CPU#

class impacthpc.src.CPU(embodied_impacts: Impacts | None = None, electric_power: ReplicableValue | str | None = None, peak_power: ReplicableValue | str | None = None, idle_power: ReplicableValue | str | None = None, name: Name | None = None, tdp: float | None = None, model_range: str | None = None, power_measures: List[PowerMeasure] = [], die_size: ReplicableValue | str | None = None, cpu_family: str | None = None, cores: int | None = None, release_date_range: DateRange | None = None)#

Bases: HasConsumptionAndEmbodiedImpacts

Subclass of Component representing a CPU.

This class uses the same estimation as BoavitzAPI, described in

Thibault Simon, David Ekchajzer, Adrien Berthelot, Eric Fourboul, Samuel Rince, et al.. BoaviztAPI: a bottom-up model to assess the environmental impacts of cloud services. HotCarbon’24 - 3rd Workshop on Sustainable Computer Systems, Jul 2024, Santa Cruz, United States. hal-04621947v3"

See the documentation of BoavitzAPI : https://doc.api.boavizta.org/Explanations/components/cpu/

Examples

from impacthpc import CPU, ExactName

impacts = CPU(name=ExactName("Intel Xeon Gold 6248")).estimate_embodied_impacts()

impacts_climate_change = impacts["gwp"]

print(impacts_climate_change.explain())

If you don’t know the exact name of you CPU, use find_close_cpu_model_name()

from impacthpc import CPU, ExactName, find_close_cpu_model_name

impacts = CPU(name=find_close_cpu_model_name("Intel 6248")).estimate_embodied_impacts()

impacts_climate_change = impacts["gwp"]

print(impacts_climate_change.explain())

Electric power of this cpu :

from impacthpc import CPU, ExactName

power = CPU(name=ExactName("Intel Xeon Gold 6248")).estimate_electric_power()

print(power.explain())
name#

The name of a CPU model present in the database. Can be an instance of ExactName or the return value of find_close_cpu_model_name() for fuzzy matching name. Defaults to None.

Type:

Name, optional

tdp#

The Thermal Design Power of the CPU, used to determine the CPU’s instantaneous power consumption. Defaults to None.

Type:

float | None, optional

model_range#

The model range of the CPU, for example EPYC, Athlon X4, Core i5, Xeon Gold… The supported model ranges are listed in the CSV file at the path defined in the config file under csv > cpu_profile.csv. If tdp or power_measures are provided, model_range will be ignored. Defaults to None.

Type:

str | None, optional

power_measures#

Measurements of the CPU’s power consumption depending on the workload. Defaults to [].

Type:

List[PowerMeasure], optional

die_size#

The surface area of the CPU die. Used to determine the embodied impacts of the CPU. Defaults to None.

Type:

ReplicableValue | None, optional

cpu_family#

The family of the CPU. Ignored if the name is provided and the die size is in the database for the CPU name. Defaults to None.

Type:

str | None, optional

cores#

The number of physical cores of the CPU. Defaults to None.

Type:

int | None, optional

release_date_range#

The date range in which the CPU was released. Used in computations when the exact name of the CPU isn’t known and estimations are done with average values. Defaults to the ten last years.

Type:

DateRange | None, optional

Methods Summary

Methods Documentation

_cpu_profile() → tuple[ReplicableValue, ReplicableValue, ReplicableValue, ReplicableValue]#

Estimate the profile of the CPU.

CPU energy consumption is estimated using the formula a * ln(b * (workload + c)) + d, where: - workload is a percentage indicating how much the CPU is being used. - a, b, c, and d are parameters estimated using power measurements and TDP.

The CPU profile is the set of these four parameters: a, b, c, and d. This method is used to estimate them.

If the TDP and/or power measures are known, we use scipy.optimize to find a, b, c, and d values that fit best. Otherwise, we use default values for the model_range.

Returns:

The CPU profile consisting of the four parameters a, b, c, and d.

Return type:

tuple[ReplicableValue, ReplicableValue, ReplicableValue, ReplicableValue]

_estimate_die_size() → ReplicableValue#

Estimate the Die size of a CPU.

../_images/estimate_die_size_schema.svg
Alt:

Schema that explains the logic used in this method

Returns:

The estimated Die size.

Return type:

ReplicableValue

_get_cpus_matching_by_name() → DataFrame#

Return a sub-dataframe of CPUs_specs containing only the row matching name.

Return an empty dataframe with the same columns as CPUs_specs if name is None.

Raises:

ValueError – name should be an instance of the abstract class Name or None. Therefore, it can be an ExactName, a FuzzymatchSingleResult, a FuzzymatchMultipleResult, or None.

Returns:

A sub-dataframe of CPUs_specs containing only the row matching name.

Return type:

pandas.DataFrame

_linear_regression_core_to_die_size(cpus: DataFrame) → SourcedValue#

Perform a linear regression on the die sizes and numbers of cores of different CPUs to estimate the die size of the current CPU.

Parameters:

cpus (pandas.DataFrame) – A dataframe with at least two columns, “cores” and “total_die_size”, which describes CPUs and is used to perform the linear regression.

Raises:

ValueError – all pairs of cores and total_die_size in cpus are (nan, nan).

Returns:

The estimated die size of the CPU.

Return type:

SourcedValue

estimate_electric_power(workload: ReplicableValue | None = None) → ReplicableValue#

Estimate the electric power of the CPU.

CPU electric power is estimated using the formula a * ln(b * (workload + c)) + d, where:

  • workload is a percentage indicating how much the CPU is being used.

  • a, b, c, and d are parameters estimated using power measurements and TDP.

These four parameters a, b, c, and d are called the CPU profile and are estimated by the method _cpu_profile().

Parameters:

workload (ReplicableValue, optional) – The :arg:`workload` of the CPU, between 0 and 100, as a ReplicableValue. The unit should be dimensionless. Defaults to the value defined in the config file under default_values_cpu > workload.

Returns:

The estimated instant consumption of the CPU.

Return type:

ReplicableValue

estimate_embodied_impacts() → Impacts#

Estimate the embodied impacts of the CPU.

CPU embodied impacts are estimated using: self._die_size_factor * self.die_size + self._base_impact, where:

  • _die_size_factor is the impact of 1 mm² of die. It is defined in the config file under default_values_cpu > die_size_factor.

  • die_size is the surface of the die in mm², either provided by the user or estimated in _estimate_die_size().

  • base_impact is the impact of everything that isn’t the die in a CPU, and is defined in the config file under default_values_cpu > base_impact.

Returns:

The embodied impacts of the CPU.

Return type:

Impact

estimate_idle_power()#

Estimate the static electric power of the CPU when the workload is 0%.

The CPU’s static instant consumption is estimated using the estimate_electric_power() method with a workload of 0%.

Returns:

The electric power of the CPU.

Return type:

Impact

estimate_peak_power()#

Estimate the peak (maximum) electric power of the CPU when the workload is 100%.

The CPU’s peak instant consumption is estimated using the estimate_electric_power() method with a workload of 100%.

Returns:

The electric power of the CPU.

Return type:

Impact