mblt-tracker 0.2.4

A Python library to track device usage, such as Mobilint NPU, NVIDIA GPU, Intel CPU, ...

Mobilint Device Tracker

A lightweight Python library and CLI for tracking dynamic hardware metrics and collecting static system metadata across CPU, DRAM, GPU, and NPU.

Overview

mblt-tracker is designed to help developers and researchers measure hardware performance with fair and consistent criteria. It provides a unified interface to poll metrics in the background while your code runs, producing both summarized statistics and detailed time-series traces.

✨ Key Features

• Multi-Backend Support: Unified interface for Intel CPU/DRAM, NVIDIA GPU, and Mobilint NPU.
• Background Tracking: Uses a background scheduler to poll metrics without blocking your main execution.
• Comprehensive Metrics: Capture Power (Watts), Utilization (%), Memory Usage (MB/%), and Temperature (C).
• Statistical Summaries: Automatically calculates averages, peaks (max), and p99 values.
• Time-Series Traces: Export raw data for custom plotting and analysis.
• Static Metadata: Collect best-effort host, OS, CPU, DRAM capacity/type/speed, motherboard, PCIe, driver, firmware, and device information for reproducible benchmarks.
• CLI Collection Tool: Use mblt-tracker collect to export static host and PCIe information as JSON.
• PCIe Discovery: Detect GPU/NPU-related PCIe devices on Linux and Windows, including current and maximum link speed/width where available.
• Lightweight: Minimal overhead, designed for production and research environments.

---

🚀 Installation

pip install mblt-tracker

For the latest features, install directly from source:

git clone https://github.com/mobilint/mblt-tracker.git
cd mblt-tracker
pip install -e .

---

📖 Quick Start Guide

The typical workflow involves initializing a tracker, starting it before your target workload, and stopping it after.

from mblt_tracker import CPUDeviceTracker # or DRAMDeviceTracker, GPUDeviceTracker, NPUDeviceTracker

# 1. Initialize with a polling interval (seconds)
tracker = CPUDeviceTracker(interval=0.1)

# 2. Start tracking (best to run after warm-up)
tracker.start()

# --- Your workload starts here ---
# e.g., model.inference(data)
# --- Your workload ends here ---

# 3. Stop tracking
tracker.stop()

# 4. Access results
metrics = tracker.get_metric()

def format_metric(value, unit):
    return f"{value:.2f} {unit}" if value is not None else f"N/A {unit}"

print(f"Average Power: {format_metric(metrics['avg_power_w'], 'W')}")
print(f"Max Utilization: {format_metric(metrics['max_utilization_pct'], '%')}")
print(f"Max Temperature: {format_metric(metrics['max_temperature_c'], 'C')}")

# 5. Export time-series traces
power_trace = tracker.get_trace()      # list of (timestamp, power_w)
util_trace = tracker.get_util_trace()  # list of (timestamp, utilization_pct)
temp_trace = tracker.get_temp_trace()  # list of (timestamp, temperature_c)

# 6. Collect static metadata for reproducibility
static_info = tracker.get_static_info()

---

Command Line Interface

mblt-tracker provides a CLI for collecting static host and PCIe metadata without writing Python code.

# Print collected information to stdout
mblt-tracker collect

# Save collected information as JSON
mblt-tracker collect -o static-info.json

# Include all PCIe devices instead of only GPU/NPU-related devices
mblt-tracker collect --all-pcie-devices

# Filter NPU PCIe discovery by vendor/device/class
mblt-tracker collect --pcie-vendor-id 0x1ed5
mblt-tracker collect --pcie-vendor-id 1ed5 --pcie-device-id 0100
mblt-tracker collect --pcie-class-filter 0x12

The CLI output is a JSON document containing best-effort host CPU, DRAM, motherboard, OS, GPU, NPU, driver, and PCIe information. For NVIDIA GPU entries, NVML is the source of truth for GPU identity and PCIe link metadata; OS PCIe discovery is used only to attach non-link PCIe identifiers and descriptive fields where available. On Linux, non-NVIDIA PCIe information is read from sysfs. On Windows, non-NVIDIA PCI devices are collected through PowerShell/CIM/PnP queries.

