Workload execution in deterministic pipelines
A computing system includes a sensor data input chiplet to obtain sensor data from a sensor system of a vehicle, and one or more workload processing chiplets executing workloads in a set of independent pipelines based on the sensor data.
1 . A computing system, comprising:
a sensor data input chiplet to obtain sensor data from a sensor system of a vehicle; and
one or more workload processing chiplets to execute a plurality of workloads in a set of independent pipelines based on the sensor data, the one or more workload processing chiplets implementing a reservation table that includes a plurality of workload entries, each workload entry identifying a corresponding workload and dependency information for the corresponding workload, the dependency information indicating whether a dependency is to be resolved with the corresponding workload before the corresponding workload is executed;
wherein for each workload of the plurality of workloads that is completed, the one or more workload processing chiplets update the dependency information of any workload in the reservation table that is identified as having the dependency on the completed workload, to indicate the dependency has been resolved.
2 . The computing system of claim 1 , wherein the one or more workload processing chiplets execute the workloads in the set of independent pipelines to perform a set of tasks for operating the vehicle.
3 . The computing system of claim 2 , wherein the set of tasks comprise a plurality of: image stitching tasks, sensor fusion tasks, machine learning inference tasks, object detection tasks, object classification tasks, scene understanding tasks, and motion prediction tasks.
4 . The computing system of claim 3 , wherein the one or more workload processing chiplets provide output of the set of independent pipelines to an application program for autonomously operating the vehicle.
5 . The computing system of claim 4 , wherein the output of the set of independent pipelines comprises an inferred sensor view of a surrounding environment of the vehicle to facilitate autonomous operation of the vehicle.
6 . The computing system of claim 1 , wherein the one or more workload processing chiplets execute the workloads in the set of independent pipelines deterministically using the reservation table for the workloads as an out-of-order buffer.
7 . The computing system of claim 1 , wherein each workload entry of the plurality of workload entries includes a cache address for workload data to execute the respective workload.
8 . The computing system of claim 1 , wherein the one or more workload processing chiplets include a central chiplet comprising at least one transient-resistant CPU that executes a subset of the workloads in a plurality of parallel independent pipelines to enable autonomous operation of the vehicle.
9 . The computing system of claim 8 , wherein the central chiplet dynamically compares and verifies output of the plurality of parallel independent pipelines in a functional safety (FuSa) pipeline.
10 . The computing system of claim 9 , wherein the at least one transient-resistant CPU and the FuSa pipeline are provided to facilitate an automotive safety integrity level (ASIL) grade of an autonomous drive system of the vehicle.
11 . The computing system of claim 1 , wherein the one or more workload processing chiplets are included on a system-on-chip (SoC) arrangement of the computing system.
12 . The computing system of claim 11 , wherein the SoC arrangement comprises the sensor data input chiplet, a central chiplet, one or more general compute chiplets, an autonomous drive chiplet, a machine learning accelerator chiplet, and one or more high-bandwidth memory chiplets.
13 . A non-transitory computer readable medium storing instructions that, when executed by one or more processors, cause the one or more processors to:
obtain, by a sensor data input chiplet, sensor data from a sensor system of a vehicle;
execute, by one or more workload processing chiplets, a plurality of workloads in a set of independent pipelines based on the sensor data; and
implement a reservation table, the reservation table including a plurality of workload entries, each workload entry identifying a corresponding workload and dependency information for the corresponding workload, the dependency information indicating whether a dependency is to be resolved with the corresponding workload before the corresponding workload is executed;
wherein for each workload of the plurality of workloads that is completed, update the dependency information of any workload in the reservation table that is identified as having the dependency on the completed workload, to indicate the dependency has been resolved.
14 . The non-transitory computer readable medium of claim 13 , wherein the one or more workload processing chiplets execute the workloads in the set of independent pipelines to perform a set of tasks for operating the vehicle.
15 . The non-transitory computer readable medium of claim 14 , wherein the set of tasks comprise a plurality of: image stitching tasks, sensor fusion tasks, machine learning inference tasks, object detection tasks, object classification tasks, scene understanding tasks, and motion prediction tasks.
16 . The non-transitory computer readable medium of claim 15 , wherein the one or more workload processing chiplets provide output of the set of independent pipelines to an application program for autonomously operating the vehicle.
17 . The non-transitory computer readable medium of claim 16 , wherein the output of the set of independent pipelines comprises an inferred sensor view of a surrounding environment of the vehicle to facilitate autonomous operation of the vehicle.
18 . The non-transitory computer readable medium of claim 13 , wherein the one or more workload processing chiplets execute the workloads in the set of independent pipelines deterministically using the reservation table for the workloads as an out-of-order buffer.
19 . The non-transitory computer readable medium of claim 13 , wherein each workload entry of the plurality of workload entries includes a cache address for workload data to execute the respective workload.
20 . A computer-implemented method, comprising:
obtaining, by a sensor data input chiplet, sensor data from a sensor system of a vehicle;
executing, by one or more workload processing chiplets to execute a plurality of workloads in a set of independent pipelines based on the sensor data; and
implementing a reservation table, the reservation table including a plurality of workload entries, each workload entry identifying a corresponding workload and dependency information for the corresponding workload, the dependency information indicating whether a dependency is to be resolved with the corresponding workload before the corresponding workload is executed;
wherein for each workload of the plurality of workloads that is completed, updating the dependency information of any workload in the reservation table that is identified as having the dependency on the completed workload, to indicate the dependency has been resolved.