DE102017206808A1 - Method and device for assigning arithmetic tasks - Google Patents

Abstract

The invention describes a method for assigning secondary computing tasks to at least two locally distributed computing capacities (103a, 103b), in particular to computing capacities arranged in vehicles. At least one computing capacity (103a) is designed in addition to the calculation of the secondary computational tasks for the calculation of primary computational tasks and on this computational capacity, the calculation of the primary computational tasks is prioritized higher than the calculation of secondary computational tasks. The method comprises the following steps: obtaining and / or determining secondary computation tasks allocating the secondary computation tasks to locally distributed computational capacities (103a, 103b) as a function of computation-specific data, which includes information on utilization and / or availability of the computational capacities in the past ,

Description

State of the art

The invention relates to a method and a device for assigning arithmetic tasks.

The trend in automobiles goes to a central control unit with high computing power. Especially for fully autonomous vehicles a very high computing power is needed. As soon as the vehicle is parked, this computing power is currently not used.

Disclosure of the invention

Against this background, a method for the distribution of computing power, for the calculation of distributed computing tasks and corresponding devices claimed.

• - Erhalten und/oder Bestimmen von Sekundärrechenaufgaben;
• - Zuweisen der Sekundärrechenaufgaben an lokal verteilte Rechenkapazitäten in Abhängigkeit von rechenkapazitätsspezifischen Daten, welche Informationen über eine Auslastung und/oder Verfügbarkeit der Rechenkapazitäten in der Vergangenheit umfassen.

The invention describes a method for the assignment of secondary computing tasks to at least two locally distributed computing capacities, in particular computing capacities arranged in vehicles. At least one computing capacity is designed in addition to the calculation of the secondary computational tasks for the calculation of primary computational tasks and on this computational capacity, the calculation of the primary computational tasks is prioritized higher than the calculation of the secondary computational tasks. The method comprises the following steps:

• - obtaining and / or determining secondary computing tasks;
• Assigning the secondary computing tasks to locally distributed computing capacities as a function of computing capacity-specific data, which include information on the utilization and / or availability of the computing capacities in the past.

Locally distributed computing capacities are understood to be cache capacities that are located at different locations. A computing capacity can be understood as any type of processor or device having corresponding processors capable of performing arithmetic operations. For example, computers, computing servers and / or computing units are included in control units of vehicles. Computational tasks are understood to be tasks which can be solved by the described computing capacities. For example, they can be tasks for executing computer programs and / or algorithms which as a result generate data for databases and / or generate simulation data and / or associate objects by means of image recognition algorithms. In principle, any arithmetic operation which can be carried out with the aid of processors can be regarded as a task.

In the first method step, secondary calculation tasks can be obtained and / or determined. If, for example, the method runs on a task-distributing system which is designed only for the distribution of tasks and not for the generation of these tasks, only the secondary computing tasks are obtained from this system and then locally allocated computing capacities are allocated in the next method step. Furthermore, it is possible for this system to receive tasks and, based on these secondary computation tasks, determine which are subsequently assigned to locally distributed computing capacities. The assignment of secondary computing tasks can alternatively also be performed by a central processing unit which determines or creates the secondary computing tasks and does not receive any tasks from another server.

The assignment of the secondary computing tasks to locally distributed computing capacities takes place as a function of computing capacity-specific data, which includes information about a utilization and / or availability of the computing capacities in the past. Computing capacity-specific here means that the data include information about the individual behavior of a specific locally distributed computing capacity and thus can be determined individually for each computing capacity, whether this computing capacity should be assigned secondary computing tasks / can. The decision as to whether to assign a secondary computing task to a specific locally distributed computing resource may depend on information comprising several of the computing capacity specific data.

The computing capacity-specific data can be determined by the system which is responsible for the assignment of the secondary computing tasks, for example by recording the times required by the computing capacity for the calculation of secondary computing tasks to the respective computing capacity. Furthermore, it is possible for the system to obtain individual computing capacity specific data from the computing resources.

Information about utilization and / or availability of computing capacity in the past is understood to mean information about utilization and / or availability of computational capacity in the past, ie prior to the time of allocation. In particular, the data may include periods of several hours, weeks and / or months as well as annually recurring events. Especially contain the computing capacity specific information about the utilization and / or availability of computing capacity from the past, which are several minutes (for example, at least one minute), in particular several hours ago (based on the time of assignment of secondary computing tasks to the respective computing capacity). In another embodiment, the data explicitly only includes information about the utilization and / or availability in the past.