Example Output

The following examples show representative mblt-tracker collect outputs across Windows and Linux systems. Public static output intentionally omits privacy-sensitive host and device instance identifiers: DRAM DIMM part/serial numbers, motherboard serial/asset tags, and PCIe bus_address / Windows pnp_device_id are not exposed, including when --all-pcie-devices is used. DRAM byte counts are accompanied by MB/GB display units; DRAM speed, type, module capacity, and estimated theoretical bandwidth are kept to make benchmark results easier to interpret.

Windows host with Intel UHD Graphics, NVIDIA RTX 3090, and Mobilint NPU

$ mblt-tracker collect
WARNING:root:imports error 
 You need to install pymongo>=3.9.0 in order to use MongoOutput 
WARNING:root:imports error 
  You need to install pandas>=0.25.1 in order to use DataFrameOutput 
{
  "hardware": {
    "cpu": {
      "architecture": "AMD64",
      "model_name": "13th Gen Intel(R) Core(TM) i5-13500",
      "vendor": "GenuineIntel"
    },
    "dram": {
      "available_bytes": 15580520448,
      "available_gb": 14.51,
      "available_mb": 14858.74,
      "configured_speed_mhz": 5600,
      "module_count": 2,
      "modules": [
        {
          "capacity_bytes": 17179869184,
          "capacity_gb": 16.0,
          "capacity_mb": 16384.0,
          "configured_speed_mhz": 5600,
          "data_width_bits": 64,
          "ram_type": "DDR5",
          "speed_mhz": 5600,
          "theoretical_bandwidth_gbps": 44.8,
          "total_width_bits": 64
        },
        {
          "capacity_bytes": 17179869184,
          "capacity_gb": 16.0,
          "capacity_mb": 16384.0,
          "configured_speed_mhz": 5600,
          "data_width_bits": 64,
          "ram_type": "DDR5",
          "speed_mhz": 5600,
          "theoretical_bandwidth_gbps": 44.8,
          "total_width_bits": 64
        }
      ],
      "ram_type": "DDR5",
      "speed_mhz": 5600,
      "theoretical_bandwidth_gbps": 89.6,
      "total_bytes": 34113015808,
      "total_gb": 31.77,
      "total_mb": 32532.71
    },
    "gpus": [
      {
        "class": "0x038000",
        "dev_no": 0,
        "device_id": "0x4680",
        "driver_date": "/Date(1764547200000)/",
        "driver_description": "Intel(R) UHD Graphics 770",
        "driver_provider": "Intel Corporation",
        "driver_version": "32.0.101.7082",
        "manufacturer": "Intel Corporation",
        "name": "Intel(R) UHD Graphics 770",
        "revision": "0x0c",
        "status": "OK",
        "subsystem_device_id": "0x7d96",
