently run in any CI job def test_swap_identical_gpus(self): """Swap context between two GPUs of the same chip type. diff --git a/cuda_core/tests/test_linker.py b/cuda_core/tests/test_linker.py index 5b0bc0a9cbc..aaec64f18a3 100644 --- a/cuda_core/tests/test_linker.py +++ b/cuda_core/tests/test_linker.py @@ -127,6 +127,7 @@ def test_linker_link_ptx_nvjitlink(compile_ltoir_functions): assert linked_code.name == options.name +# This test is not currently run in any CI job @pytest.mark.skipif(not is_culink_backend, reason="nvjitlink requires lto for ptx linking") def test_linker_link_ptx_culink(compile_ptx_functions): options = LinkerOptions(arch=ARCH) @@ -311,6 +312,7 @@ def test_linker_handle(compile_ptx_functions): assert int(handle) != 0 +# This test is not currently run in any CI job @pytest.mark.agent_authored(model="gpt-5") @pytest.mark.skipif(not is_culink_backend, reason="driver backend regression test") def test_driver_linker_lifetime_no_heap_corruption(compile_ptx_functions): @@ -330,6 +332,7 @@ def test_driver_linker_lifetime_no_heap_corruption(compile_ptx_functions): linker.close() +# This test is not currently run in any CI job @pytest.mark.agent_authored(model="gpt-5") @pytest.mark.skipif(not is_culink_backend, reason="driver backend regression test") def test_driver_linker_preserves_error_log_after_close(init_cuda): diff --git a/cuda_core/tests/test_managed_memory_warning.py b/cuda_core/tests/test_managed_memory_warning.py index 811bb6c3ce5..60e99439b26 100644 --- a/cuda_core/tests/test_managed_memory_warning.py +++ b/cuda_core/tests/test_managed_memory_warning.py @@ -65,6 +65,7 @@ def test_default_pool_error_without_concurrent_access(device_without_concurrent_ raise +# This test is not currently run in any CI job @requires_cuda_13 def test_warning_emitted(device_without_concurrent_managed_access): """ManagedMemoryResource emits a warning when concurrent managed access is unsupported.""" @@ -85,6 +86,7 @@ def test_warning_emitted(device_without_concurrent_managed_access): mr.close() +# This test is not currently run in any CI job @requires_cuda_13 def test_warning_emitted_only_once(device_without_concurrent_managed_access): """Warning fires only once even when multiple ManagedMemoryResources are created.""" diff --git a/cuda_core/tests/test_memory.py b/cuda_core/tests/test_memory.py index ac1099244ab..9caac3f7891 100644 --- a/cuda_core/tests/test_memory.py +++ b/cuda_core/tests/test_memory.py @@ -2822,6 +2822,8 @@ def test_pinned_mr_numa_id_negative_error(init_cuda): PinnedMemoryResource(PinnedMemoryResourceOptions(numa_id=-42)) +# For ManagedMR with ipc_enabled=True (all property names): +# This test is not currently run in any CI job @pytest.mark.parametrize("ipc_enabled", [True, False]) @pytest.mark.parametrize( "property_name,expected_type", diff --git a/cuda_core/tests/test_memory_peer_access.py b/cuda_core/tests/test_memory_peer_access.py index 6c392404ac8..7bce09245a7 100644 --- a/cuda_core/tests/test_memory_peer_access.py +++ b/cuda_core/tests/test_memory_peer_access.py @@ -57,6 +57,7 @@ def test_peer_access_basic(mempool_device_x2): zero_on_dev0.copy_from(buf_on_dev1, stream=stream_on_dev0) +# This test is not currently run in any CI job def test_peer_access_transitions(mempool_device_x3): """Advanced tests for dmr.peer_accessible_by.""" diff --git a/cuda_core/tests/test_module.py b/cuda_core/tests/test_module.py index 0e2ac833124..c6d42e0a14e 100644 --- a/cuda_core/tests/test_module.py +++ b/cuda_core/tests/test_module.py @@ -662,6 +662,7 @@ def test_occupancy_max_potential_block_size_constant(get_saxpy_kernel_cubin, blo