At least one of the locally distributed computing capacities is designed in addition to the calculation of the secondary computational tasks for the calculation of primary computational tasks, whereby the computation capacity of the primary computational tasks are prioritized higher than the calculation of the secondary computational tasks. For example, the computing capacity may be one and / or more controller / s in a highly automated vehicle. The computing capacity preferably calculates primary computing tasks during operation of the vehicle so that safe driving of the vehicle is possible. If secondary computing tasks are simultaneously provided to this computing capacity, they will only be calculated if safe operation of the vehicle can be ensured. If this is not the case, only the primary calculation tasks are calculated and the calculation of the secondary calculation tasks is postponed to a later date. The calculation of the secondary computational tasks can take place, for example, in the parked state of the vehicle and / or at a time at which the vehicle is charged at a charging station, if it is an electrically operated vehicle.

This method offers the advantage that possibly unused computing capacity can be used for the calculation of secondary computing tasks and consequently a better utilization of the available resources can be ensured. As a result, costs can be saved on the part of the job creator and on the part of the owner or owners of the locally distributed computing capacities.

If the locally distributed computing capacities are located in vehicles, this results in a secondary benefit, which can be used profitably for the owner of the vehicle and the environment. In addition, the use of freely available computing capacity can create an extremely powerful network that can be used to calculate highly complex computing tasks in extremely short times.

In a further embodiment of the method, this includes the additional step of determining a selection of locally distributed computing capacities and / or a scope of secondary computing tasks to be assigned to the locally distributed computing capacities as a function of the computing capacity-specific data, the assignment depending on the particular selection and / or the Scope takes place.

This embodiment of the method offers the advantage that it can be determined based on the computing capacity-specific data, which locally distributed computing capacities are expected to be available for currently pending secondary computing tasks and are selected based on this determination for the calculation. This prediction is possible based on the information contained in the computing capacity specific data.

If, for example, several locally distributed computing units are available, an evaluation of the respective computing capacity-specific data indicates that a certain number of computing capacities are likely to be able to provide their full computing power in the near future (for example within the next two hours) and the availability of further computing capacities in the near future Future is associated with a great deal of uncertainty, a selection of the computing capacities available in the near future takes place. In this case, only the selected computational resources receive secondary compute tasks, whereas the secondary compute capacities are not assigned secondary compute tasks. Such a distribution of secondary computation tasks makes it possible to achieve as fast and complete a computation of the secondary computation tasks as possible.

In a further embodiment of the method, the computing capacity-specific data contain information about calculated secondary computation tasks, in particular about required time periods for the computation of the secondary computation tasks.

This embodiment of the method offers the advantage that the assignment of secondary computing tasks can be further optimized. If data are available on the locally distributed computing capacities, from which their average calculation speeds can be derived, these computing capacities can be assigned secondary computing tasks tailored to their computing speed. The computing capacity-specific data can in this case be retrieved and / or determined individually for each computing capacity available in the pool.

In a further embodiment of the method, this includes the additional step of Receiving computing capacity specific data.

This embodiment has the advantage that the system that distributes the secondary computing tasks / the computing tasks that distribute the tasks does not necessarily have to acquire and / or store data of individual computing capacities that are specific to the computing capacity, since these data are sent from the computing capacities to the system. This also makes it possible to directly assess the performance of a locally distributed computing capacity newly available to the system, even if it has not yet been assigned any secondary computing tasks by the system. In addition, the locally distributed computing capacities may be used to determine more exact evaluations of their utilization in the past, which increases the quality of the computing capacity-specific data. If locally distributed computing capacities are allocated by secondary systems to secondary computing tasks, for example, this information may not be captured by the systems distributing the individual secondary computing tasks.

In a further embodiment of the method, the received computing capacity-specific data comprise information about a location profile of the computing capacity.

This embodiment of the method offers the advantage that based on the course of the location, a prediction can be made about the future availability of the respective computing capacity. If the locally distributed computing capacity is located, for example, in a vehicle, the course of the location can be used to determine at what times the vehicle preferably stays at a charging station or is not used for driving. Based on this, secondary computing tasks can be assigned to the corresponding computing capacity primarily at these times.