        "subsystem_vendor_id": "0x1462",
        "vendor_id": "0x8086"
      },
      {
        "architecture": "Ampere",
        "class": "0x030000",
        "dev_no": 0,
        "device_id": "0x2204",
        "driver_date": "/Date(1773705600000)/",
        "driver_description": "NVIDIA GeForce RTX 3090",
        "driver_provider": "NVIDIA",
        "driver_version": "595.97",
        "lane_width": "x4",
        "link_generation": "Gen2",
        "manufacturer": "NVIDIA",
        "max_lane_width": "x16",
        "max_link_generation": "Gen4",
        "memory_total_bytes": 25769803776,
        "name": "NVIDIA GeForce RTX 3090",
        "revision": "0xa1",
        "status": "OK",
        "subsystem_device_id": "0x1454",
        "subsystem_vendor_id": "0x10de",
        "vendor_id": "0x10de"
      }
    ],
    "motherboard": {
      "chipset": "INTEL Intel(R) SMBus - 7A23",
      "manufacturer": "Micro-Star International Co., Ltd.",
      "model_name": "MAG B760 TOMAHAWK WIFI (MS-7D96)",
      "pcie": {
        "max_lane_width": "x16",
        "max_link_generation": "Gen4",
        "max_link_speed": "16.0 GT/s PCIe"
      },
      "version": "2.0"
    },
    "npus": [
      {
        "class": "0x078000",
        "current_link_speed": "16.0 GT/s PCIe",
        "current_link_width": "8",
        "dev_no": 0,
        "device_id": "0x0000",
        "driver_date": "/Date(1774828800000)/",
        "driver_description": "MOBILINT NPU Accelerator",
        "driver_provider": "MOBILINT, Inc.",
        "lane_width": "x8",
        "link_generation": "Gen4",
        "manufacturer": "MOBILINT, Inc.",
        "max_lane_width": "x8",
        "max_link_generation": "Gen4",
        "max_link_speed": "16.0 GT/s PCIe",
        "max_link_width": "8",
        "name": "MOBILINT NPU Accelerator",
        "revision": "0x02",
        "status": "OK",
        "subsystem_device_id": "0x1093",
        "subsystem_vendor_id": "0x0402",
        "vendor_id": "0x209f"
      }
    ]
  },
  "inference": {
    "cpu": {
      "governor": null,
      "max_processor_state_pct": 100,
      "min_processor_state_pct": 100,
      "power_plan": "High performance"
    },
    "cuda": {
      "version": "not_found"
    },
    "gpu": {
      "cuda_driver": {
        "version": "13.2"
      },
      "driver": {
        "version": "595.97"
      }
    },
    "npu_driver_version": "1.8.1.1348",
    "os": {
      "kernel_version": "11",
      "name": "Windows",
      "version": "10.0.26200"
    },
    "qbcompiler": {
      "version": "not_installed"
    },
    "qbruntime": {
      "version": "v1.2.0"
    }
  }
}