kernel.occupancy.max_potential_block_size(invalid_dsmem, block_size_limit) +# This test is not currently run in any CI job @pytest.mark.skipif(numba is None, reason="Test requires numba to be installed") @pytest.mark.parametrize("block_size_limit", [32, 64, 96, 120, 128, 277, 0]) def test_occupancy_max_potential_block_size_b2dsize(get_saxpy_kernel_cubin, block_size_limit): diff --git a/cuda_core/tests/test_object_protocols.py b/cuda_core/tests/test_object_protocols.py index 1e1e410f807..7807b2b4fbd 100644 --- a/cuda_core/tests/test_object_protocols.py +++ b/cuda_core/tests/test_object_protocols.py @@ -908,6 +908,9 @@ def test_usable_in_weak_set(sample_object): # ============================================================================= +# For sample_alloc_node, sample_empty_node, sample_free_node, sample_memcpy_node, sample_memset_node, +# sample_object_code_ptx, and sample_program_ptx: +# This test is not currently run in any CI job @pytest.mark.parametrize("sample_object,pattern", REPR_PATTERNS, indirect=["sample_object"]) def test_repr_format(sample_object, pattern): """repr() returns a properly formatted string.""" diff --git a/cuda_core/tests/test_program.py b/cuda_core/tests/test_program.py index 6dc422c9c46..4376519a90b 100644 --- a/cuda_core/tests/test_program.py +++ b/cuda_core/tests/test_program.py @@ -172,6 +172,8 @@ def ptx_code_object(): return ptx_object_code +# For the options31 parameter case: +# This test is not currently run in any CI job @pytest.mark.parametrize( "options", [ diff --git a/cuda_core/tests/test_tensor_map.py b/cuda_core/tests/test_tensor_map.py index 3f05a2e477b..2cc348fceae 100644 --- a/cuda_core/tests/test_tensor_map.py +++ b/cuda_core/tests/test_tensor_map.py @@ -516,6 +516,7 @@ def test_replace_address_rejects_tensor_from_other_device(self, dev, skip_if_no_ with pytest.raises(ValueError, match=r"replace_address expects tensor on device 0, got 1"): desc.replace_address(buf1) + # This test is not currently run in any CI job def test_replace_address_accepts_managed_buffer_on_nonzero_device(self, init_cuda): if len(Device.get_all_devices()) < 2: pytest.skip("requires multi-GPU") @@ -609,6 +610,7 @@ def test_from_im2col_wide_rejects_tensor_from_other_device(self, init_cuda): pytest.skip("Im2col-wide requires cuda.core built with CUDA 13+") raise + # This test is not currently run in any CI job def test_from_tiled_accepts_managed_buffer_on_nonzero_device(self, init_cuda): if len(Device.get_all_devices()) < 2: pytest.skip("requires multi-GPU") @@ -738,6 +740,7 @@ def skip_if_no_im2col_wide(self, dev): pytest.skip("Im2col-wide unsupported on this driver/GPU combination") raise + # This test is not currently run in any CI job def test_from_im2col_wide_3d(self, dev, skip_if_no_im2col_wide): # 3D tensor: batch=1, width=32, channels=64 buf = dev.allocate(1 * 32 * 64 * 4, stream=dev.default_stream) @@ -752,6 +755,7 @@ def test_from_im2col_wide_3d(self, dev, skip_if_no_im2col_wide): ) assert desc is not None + # This test is not currently run in any CI job def test_from_im2col_wide_4d(self, dev, skip_if_no_im2col_wide): # NHWC layout: N=1, H=8, W=8, C=64 # Wide mode only uses scalar W corners, even with higher rank @@ -767,6 +771,7 @@ def test_from_im2col_wide_4d(self, dev, skip_if_no_im2col_wide): ) assert desc is not None + # This test is not currently run in any CI job def test_from_im2col_wide_5d(self, dev, skip_if_no_im2col_wide): # NDHWC layout: N=1, D=4, H=8, W=8, C=64 # Max rank boundary — verifies all 5 dim/stride slots are filled @@ -782,6 +787,7 @@ def test_from_im2col_wide_5d(self, dev, skip_if_no_im2col_wide): ) assert desc is not None + # This test is not currently run in