codecarbon 3.2.3__tar.gz → 3.2.5__tar.gz

{codecarbon-3.2.3 → codecarbon-3.2.5}/PKG-INFO

@@ -1,23 +1,21 @@
 Metadata-Version: 2.4
 Name: codecarbon
-Version: 3.2.3
+Version: 3.2.5
 Author: Mila, DataForGood, BCG GAMMA, Comet.ml, Haverford College
 License-Expression: MIT
 Project-URL: Homepage, https://codecarbon.io/
 Project-URL: Repository, https://github.com/mlco2/codecarbon
 Project-URL: Dashboard, http://dashboard.codecarbon.io/
-Project-URL: Documentation, https://mlco2.github.io/codecarbon/
+Project-URL: Documentation, https://docs.codecarbon.io/
 Project-URL: Issues, https://github.com/mlco2/codecarbon/issues
 Project-URL: Changelog, https://github.com/mlco2/codecarbon/releases
 Classifier: Natural Language :: English
-Classifier: Programming Language :: Python :: 3.8
-Classifier: Programming Language :: Python :: 3.9
 Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
 Classifier: Programming Language :: Python :: 3.12
 Classifier: Programming Language :: Python :: 3.13
 Classifier: Programming Language :: Python :: 3.14
-Requires-Python: >=3.8
+Requires-Python: >=3.10
 Description-Content-Type: text/markdown
 License-File: LICENSE
 Requires-Dist: arrow
@@ -35,6 +33,7 @@ Requires-Dist: requests
 Requires-Dist: questionary
 Requires-Dist: rich
 Requires-Dist: typer
+Requires-Dist: pycountry
 Provides-Extra: carbonboard
 Requires-Dist: dash; extra == "carbonboard"
 Requires-Dist: dash_bootstrap_components>1.0.0; extra == "carbonboard"
@@ -43,34 +42,11 @@ Provides-Extra: viz-legacy
 Requires-Dist: dash; extra == "viz-legacy"
 Requires-Dist: dash_bootstrap_components>1.0.0; extra == "viz-legacy"
 Requires-Dist: fire; extra == "viz-legacy"
-Provides-Extra: api
-Requires-Dist: alembic<2.0.0; extra == "api"
-Requires-Dist: authlib>=1.2.1; extra == "api"
-Requires-Dist: bcrypt<5.0.0; extra == "api"
-Requires-Dist: python-dateutil<3.0.0; extra == "api"
-Requires-Dist: dependency-injector<5.0.0; extra == "api"
-Requires-Dist: fastapi<1.0.0; extra == "api"
-Requires-Dist: fief-client[fastapi]; extra == "api"
-Requires-Dist: httpx; extra == "api"
-Requires-Dist: itsdangerous; extra == "api"
-Requires-Dist: pydantic[email]<2.0.0; extra == "api"
-Requires-Dist: psycopg2-binary<3.0.0; extra == "api"
-Requires-Dist: requests<3.0.0; extra == "api"
-Requires-Dist: sqlalchemy<2.0.0; extra == "api"
-Requires-Dist: uvicorn[standard]<1.0.0; extra == "api"
-Requires-Dist: fastapi-pagination<1.0.0; extra == "api"
-Requires-Dist: pytest; extra == "api"
-Requires-Dist: mock; extra == "api"
-Requires-Dist: responses; extra == "api"
-Requires-Dist: fastapi-oidc; python_version >= "3.10" and extra == "api"
-Requires-Dist: numpy; extra == "api"
-Requires-Dist: psutil; extra == "api"
-Requires-Dist: requests-mock; extra == "api"
-Requires-Dist: rapidfuzz; extra == "api"
-Requires-Dist: PyJWT; extra == "api"
+Provides-Extra: amdsmi
+Requires-Dist: amdsmi>=6.0.0; extra == "amdsmi"
 Dynamic: license-file
 
-![banner](docs/edit/images/banner.png)
+![banner](docs/images/banner.png)
 
 Estimate and track carbon emissions from your computer, quantify and analyze their impact.
 