If the locally distributed computing capacities are located, for example, in mobile telephones or laptops, the location profiles can also be used for the prediction of the availability of computing power of this computing capacity.

In a further embodiment of the method, the computing capacity-specific data comprise information about the utilization of the computing capacity by the calculation of primary computing tasks.

The information about the utilization of the computing capacity by primary computing tasks can be sent directly from the locally distributed computing capacity to the secondary computing tasks distributing system. Based on the utilization of the computing capacity in the past, it can again be predicted at what times it is expected how much computing power of the computing capacity will be available for the calculation of secondary computing tasks.

In a further embodiment of the method, the computing capacity-specific data comprise information about the number and / or the extent of calculated primary computing tasks, in particular within a defined period and / or season and / or month and / or weekday and / or time-dependent.

If the computing capacity-specific data also contain information about the number and / or the extent of calculated primary computing tasks in the past, a further optimized prediction of the available computing power of this computing capacity can be carried out in the future. If the computing capacity-specific data also contains information about the capacity utilization over certain periods of time, it is possible to detect patterns which make it possible to make almost exact predictions about the availability of the respective computing capacity.

In a further embodiment of the method, the computing capacity is a computing capacity arranged in a mobile unit, in particular a highly or fully automated vehicle and / or a smartphone and / or a laptop, and the selection and / or the scope will / will become determined as a function of a state of charge of the mobile unit.

This embodiment of the method offers the advantage that a further important criterion is available for the assignment of the secondary computing tasks to locally distributed computing capacities, by means of which the available computing power of the computing capacities can be predicted in the future. Since the computation of secondary computation tasks on the locally distributed units may require a lot of computational power, it is necessary to ensure sufficient power supply of the locally distributed computation units while calculating secondary computation tasks. By adapting secondary calculation tasks to locally distributed computing capacities, the calculation of the secondary calculation tasks can be optimized as a whole. In addition, the acceptance of the owner of a locally distributed computing capacity increases when the state of charge is taken into account when assigning secondary computing tasks. In addition, it can be ensured that the respective locally distributed computing capacity continues to be used for the calculation of the secondary calculation tasks Primary computational tasks is available without having to initially charge the computing capacity containing mobile unit. Under a state of charge, for example, the available capacity of a battery and / or a battery can be understood. In addition, a state of charge may indicate whether a device is currently being charged and / or powered by an external power source.

The use of locally distributed computing capacities, which are located in vehicles, smartphones and / or laptops, is particularly advantageous due to the very high-performance processors / computing capacities of these devices and the low utilization. For example, highly automated vehicles spend most of their life in garages or on roads and are not operated. The computing power available in the vehicles is therefore also not exploited or made available. By providing this computing power, the owner of the vehicle could take additional money while relieving the environment by making better use of resources.

In another embodiment of the method, the computing capacity specific data includes information from external data sources about the future, such as weather forecasts or calendar entries of the owner of the computing capacity.

This embodiment of the method offers the advantage that the duration of the availability of the free computing capacity can be optimized more precisely for individual behavior patterns of the owner and interruptions of arithmetic tasks are minimized.

• - Aussenden von rechenkapazitätsspezifischen Daten, welche Informationen über eine Auslastung und/oder Verfügbarkeit der Rechenkapazität in der Vergangenheit umfassen;
• - Empfangen von Sekundärrechenaufgaben basierend auf den ausgesendeten Daten;
• - Berechnen von Ergebnissen und/oder Zwischenergebnissen der Sekundärrechenaufgaben;
• - Aussenden der Ergebnisse und/oder Zwischenergebnisse.

Furthermore, a method for calculating secondary computing tasks on a computing capacity is claimed. This includes the following steps:

• - transmitting computing capacity specific data that includes information about past utilization and / or availability of computing capacity;
• - receiving secondary computational tasks based on the transmitted data;
• - calculating results and / or intermediate results of the secondary calculation tasks;
• - sending the results and / or intermediate results.

This method is preferably based on a locally distributed computing capacity and also offers the advantages mentioned above.

Furthermore, a control unit is claimed, in particular a control unit arranged in a vehicle, having a computing capacity and a transmitting and receiving interface. The control unit is characterized in that it is adapted to carry out a method according to the method for calculating secondary calculation tasks described above.