any CI job def test_from_im2col_wide_rank_validation(self, dev, skip_if_no_im2col_wide): buf = dev.allocate(1024 * 4, stream=dev.default_stream) with pytest.raises(ValueError, match="Im2col-wide tensor rank must be between 3 and 5"): diff --git a/cuda_core/tests/test_utils.py b/cuda_core/tests/test_utils.py index 1e961e173ed..fc384d3dcdd 100644 --- a/cuda_core/tests/test_utils.py +++ b/cuda_core/tests/test_utils.py @@ -245,6 +245,8 @@ def gpu_array_ptr(arr): @pytest.mark.parametrize(("in_arr", "use_stream"), gpu_array_samples()) class TestViewGPU: + # For the empty parameter set ([NOTSET]): + # This test is not currently run in any CI job def test_args_viewable_as_strided_memory_gpu(self, in_arr, use_stream): # TODO: use the device fixture? dev = Device() @@ -259,6 +261,8 @@ def my_func(arr): my_func(in_arr) + # For the empty parameter set ([NOTSET]): + # This test is not currently run in any CI job def test_strided_memory_view_cpu(self, in_arr, use_stream): # TODO: use the device fixture? dev = Device() @@ -269,6 +273,8 @@ def test_strided_memory_view_cpu(self, in_arr, use_stream): view = StridedMemoryView.from_any_interface(in_arr, stream_ptr=s.handle if s else -1) self._check_view(view, in_arr, dev) + # For the empty parameter set ([NOTSET]): + # This test is not currently run in any CI job def test_strided_memory_view_init(self, in_arr, use_stream): # TODO: use the device fixture? dev = Device() @@ -338,6 +344,8 @@ def test_strided_memory_view_exposes_dlpack_c_exchange_api_capsule(): @pytest.mark.skipif(cp is None, reason="CuPy is not installed") @pytest.mark.parametrize("in_arr,use_stream", (*gpu_array_samples(),)) class TestViewCudaArrayInterfaceGPU: + # For the empty parameter set ([NOTSET]): + # This test is not currently run in any CI job def test_cuda_array_interface_gpu(self, in_arr, use_stream): # TODO: use the device fixture? dev = Device() @@ -753,6 +761,7 @@ def test_ml_dtypes_bfloat16_dlpack(init_cuda, slices): assert smv.strides == strides_in_counts +# This test is not currently run in any CI job @pytest.mark.parametrize( "slices", [ @@ -953,6 +962,7 @@ def test_array_interface_mask_rejected(): _torch_skip = pytest.mark.skipif(torch is None, reason="PyTorch is not installed") +# This test is not currently run in any CI job @_torch_skip @pytest.mark.parametrize( "dtype", diff --git a/cuda_pathfinder/tests/test_driver_lib_loading.py b/cuda_pathfinder/tests/test_driver_lib_loading.py index defca06abed..f71ab102e06 100644 --- a/cuda_pathfinder/tests/test_driver_lib_loading.py +++ b/cuda_pathfinder/tests/test_driver_lib_loading.py @@ -128,6 +128,8 @@ def test_load_lib_no_cache_does_not_dispatch_ctk_lib_to_driver_path(mocker): # --------------------------------------------------------------------------- +# For libname=nvcudla: +# This test is not currently run in any CI job @pytest.mark.parametrize("libname", sorted(_DRIVER_ONLY_LIBNAMES)) def test_real_load_driver_lib(info_summary_append, libname): """Load a real driver library in a dedicated subprocess. diff --git a/cuda_pathfinder/tests/test_load_nvidia_dynamic_lib.py b/cuda_pathfinder/tests/test_load_nvidia_dynamic_lib.py index a6a69e5c6ac..4cc524c49e8 100644 --- a/cuda_pathfinder/tests/test_load_nvidia_dynamic_lib.py +++ b/cuda_pathfinder/tests/test_load_nvidia_dynamic_lib.py @@ -157,6 +157,8 @@ def test_cutensor_mg_expected_failure_follows_installed_cutensor_version( have_distribution.cache_clear() +# For libname=nvcudla: +# This test is not currently run in any CI job @pytest.mark.parametrize( "libname", supported_nvidia_libs.SUPPORTED_WINDOWS_DLLS if IS_WINDOWS else supported_nvidia_libs.SUPPORTED_LINUX_SONAMES,