@@ -94,14 +70,18 @@ CodeCarbon websites:
 
 - [About CodeCarbon 💡](#about-codecarbon-)
 - [Quickstart 🚀](#quickstart-)
-    - [Installation 🔧](#installation-)
-    - [Start to estimate your impact 📏](#start-to-estimate-your-impact-)
-      - [Monitoring your whole machine](#monitoring-your-machine-)
-      - [In your python code](#in-your-python-code-)
-      - [Visualize](#visualize-)
+  - [Installation 🔧](#installation-)
+  - [Start to estimate your impact 📏](#start-to-estimate-your-impact-)
+    - [Without using the online dashboard](#without-using-the-online-dashboard)
+    - [With the online dashboard](#with-the-online-dashboard)
+    - [Monitoring your machine 💻](#monitoring-your-machine-)
+    - [Detecting your hardware 🔍](#detecting-your-hardware-)
+    - [In your Python code 🐍](#in-your-python-code-)
+  - [Visualize 📊](#visualize-)
 - [Contributing 🤝](#contributing-)
 - [How To Cite 📝](#how-to-cite-)
 - [Contact 📝](#contact-)
+  - [Star History](#star-history)
 
 # About CodeCarbon 💡
 
@@ -117,7 +97,7 @@ At **CodeCarbon**, we believe, along with Niels Bohr, that "Nothing exists until
 
 We created a Python package that estimates your hardware electricity power consumption (GPU + CPU + RAM) and we apply to it the carbon intensity of the region where the computing is done.
 
-![calculation Summary](docs/edit/images/calculation.png)
+![calculation Summary](docs/images/calculation.png)
 
 We explain more about this calculation in the [**Methodology**](https://mlco2.github.io/codecarbon/methodology.html#) section of the documentation.
 
@@ -200,7 +180,7 @@ There is other ways to use **codecarbon** package, please refer to the documenta
 ## Visualize 📊
 
 You can now visualize your experiment emissions on the [dashboard](https://dashboard.codecarbon.io/).
-![dashboard](docs/edit/images/dashboard.png)
+![dashboard](docs/images/dashboard.png)
 
 
 > Hope you enjoy your first steps monitoring your carbon computing impact!

{codecarbon-3.2.3 → codecarbon-3.2.5}/README.md

@@ -1,4 +1,4 @@
-![banner](docs/edit/images/banner.png)
+![banner](docs/images/banner.png)
 
 Estimate and track carbon emissions from your computer, quantify and analyze their impact.
 
@@ -22,14 +22,18 @@ CodeCarbon websites:
 
 - [About CodeCarbon 💡](#about-codecarbon-)
 - [Quickstart 🚀](#quickstart-)
-    - [Installation 🔧](#installation-)
-    - [Start to estimate your impact 📏](#start-to-estimate-your-impact-)
-      - [Monitoring your whole machine](#monitoring-your-machine-)
-      - [In your python code](#in-your-python-code-)
-      - [Visualize](#visualize-)
+  - [Installation 🔧](#installation-)
+  - [Start to estimate your impact 📏](#start-to-estimate-your-impact-)
+    - [Without using the online dashboard](#without-using-the-online-dashboard)
+    - [With the online dashboard](#with-the-online-dashboard)
+    - [Monitoring your machine 💻](#monitoring-your-machine-)
+    - [Detecting your hardware 🔍](#detecting-your-hardware-)
+    - [In your Python code 🐍](#in-your-python-code-)
+  - [Visualize 📊](#visualize-)
 - [Contributing 🤝](#contributing-)
 - [How To Cite 📝](#how-to-cite-)
 - [Contact 📝](#contact-)
+  - [Star History](#star-history)
 
 # About CodeCarbon 💡
 
@@ -45,7 +49,7 @@ At **CodeCarbon**, we believe, along with Niels Bohr, that "Nothing exists until
 
 We created a Python package that estimates your hardware electricity power consumption (GPU + CPU + RAM) and we apply to it the carbon intensity of the region where the computing is done.
 