Linux host with NVIDIA RTX PRO 6000 Blackwell GPUs and MLA100

$ mblt-tracker collect
WARNING:root:imports error
 You need to install pymongo>=3.9.0 in order to use MongoOutput
WARNING:root:imports error
  You need to install pandas>=0.25.1 in order to use DataFrameOutput
{
  "hardware": {
    "cpu": {
      "architecture": "x86_64",
      "model_name": "INTEL(R) XEON(R) GOLD 6542Y",
      "vendor": "GenuineIntel"
    },
    "dram": {
      "available_bytes": 322689507328,
      "available_gb": 300.53,
      "available_mb": 307740.7,
      "total_bytes": 405389791232,
      "total_gb": 377.55,
      "total_mb": 386609.83
    },
    "gpus": [
      {
        "class": "0x030000",
        "dev_no": 0,
        "device_id": "0x2000",
        "manufacturer": "ASPEED Technology, Inc.",
        "name": "ASPEED Graphics Family",
        "revision": "0x52",
        "subsystem_device_id": "0x1c6b",
        "subsystem_vendor_id": "0x15d9",
        "vendor_id": "0x1a03"
      },
      {
        "architecture": "Blackwell",
        "class": "0x030000",
        "dev_no": 0,
        "device_id": "0x2bb1",
        "driver_version": "580.95.05",
        "lane_width": "x16",
        "link_generation": "Gen1",
        "manufacturer": "NVIDIA Corporation",
        "max_lane_width": "x16",
        "max_link_generation": "Gen5",
        "memory_total_bytes": 102641958912,
        "name": "NVIDIA RTX PRO 6000 Blackwell Workstation Edition",
        "revision": "0xa1",
        "subsystem_device_id": "0x204b",
        "subsystem_vendor_id": "0x10de",
        "vendor_id": "0x10de"
      },
      {
        "architecture": "Blackwell",
        "class": "0x030000",
        "dev_no": 1,
        "device_id": "0x2bb1",
        "driver_version": "580.95.05",
        "lane_width": "x16",
        "link_generation": "Gen1",
        "manufacturer": "NVIDIA Corporation",
        "max_lane_width": "x16",
        "max_link_generation": "Gen5",
        "memory_total_bytes": 102641958912,
        "name": "NVIDIA RTX PRO 6000 Blackwell Workstation Edition",
        "revision": "0xa1",
        "subsystem_device_id": "0x204b",
        "subsystem_vendor_id": "0x10de",
        "vendor_id": "0x10de"
      }
    ],
    "motherboard": {
      "chipset": "ASPEED Technology, Inc. AST1150 PCI-to-PCI Bridge",
      "manufacturer": "Supermicro",
      "model_name": "X13DEG-QT",
      "pcie": {
        "max_lane_width": "x16",
        "max_link_generation": "Gen5",
        "max_link_speed": "32.0 GT/s PCIe"
      },
      "version": "1.10"
    },
    "npus": [
      {
        "board_name": "aries0",
        "card_id": 0,
        "card_model": "MLA100",
        "class": "0x7800002",
        "current_link_speed": "16.0 GT/s PCIe",
        "current_link_width": "8",
        "dev_no": 0,
        "device_id": "0x0",
        "firmware": {
          "revision": "0",
          "version": "1.1"
        },
        "lane_width": "8",
        "link_generation": "4",
        "manufacturer": "Mobilint, Inc.",
        "max_lane_width": "x8",
        "max_link_generation": "Gen4",
        "max_link_speed": "16.0 GT/s PCIe",
        "max_link_width": "8",
        "memory_total_bytes": 17179869184,
        "name": "Aries",
        "product": "Aries",
        "revision": "0x2",
        "subsystem_device_id": "0x1093",
        "subsystem_vendor_id": "0x401",
        "vendor_id": "0x209F"
      }
    ]
  },
  "inference": {
    "cpu": {
      "governor": "schedutil",
      "max_processor_state_pct": null,
      "min_processor_state_pct": null,
      "power_plan": null
    },
    "cuda": {
      "version": "12.8"
    },
    "driver": {
      "aries_version": "1.12.0",
      "regulus_version": "N/A"
    },
    "gpu": {
      "cuda_driver": {
        "version": "13.0"
      },
      "driver": {
        "version": "580.95.05"
      }
    },
    "npu_driver_version": "1.12.0",
    "os": {
      "kernel_version": "6.8.0-110-generic",
      "name": "Linux",
      "version": "Ubuntu 24.04 LTS"
    },
    "qbcompiler": {
      "version": "not_installed"
    },
    "qbruntime": {
      "version": "v1.2.0"
    }
  }
}

