Skip to content

test_nvidia_telemetry

ipw.tests.integration.test_nvidia_telemetry

Tier 3 integration tests for NVIDIA GPU telemetry.

These tests require actual NVIDIA GPU hardware and drivers. They are skipped unless the environment provides NVIDIA tooling.

test_nvidia_smi_available()

Verify nvidia-smi is accessible.

Source code in intelligence-per-watt/src/ipw/tests/integration/test_nvidia_telemetry.py 
23
24
25
def test_nvidia_smi_available() -> None:
    """Verify nvidia-smi is accessible."""
    assert shutil.which("nvidia-smi") is not None

test_nvml_telemetry_collection()

Placeholder: collect NVML telemetry readings from a real GPU.

Source code in intelligence-per-watt/src/ipw/tests/integration/test_nvidia_telemetry.py 
28
29
30
def test_nvml_telemetry_collection() -> None:
    """Placeholder: collect NVML telemetry readings from a real GPU."""
    pytest.skip("Full NVML integration test not yet implemented")

test_gpu_energy_counter_monotonic()

Placeholder: verify GPU energy counter is monotonically increasing.

Source code in intelligence-per-watt/src/ipw/tests/integration/test_nvidia_telemetry.py 
33
34
35
def test_gpu_energy_counter_monotonic() -> None:
    """Placeholder: verify GPU energy counter is monotonically increasing."""
    pytest.skip("Full NVML energy counter test not yet implemented")