-![calculation Summary](docs/edit/images/calculation.png)
+![calculation Summary](docs/images/calculation.png)
 
 We explain more about this calculation in the [**Methodology**](https://mlco2.github.io/codecarbon/methodology.html#) section of the documentation.
 
@@ -128,7 +132,7 @@ There is other ways to use **codecarbon** package, please refer to the documenta
 ## Visualize 📊
 
 You can now visualize your experiment emissions on the [dashboard](https://dashboard.codecarbon.io/).
-![dashboard](docs/edit/images/dashboard.png)
+![dashboard](docs/images/dashboard.png)
 
 
 > Hope you enjoy your first steps monitoring your carbon computing impact!

codecarbon-3.2.5/codecarbon/_version.py

@@ -0,0 +1 @@
+__version__ = "3.2.5"

{codecarbon-3.2.3 → codecarbon-3.2.5}/codecarbon/cli/main.py

@@ -69,13 +69,13 @@ def version(
 
 def show_config(path: Path = Path("./.codecarbon.config")) -> None:
     d = get_config(path)
-    api_endpoint = get_api_endpoint(path)
-    api = ApiClient(endpoint_url=api_endpoint)
-    api.set_access_token(get_access_token())
     print("Current configuration : \n")
     print("Config file content : ")
     print(d)
     try:
+        api_endpoint = get_api_endpoint(path)
+        api = ApiClient(endpoint_url=api_endpoint)
+        api.set_access_token(get_access_token())
         if "organization_id" not in d:
             print(
                 "No organization_id in config, follow setup instruction to complete your configuration file!",
@@ -102,8 +102,8 @@ def show_config(path: Path = Path("./.codecarbon.config")) -> None:
                     print("\nOrganization :")
                     print(org)
     except Exception as e:
-        raise ValueError(
-            f"Your configuration is invalid, please verify your configuration file at {path}. To start from scratch, run `codecarbon config` and overwrite your configuration file. (error: {e})"
+        print(
+            f"[yellow]Could not validate remote configuration details[/yellow]. You can continue with local configuration setup. (error: {e})"
         )
 
 
@@ -345,7 +345,8 @@ def monitor(
     if offline:
         if not country_iso_code:
             print(
-                "ERROR: country_iso_code is required for offline mode", file=sys.stderr
+                "ERROR: Country ISO code is required for offline mode. Add it to your configuration or provide it via the command line: `--country-iso-code FRA`",
+                file=sys.stderr,
             )
             raise typer.Exit(1)
 
@@ -358,7 +359,7 @@ def monitor(
         experiment_id = get_existing_local_exp_id()
         if api and experiment_id is None:
             print(
-                "ERROR: No experiment id, call 'codecarbon config' first.",
+                "ERROR: No experiment id, call 'codecarbon config' first. Or run in offline mode with `--offline --country-iso-code FRA` flag if you don't want to connect to the API.",
                 file=sys.stderr,
             )
             raise typer.Exit(1)

{codecarbon-3.2.3 → codecarbon-3.2.5}/codecarbon/core/config.py

@@ -1,7 +1,7 @@
 import configparser
 import os
 from pathlib import Path
-from typing import List, Union
+from typing import List, Optional, Union
 
 from codecarbon.external.logger import logger
 
@@ -73,6 +73,43 @@ def parse_gpu_ids(gpu_ids: Union[str, List[int]]) -> List[str]:
         )
 