Linux host with MLA400 NPUs and NVML unavailable

$ mblt-tracker collect
WARNING:root:imports error
 You need to install pymongo>=3.9.0 in order to use MongoOutput
WARNING:root:imports error
  You need to install pandas>=0.25.1 in order to use DataFrameOutput
Warning: NVML not available. GPU information will not be collected.
{
  "hardware": {
    "cpu": {
      "architecture": "x86_64",
      "model_name": "13th Gen Intel(R) Core(TM) i5-13600K",
      "vendor": "GenuineIntel"
    },
    "dram": {
      "available_bytes": 27281420288,
      "available_gb": 25.41,
      "available_mb": 26017.59,
      "total_bytes": 33379606528,
      "total_gb": 31.09,
      "total_mb": 31833.27
    },
    "motherboard": {
      "chipset": "Intel Corporation Raptor Lake PCI Express Root Port #25",
      "manufacturer": "ASUSTeK COMPUTER INC.",
      "model_name": "ROG STRIX B760-I GAMING WIFI",
      "pcie": {
        "max_lane_width": "x16",
        "max_link_generation": "Gen5",
        "max_link_speed": "32.0 GT/s PCIe"
      },
      "version": "Rev 1.xx"
    },
    "npus": [
      {
        "board_name": "aries0",
        "card_id": 0,
        "card_model": "MLA400",
        "class": "0x7800002",
        "current_link_speed": "16.0 GT/s PCIe",
        "current_link_width": "8",
        "dev_no": 0,
        "device_id": "0x0",
        "firmware": {
          "revision": "0",
          "version": "1.2.5"
        },
        "lane_width": "8",
        "link_generation": "4",
        "manufacturer": "MOBILINT, Inc.",
        "max_lane_width": "x8",
        "max_link_generation": "Gen4",
        "max_link_speed": "16.0 GT/s PCIe",
        "max_link_width": "8",
        "memory_total_bytes": 17179869184,
        "name": "MOBILINT NPU Accelerator",
        "product": "Aries",
        "revision": "0x2",
        "subsystem_device_id": "0x108B",
        "subsystem_vendor_id": "0x402",
        "vendor_id": "0x209F"
      },
      {
        "board_name": "aries1",
        "card_id": 0,
        "card_model": "MLA400",
        "class": "0x7800002",
        "current_link_speed": "16.0 GT/s PCIe",
        "current_link_width": "8",
        "dev_no": 1,
        "device_id": "0x0",
        "firmware": {
          "revision": "0",
          "version": "1.2.5"
        },
        "lane_width": "8",
        "link_generation": "4",
        "manufacturer": "MOBILINT, Inc.",
        "max_lane_width": "x8",
        "max_link_generation": "Gen4",
        "max_link_speed": "16.0 GT/s PCIe",
        "max_link_width": "8",
        "memory_total_bytes": 17179869184,
        "name": "MOBILINT NPU Accelerator",
        "product": "Aries",
        "revision": "0x2",
        "subsystem_device_id": "0x108B",
        "subsystem_vendor_id": "0x402",
        "vendor_id": "0x209F"
      },
      {
        "board_name": "aries2",
        "card_id": 0,
        "card_model": "MLA400",
        "class": "0x7800002",
        "current_link_speed": "16.0 GT/s PCIe",
        "current_link_width": "8",
        "dev_no": 2,
        "device_id": "0x0",
        "firmware": {
          "revision": "0",
          "version": "1.2.5"
        },
        "lane_width": "8",
        "link_generation": "4",
        "manufacturer": "MOBILINT, Inc.",
        "max_lane_width": "x8",
        "max_link_generation": "Gen4",
        "max_link_speed": "16.0 GT/s PCIe",
        "max_link_width": "8",
        "memory_total_bytes": 17179869184,
        "name": "MOBILINT NPU Accelerator",
        "product": "Aries",
        "revision": "0x2",
        "subsystem_device_id": "0x108B",
        "subsystem_vendor_id": "0x402",
        "vendor_id": "0x209F"
      },
      {
        "board_name": "aries3",
        "card_id": 0,
        "card_model": "MLA400",
        "class": "0x7800002",
        "current_link_speed": "16.0 GT/s PCIe",
        "current_link_width": "8",
        "dev_no": 3,
        "device_id": "0x0",
        "firmware": {
          "revision": "0",
          "version": "1.2.5"
        },
        "lane_width": "8",
        "link_generation": "4",
        "manufacturer": "MOBILINT, Inc.",
        "max_lane_width": "x8",
        "max_link_generation": "Gen4",
        "max_link_speed": "16.0 GT/s PCIe",
        "max_link_width": "8",
        "memory_total_bytes": 17179869184,
        "name": "MOBILINT NPU Accelerator",
        "product": "Aries",
        "revision": "0x2",
        "subsystem_device_id": "0x108B",
        "subsystem_vendor_id": "0x402",
        "vendor_id": "0x209F"
      }
    ]
  },
  "inference": {
    "cpu": {
      "governor": "powersave",
      "max_processor_state_pct": null,
      "min_processor_state_pct": null,
      "power_plan": null
    },
    "cuda": {
      "version": "12.8"
    },
    "driver": {
      "aries_version": "1.12.0",
      "regulus_version": "N/A"
    },
    "npu_driver_version": "1.12.0",
    "os": {
      "kernel_version": "6.17.0-23-generic",
      "name": "Linux",
      "version": "Ubuntu 24.04.4 LTS"
    },
    "qbcompiler": {
      "version": "not_installed"
    },
    "qbruntime": {
      "version": "v1.2.0"
    }
  }
}