In addition, a central processing unit for assigning secondary computing tasks to locally distributed computing capacity is claimed. In this case, at least one of the distributed computing capacities is designed for the calculation of secondary computing tasks provided by the central processing unit and the calculation of primary computing tasks, wherein this computing capacity is designed such that it prioritizes the primary computing tasks higher than the secondary computing tasks assigned by the central processing unit for the distribution of its capacity , The central processing unit is characterized in that it is designed in such a way that the allocation takes place as a function of computing capacity-specific data, which includes information about a utilization and / or availability of the computing capacities in the past.

In a further embodiment of the central processing unit, this is also designed to execute one of the above-described methods for the assignment of secondary computing tasks to at least two locally distributed computing capacities.

In addition, a computer program is claimed which includes instructions that cause a computer to execute one of the methods described above when executing the computer program.

Furthermore, a computer-readable data carrier is claimed on which a corresponding computer program is stored.

• 1 zeigt eine beispielhafte Anwendung des Verfahrens.
• 2 zeigt ein schematisches Verfahrensdiagramm.
• 3 zeigt ein weiteres schematisches Verfahrensdiagramm.

drawings

• 1 shows an exemplary application of the method.
• 2 shows a schematic process diagram.
• 3 shows another schematic process diagram.

embodiments

In 1 An example application of the claimed method and apparatus is shown. 1 shows a user interface 102 , assigning a secondary calculation task unit 101 and locally distributed computing resources 103a . 103b ,

From a user in this embodiment, the user interface 102 Calculation packages to secondary computing tasks assigning unit 101 Posted. On this unit 101 this is going on 2 shown method, which in step 201 starts.

In step 202 receives the unit 101 the one from the user interface 102 received computational packets and determined based on these secondary computational tasks. In this embodiment, a simulation is required by a user. The calculations required for this purpose become a unit assigned by the secondary calculation tasks 101 decomposed into secondary computational tasks, which can be calculated from several separated computational capacities.

In step 203 the previously determined secondary computational tasks are taken from the unit 101 locally distributed computing capacity 103a . 103b assigned. The assignment takes place here as a function of computing capacity-specific data, which information about a load and in this embodiment also on the availability of the respective computing capacity 103a . 103b in the past. These data were from the unit 101 in the assignment of secondary calculation tasks to the respective computing capacities 103a . 103b recorded in the past.

The procedure ends in step 204 ,

In a specific embodiment, the locally distributed computing capacities 103a . 103b computing capacity in highly automated vehicles. The computing power of this computing capacity is available in particular for calculating secondary computing tasks when the vehicles are parked and consequently do not calculate primary computing tasks that are performed, for example, within the framework of driver assistance functions.

The computing capacity in the highly automated vehicles runs in each case in 3 imaged methods, which in step 301 starts.

In step 302 Computing capacities are used to send computing capacity-specific data, which includes information about utilization and availability of computing capacity in the past. In this embodiment, this data contains information about secondary tasks computed by the computing capacities in the past, especially about required time periods for the calculation of the secondary tasks.

In step 303 Secondary computational tasks are received based on the transmitted computationally specific data.

In step 304 These secondary calculation tasks are calculated, producing intermediate results and results.

These results and intermediate results will be in step 305 via an appropriate interface to the unit 101 Posted. The transmission of the data takes place in this embodiment via the mobile network to which the vehicle is connected. If the vehicle is in a parking state and can be connected to a WLAN during this time, the transmission of the data takes place via this network. If the vehicle is coupled to a LAN network during a charging process, the transmission can also take place via this network.

The procedure ends in step 305 ,

In a further embodiment, the assignment of secondary computing tasks in step 203 depending on a state of charge of the vehicles.

In another embodiment, the unit 101 computing capacity specific data received. These are sent out by the respective computing capacities and contain information about the availability and / or utilization of these computing capacities in the past.

In a further embodiment, the method for assigning secondary computing tasks is performed on a central processing unit, which independently creates secondary computing tasks. In this case, it is a central computer on which complex CAD simulations are calculated. This central computer divides the simulation packages into secondary calculation tasks and distributes these according to the method to a plurality of locally distributed computing capacities. For this purpose, a computer program stored on a data carrier accessible to the central computer is executed, which causes the central computer to execute the assignment.