 
+def normalize_gpu_ids(
+    gpu_ids: Optional[Union[str, List[Union[int, str]]]],
+) -> Optional[List[Union[int, str]]]:
+    """
+    Normalize GPU IDs from config/user input into a list of ids consumable by hardware
+    resolution code.
+
+    Supports:
+    - comma-separated string values (sanitized via parse_gpu_ids)
+    - lists containing ints and/or strings
+    """
+    if gpu_ids is None:
+        return None
+
+    if isinstance(gpu_ids, str):
+        return parse_gpu_ids(gpu_ids)
+
+    if isinstance(gpu_ids, list):
+        normalized_gpu_ids: List[Union[int, str]] = []
+        for gpu_id in gpu_ids:
+            if isinstance(gpu_id, int):
+                normalized_gpu_ids.append(gpu_id)
+            elif isinstance(gpu_id, str):
+                normalized_gpu_ids.extend(parse_gpu_ids(gpu_id))
+            else:
+                logger.warning(
+                    "Ignoring invalid gpu_id entry of type %s; expected int or str.",
+                    type(gpu_id).__name__,
+                )
+        return normalized_gpu_ids
+
+    logger.warning(
+        "Invalid gpu_ids format. Expected a string or a list of ints/strings."
+    )
+    return None
+
+
 def get_hierarchical_config():
     """
     Get the user-defined codecarbon configuration ConfigParser dictionnary

{codecarbon-3.2.3 → codecarbon-3.2.5}/codecarbon/core/cpu.py

@@ -17,7 +17,7 @@ from rapidfuzz import fuzz, process, utils
 
 from codecarbon.core.rapl import RAPLFile
 from codecarbon.core.units import Time
-from codecarbon.core.util import detect_cpu_model
+from codecarbon.core.util import count_cpus, detect_cpu_model
 from codecarbon.external.logger import logger
 from codecarbon.input import DataSource
 
@@ -1001,7 +1001,7 @@ class TDP:
             )
             if is_psutil_available():
                 # Count thread of the CPU
-                threads = psutil.cpu_count(logical=True)
+                threads = count_cpus()
                 estimated_tdp = threads * DEFAULT_POWER_PER_CORE
                 logger.warning(
                     f"We will use the default power consumption of {DEFAULT_POWER_PER_CORE} W per thread for your {threads} CPU, so {estimated_tdp}W."

{codecarbon-3.2.3 → codecarbon-3.2.5}/codecarbon/core/emissions.py

@@ -156,7 +156,7 @@ class Emissions:
                 )
 
         compute_with_regional_data: bool = (geo.region is not None) and (
-            geo.country_iso_code.upper() in ["USA", "CAN"]
+            geo.country_iso_code.upper() in ["USA", "CAN", "SWE", "NOR", "FIN"]
         )
 
         if compute_with_regional_data:
@@ -170,16 +170,72 @@ class Emissions:
                 )
         return self.get_country_emissions(energy, geo)
 
+    def _try_get_nordic_region_emissions(
+        self, energy: Energy, geo: GeoMetadata
+    ) -> Optional[float]:
+        nordic_regions = {
+            "SE1",
+            "SE2",
+            "SE3",
+            "SE4",
+            "NO1",
+            "NO2",
+            "NO3",
+            "NO4",
+            "NO5",
+            "FI",
+        }
+        if geo.region is None:
+            return None
+
+        region_upper = geo.region.upper()
+        if region_upper not in nordic_regions:
+            return None
+
+        try:
+            nordic_data = self._data_source.get_nordic_country_energy_mix_data()
+            region_data = nordic_data["data"].get(region_upper)
+            if region_data:
+                emission_factor_g = region_data["emission_factor"]
+                emission_factor_kg = emission_factor_g / 1000
+                emissions = emission_factor_kg * energy.kWh
+                logger.debug(
+                    f"Nordic region {geo.region}: Retrieved emissions using static factor "
+                    + f"{emission_factor_g} gCO2eq/kWh: {emissions * 1000} g CO2eq"
+                )
+                return emissions
+        except Exception as e:
+            logger.warning(
+                f"Error loading Nordic emissions data for {geo.region}: {e}. "
+                + "Falling back to default emission calculation."
+            )
+        return None
+
     def get_region_emissions(self, energy: Energy, geo: GeoMetadata) -> float:
         """
         Computes emissions for a region on private infra.
         Given an quantity of power consumed, use regional data
          on emissions per unit power consumed or the mix of energy sources.
         https://github.com/responsibleproblemsolving/energy-usage#calculating-co2-emissions
+
+        get_private_infra_emissions
+        ├─ Electricity Maps API (si token)
+        ├─ get_region_emissions (USA/CAN/SWE/NOR/FIN)
+        │   └─ _try_get_nordic_region_emissions (pour SWE/NOR/FIN)
+        │   └─ country_emissions_data (pour USA)
+        │   └─ country_energy_mix_data (pour CAN)
+        └─ get_country_emissions (fallback)
+
         :param energy: Mean power consumption of the process (kWh)
         :param geo: Country and region metadata.
         :return: CO2 emissions in kg
         """
+        # Handle Nordic regions (Sweden, Norway, Finland electricity bidding zones)
+        nordic_emissions = self._try_get_nordic_region_emissions(energy, geo)
+        if nordic_emissions is not None:
+            return nordic_emissions
+
+        # Handle USA and Canada regional data
         try:
             country_emissions_data = self._data_source.get_country_emissions_data(
                 geo.country_iso_code.lower()