---

Metrics Coverage

Metric	Intel CPU	Host DRAM	NVIDIA GPU	Mobilint NPU
Power (W)	✅ (RAPL)	✅ (RAPL DRAM)	✅ (NVML)	✅ (mobilint-cli)
Utilization (%)	✅ (psutil)	N/A	✅ (NVML)	✅ (mobilint-cli)
Memory (MB/%)	✅ (psutil)	N/A	✅ (NVML)	✅ (mobilint-cli)
Temperature (C)	✅ (psutil)	N/A	✅ (NVML)	✅ (mobilint-cli)
Static Info	✅ Host/OS/DRAM/Motherboard	✅ Host/OS/DRAM/Motherboard	✅ NVML + PCIe	✅ PCIe + mobilint-cli
Per-Device Stats	✅ (Sockets)	✅ (Sockets)	✅ (GPU Indices)	✅ (Logical NPU Cards)

---

🛠️ Hardware Specifics

Intel CPU

Uses pyRAPL for power measurements and psutil for utilization/memory.

• Permission: Requires read access to Intel RAPL sysfs.

  sudo chmod -R a+r /sys/class/powercap/intel-rapl/
  

• Docker: Run containers with --privileged or mount the powercap directory.

• Features: Tracks total system CPU usage or specific indices (e.g., CPUDeviceTracker(cpu_id=[0, 1])).

• Temperature: Uses psutil.sensors_temperatures() when the platform exposes CPU thermal sensors.

• Static Info: Reports CPU architecture, model, vendor, DRAM capacity/type/speed, motherboard metadata, OS details, and CPU power policy when available.

NVIDIA GPU

Uses NVML (via nvidia-ml-py) for high-fidelity hardware monitoring.

• Features: Tracks total system GPU usage or specific indices (e.g., GPUDeviceTracker(gpu_id=[0, 1])).
• Dependencies: Requires NVIDIA Drivers and NVML library installed.
• Temperature: Reads on-die GPU temperature through NVML.
• Static Info: GPUDeviceTracker.get_static_info() reports the detected GPU count, tracked device names, NVIDIA driver version, and the raw NVML CUDA driver version. The mblt-tracker collect CLI provides richer NVML-discovered GPU metadata; for NVIDIA GPUs, PCIe link generation/width fields come from NVML, while OS PCIe discovery may add non-link identifiers and descriptive metadata.

Host DRAM

Uses the Intel RAPL DRAM domain through pyRAPL for host DRAM power measurements.

• Platform/Hardware: Requires a host that exposes RAPL DRAM energy counters.
• Permission: Requires read access to Intel RAPL sysfs, similar to CPU power tracking.
• Features: Tracks all detected CPU socket DRAM domains by default, or specific socket IDs with DRAMDeviceTracker(socket_id=0) / DRAMDeviceTracker(socket_id=[0, 1]).
• Metrics: Reports total host DRAM power through standard keys (avg_power_w, p99_power_w, max_power_w) and DRAM-specific aliases (avg_dram_power_w, p99_dram_power_w, max_dram_power_w). Per-socket statistics are returned under metrics["dram"].
• Trace: DRAMDeviceTracker.get_trace() returns total host DRAM power as list[(timestamp, power_w)].
• Static Info: DRAMDeviceTracker.get_static_info() returns the same privacy-first host CPU, aggregate DRAM capacity with MB/GB display units, optional DIMM capacity/type/speed summaries, motherboard metadata, and OS metadata as host static collection. Individual DIMM identifiers are not exposed.

Mobilint NPU

Polls the mobilint-cli status -q command, with a legacy JSON fallback for older environments.