In a further embodiment, the methods described above are used for marketing computing capacities in highly automated cars, which have control devices with large computing capacities and interfaces for connection to the Internet. A marketing company, in this embodiment an automobile manufacturer, markets computing capacity on behalf of the car owner. A customer, for whom a calculation task is to be solved, hands over a calculation package to the marketing company. The computer package is split on a computer or server of the marketing company into secondary computing tasks, which are distributed via the mobile network and via WLAN to several computing capacities of different car owners. The calculation of the secondary calculation tasks takes place in the vehicles, from which the results are then sent back to the server of the marketing company. After successful calculation, the results are transmitted to the customer and the service is billed to the customer. For this purpose, the car owner receives a pro rata remuneration, which is determined based on the secondary calculation tasks calculated by his vehicle. The marketing company reserves a commission for the assignment of the secondary calculation tasks and the associated services.

As soon as the owner wants to use his car, the calculation of the secondary calculation tasks is interrupted / aborted, so that there are no restrictions on the use of the vehicle. In order to reduce aborted calculation packages and to optimize the calculation speed of the calculation packages over time, the marketing company records analyzes regarding the usage behavior of the individual vehicles. Furthermore, the individual vehicles send vehicle-specific data about their use in the past, especially about the utilization of their computing capacity in the past, to the marketing company.

Claims (14)

Method for assigning secondary computational tasks to at least two locally distributed computing capacities (103a, 103b), in particular to computational capacities arranged in vehicles, at least one computation capacity (103a) being designed in addition to the computation of the secondary computation tasks for the computation of primary computational tasks and on this computation capacity (103a) the calculation of the primary compute tasks is higher prioritized than the computation of the secondary computation tasks, with the steps: - obtaining and / or determining secondary computing tasks; Assigning the secondary computational tasks to locally distributed computing capacities (103a, 103b) as a function of computation-specific data, which include information about a utilization and / or availability of the computational capacities in the past. Method according to Claim 1 with the additional step of determining a selection of locally distributed computing capacities (103a, 103b) and / or a scope of secondary computing tasks to be assigned to the locally distributed computing capacities (103a, 103b) depending on the computing capacity specific data, the allocation depending on the particular selection and / or scope. Method according to one of the preceding claims, characterized in that the computing capacity-specific data contain information about calculated secondary tasks, in particular about required time periods for the calculation of the secondary tasks. Method according to one of the preceding claims, with the additional step of receiving computing capacity specific data. Method according to Claim 4 , characterized in that the received computing capacity specific data include information about a location history of the computing capacity (103a). Method according to one of Claims 4 or 5 , characterized in that the computing capacity-specific data comprise information about the utilization of the computing capacity (103a) by the calculation of primary computing tasks. Method according to one of Claims 4 to 6 , characterized in that the computing capacity-specific data include information about the number and / or scope of calculated primary computational tasks, in particular within a defined period and / or season and / or month and / or weekday and / or time-dependent. Method according to one of the preceding claims, wherein the computing capacity (103a) is a computing capacity arranged in a mobile unit, in particular a highly or fully automated vehicle or a smartphone or a laptop, and the selection and / or the extent depending on a state of charge of the unit are determined. A method of calculating secondary computational tasks on a computational capacity (103a) comprising the steps of: - transmitting computationally specific data including information about past utilization and / or availability of computational capacity (103a); - receiving secondary computational tasks based on the transmitted data; - Calculating results and / or intermediate results of the secondary calculation tasks; - sending the results and / or intermediate results. Control unit, in particular a control unit arranged in a vehicle, having a computing capacity and a transmitting and receiving interface, characterized in that the control unit is adapted to a method according to Claim 9 perform. Central arithmetic unit for the assignment of secondary computing tasks to locally distributed computing capacities, wherein at least one of the distributed computing capacities is designed for the calculation of secondary computing tasks provided by the central processing unit and the calculation of primary computing tasks, wherein this computing capacity is designed such that it distributes its capacity Primary compute tasks higher prioritized over the assigned by the central processing unit secondary computational tasks, characterized in that the central arithmetic unit is designed such that the assignment takes place in dependence on computationally specific data, which includes information about a capacity utilization and / or availability of computational capacity in the past. Central processing unit after Claim 11 , characterized in that the arithmetic unit is adapted to a method according to one of Claims 1 to 8th perform. A computer program comprising instructions for causing a computer to execute a computer program upon execution of the computer program Claims 1 to 9 perform. Computer-readable disk on which the computer program is based Claim 13 is stored.

Images