codecarbon-3.2.5/codecarbon/core/gpu.py

@@ -0,0 +1,166 @@
+from typing import List
+
+from codecarbon.core import gpu_amd, gpu_nvidia
+from codecarbon.core.gpu_device import GPUDevice
+from codecarbon.core.units import Time
+from codecarbon.external.logger import logger
+
+AMDSMI_AVAILABLE = gpu_amd.AMDSMI_AVAILABLE
+PYNVML_AVAILABLE = gpu_nvidia.PYNVML_AVAILABLE
+
+AMDGPUDevice = gpu_amd.AMDGPUDevice
+NvidiaGPUDevice = gpu_nvidia.NvidiaGPUDevice
+is_rocm_system = gpu_amd.is_rocm_system
+is_nvidia_system = gpu_nvidia.is_nvidia_system
+
+# Backward-compatible module attributes
+amdsmi = gpu_amd.amdsmi
+pynvml = gpu_nvidia.pynvml
+
+
+class AllGPUDevices:
+    device_count: int
+    devices: List[GPUDevice]
+
+    def __init__(self) -> None:
+        gpu_details_available = is_gpu_details_available()
+        if gpu_details_available:
+            logger.debug("GPU available. Starting setup")
+        else:
+            logger.error("There is no GPU available")
+        self.devices = []
+
+        if PYNVML_AVAILABLE:
+            logger.debug("PyNVML available. Starting setup")
+            gpu_nvidia.pynvml.nvmlInit()
+            nvidia_devices_count = gpu_nvidia.pynvml.nvmlDeviceGetCount()
+            for i in range(nvidia_devices_count):
+                handle = gpu_nvidia.pynvml.nvmlDeviceGetHandleByIndex(i)
+                nvidia_gpu_device = NvidiaGPUDevice(handle=handle, gpu_index=i)
+                self.devices.append(nvidia_gpu_device)
+
+        if AMDSMI_AVAILABLE:
+            logger.debug("AMDSMI available. Starting setup")
+            try:
+                gpu_amd.amdsmi.amdsmi_init()
+                amd_devices_handles = gpu_amd.amdsmi.amdsmi_get_processor_handles()
+                if len(amd_devices_handles) == 0:
+                    logger.warning(
+                        "No AMD GPUs found on machine with amdsmi_get_processor_handles() !"
+                    )
+                else:
+                    for i, handle in enumerate(amd_devices_handles):
+                        # Try to get the actual device index from BDF (Bus/Device/Function)
+                        # If this fails, fall back to enumeration index
+                        try:
+                            bdf_info = gpu_amd.amdsmi.amdsmi_get_gpu_device_bdf(handle)
+                            # BDF typically contains domain, bus, device, function
+                            # The device portion often corresponds to the GPU index
+                            # For now, we'll use the enumeration index but log the BDF
+                            logger.debug(
+                                f"Found AMD GPU device with handle {handle}, enum_index {i}, BDF {bdf_info}: {gpu_amd.amdsmi.amdsmi_get_gpu_device_uuid(handle)}"
+                            )
+                            # Use enumerate index for now - this will be the index in the filtered list
+                            gpu_index = i
+                        except Exception:
+                            logger.debug(
+                                f"Found AMD GPU device with handle {handle} and index {i} : {gpu_amd.amdsmi.amdsmi_get_gpu_device_uuid(handle)}"
+                            )
+                            gpu_index = i
+
+                        amd_gpu_device = AMDGPUDevice(
+                            handle=handle, gpu_index=gpu_index