• Platform: Currently supports Linux only.
• Requirement: Ensure Mobilint Utility Tool is installed and mobilint-cli is in your PATH.
• Device Selection: Tracks all logical NPU cards by default, or selected logical card IDs with NPUDeviceTracker(npu_id=0) / NPUDeviceTracker(npu_id=[0, 1]).
• MLA100 vs MLA400: status -q output is classified best-effort. Devices with a Power.GOLDFINGER rail are treated as MLA400 and grouped as one logical card. MLA100 devices remain one logical card per PCIe card. PCIe subsystem IDs are also used as a fallback (0x401/0x1093 for MLA100, 0x402/0x108B for MLA400 observed outputs).
• NPU Power: Distinguishes between NPU core power and total card/system power. For MLA400, Power.Total is reported by the first chip, while NPU core, memory, and utilization are aggregated across the grouped Aries chips.
• DDR/PMIC/GOLDFINGER Power: Parses optional NPU board DDR, PMIC, and MLA400 GOLDFINGER power rails when present.
• Temperature: Parses NPU temperature from mobilint-cli status output when available.
• Static Info: Reports Mobilint PCIe device information and parses driver, firmware, product, board, card_model, and card_id metadata from mobilint-cli status when available.

---

📝 Metric Output Format

Calling get_metric() returns a dictionary with standardized cross-device keys where applicable. Missing or unavailable measurements are returned as None.

{
  "avg_power_w": 25.4,
  "p99_power_w": 40.1,
  "max_power_w": 45.2,
  "avg_utilization_pct": 78.5,
  "p99_utilization_pct": 90.0,
  "max_utilization_pct": 95.0,
  "avg_memory_used_mb": 2048.0,
  "p99_memory_used_mb": 3072.0,
  "max_memory_used_mb": 4096.0,
  "total_memory_mb": 8192.0,
  "avg_memory_used_pct": 25.0,
  "p99_memory_used_pct": 37.5,
  "max_memory_used_pct": 50.0,
  "avg_temperature_c": 72.3,
  "p99_temperature_c": 79.0,
  "max_temperature_c": 80.0,
  "samples": 100,
  "util_samples": 101
}

Tracker-specific fields may also be present:

• CPU: cpu contains per-socket statistics keyed by socket ID.
• GPU: gpu contains per-GPU statistics keyed by GPU index. GPU-specific summary keys include avg_gpu_util_pct, p99_gpu_util_pct, max_gpu_util_pct, avg_mem_util_pct, and p99_mem_util_pct.
• DRAM: DRAM-specific power keys include avg_dram_power_w, p99_dram_power_w, and max_dram_power_w. dram contains per-socket statistics keyed by socket ID.
• NPU: NPU-specific power keys include avg_npu_power_w, p99_npu_power_w, max_npu_power_w, avg_ddr_power_w, p99_ddr_power_w, max_ddr_power_w, avg_pmic_power_w, p99_pmic_power_w, max_pmic_power_w, avg_goldfinger_power_w, p99_goldfinger_power_w, max_goldfinger_power_w, avg_total_power_w, p99_total_power_w, and max_total_power_w. avg_power_w is mapped to total power for cross-device consistency. npu contains per logical card statistics keyed by card ID.

Time-Series Trace APIs

All trackers expose trace APIs for post-processing and plotting:

tracker.get_trace()       # Power trace: list[(timestamp, power_w)]
tracker.get_util_trace()  # Utilization trace: list[(timestamp, utilization_pct)]
tracker.get_temp_trace()  # Temperature trace: list[(timestamp, temperature_c)]

NPU trackers additionally expose rail-specific power traces:

tracker.get_npu_power_trace()         # NPU core power
tracker.get_ddr_power_trace()         # On-board NPU DDR power, when available
tracker.get_pmic_power_trace()        # NPU PMIC power, when available
tracker.get_goldfinger_power_trace()  # MLA400 GOLDFINGER input power, when available

---

🔍 Static Information

For benchmark reproducibility, each tracker exposes get_static_info():

from mblt_tracker import CPUDeviceTracker, GPUDeviceTracker, NPUDeviceTracker

tracker = CPUDeviceTracker()
info = tracker.get_static_info()

Static information is collected on a best-effort, privacy-first basis and may vary by platform and permissions.