+                        )
+                        self.devices.append(amd_gpu_device)
+            except gpu_amd.amdsmi.AmdSmiException as e:
+                logger.warning(f"Failed to initialize AMDSMI: {e}", exc_info=True)
+        self.device_count = len(self.devices)
+
+    def start(self) -> None:
+        for device in self.devices:
+            if hasattr(device, "start"):
+                device.start()
+
+    def get_gpu_static_info(self) -> List:
+        """Get all GPUs static information.
+        >>> get_gpu_static_info()
+        [
+            {
+                "name": "Tesla V100-SXM2-16GB",
+                "uuid": "GPU-4e817856-1fb8-192a-7ab7-0e0e4476c184",
+                "total_memory": 16945512448,
+                "power_limit": 300000,
+                "gpu_index": 0,
+            }
+        ]
+        """
+        try:
+            devices_static_info = []
+            for i in range(self.device_count):
+                gpu_device = self.devices[i]
+                devices_static_info.append(gpu_device.get_static_details())
+            return devices_static_info
+
+        except Exception:
+            logger.warning("Failed to retrieve gpu static info", exc_info=True)
+            return []
+
+    def get_gpu_details(self) -> List:
+        """Get all GPUs instantaneous metrics
+        >>> get_gpu_details()
+        [
+            {
+                "name": "Tesla V100-SXM2-16GB",
+                "uuid": "GPU-4e817856-1fb8-192a-7ab7-0e0e4476c184",
+                "free_memory": 16945381376,
+                "total_memory": 16945512448,
+                "used_memory": 131072,
+                "temperature": 28,
+                "total_energy_consumption":2000,
+                "power_usage": 42159,
+                "power_limit": 300000,
+                "gpu_utilization": 0,
+                "compute_mode": 0,
+                "compute_processes": [],
+                "graphics_processes": [],
+            }
+        ]
+        """
+        try:
+            devices_info = []
+            for i in range(self.device_count):
+                gpu_device = self.devices[i]
+                devices_info.append(gpu_device.get_gpu_details())
+            return devices_info
+
+        except Exception:
+            logger.warning("Failed to retrieve gpu information", exc_info=True)
+            return []
+
+    def get_delta(self, last_duration: Time) -> List:
+        """Get difference since last time this function was called
+        >>> get_delta()
+        [
+            {
+                "name": "Tesla V100-SXM2-16GB",
+                "uuid": "GPU-4e817856-1fb8-192a-7ab7-0e0e4476c184",
+                "delta_energy_consumption":2000,
+                "power_usage": 42159,
+            }
+        ]
+        """
+        try:
+            devices_info = []
+            for i in range(self.device_count):
+                gpu_device = self.devices[i]
+                devices_info.append(gpu_device.delta(last_duration))
+            return devices_info
+
+        except Exception:
+            logger.warning("Failed to retrieve gpu information", exc_info=True)
+            return []
+
+
+def is_gpu_details_available() -> bool:
+    """Returns True if the GPU details are available."""
+    return PYNVML_AVAILABLE or AMDSMI_AVAILABLE