Typical fields include:

• hardware.cpu: CPU architecture, model name, and vendor
• hardware.dram: total and available memory in bytes plus total_mb, total_gb, available_mb, and available_gb; optional privacy-safe aggregate fields include ram_type, speed_mhz, configured_speed_mhz, module_count, modules, and theoretical_bandwidth_gbps when available. Per-module entries may include capacity_bytes, capacity_mb, capacity_gb, DDR type, speed, width, and estimated bandwidth. Individual DIMM serial numbers, part numbers, and PCIe/device instance identifiers are not exposed.
• hardware.motherboard: optional baseboard manufacturer, model_name, version, best-effort chipset, and pcie support/capability summary. Motherboard serial numbers, asset tags, PCI bus addresses, and Windows instance IDs are not exposed.
• hardware.motherboard.pcie: optional maximum PCIe generation/speed/lane-width summary.
• inference.os: OS name, version, and kernel version
• inference.cpu: OS-independent CPU power policy object. Linux fills governor; Windows fills power_plan, min_processor_state_pct, and max_processor_state_pct. Unavailable OS-specific attributes are kept as null.
• hardware.gpu: GPUDeviceTracker.get_static_info() output with device_count and a devices list containing tracked GPU indices and names
• hardware.gpus: mblt-tracker collect output containing NVML-discovered NVIDIA GPU devices. For NVIDIA GPUs, current and maximum PCIe link generation/width fields are sourced from NVML; OS PCIe discovery may add non-link fields such as vendor/device IDs and descriptive metadata where available. Private PCIe instance identifiers such as bus_address and pnp_device_id are omitted from public output.
• hardware.npus: Mobilint PCIe devices, including vendor/device IDs, current/maximum link information, and firmware metadata where available. Private PCIe instance identifiers such as bus_address and pnp_device_id are omitted from public output.
• hardware.npus[].card_model: best-effort Mobilint card model classification such as MLA100 or MLA400 when mobilint-cli status -q exposes enough information
• hardware.npus[].card_id: logical NPU card ID used by NPUDeviceTracker(npu_id=...); MLA400 Aries chips share the same card ID
• inference.gpu: NVIDIA driver and CUDA driver versions. The CLI normalizes the CUDA driver version as a string such as "13.0"; GPUDeviceTracker.get_static_info() returns the raw NVML CUDA driver integer.
• hardware.npus[].firmware: per-NPU firmware metadata where available. Linux currently maps mobilint-cli status firmware rows by device order.
• inference.npu_driver_version: host Mobilint NPU driver version when available. Linux may also include inference.driver with Aries/Regulus driver metadata parsed from mobilint-cli status.

PCIe discovery supports:

• Linux: /sys/bus/pci/devices
• Windows: PowerShell/CIM/PnP PCI device queries

For tests or custom environments, MBLT_TRACKER_PCI_SYSFS can override the Linux PCI sysfs root. NPUDeviceTracker.get_static_info() can customize NPU PCIe matching with MBLT_TRACKER_NPU_PCI_VENDOR_ID, MBLT_TRACKER_NPU_PCI_DEVICE_ID, and MBLT_TRACKER_NPU_PCI_CLASS_FILTER. The mblt-tracker collect CLI uses the corresponding --pcie-vendor-id, --pcie-device-id, and --pcie-class-filter flags.

---

🤝 Contributing

We welcome contributions! To set up for development:

1. Install dev dependencies: pip install -e ".[dev]"
2. Run tests: pytest tests/

📄 License

This project is licensed under the BSD-3-Clause License. See the LICENSE file for details.

---

Developed with ❤️ by Mobilint Inc.