Precise suspend and resume of workloads in a processing unit

An apparatus comprising:

• a processing unit including a plurality of compute units configured to implement a plurality of pipelines, wherein the processing unit is configured to:
• execute a first workload in a first subset of the plurality of pipelines;
• execute a second workload in a second subset of the plurality of pipelines, wherein the second workload is dependent upon the first workload;
• suspend the first and second workloads;
• and
• store state information for the first and second workloads in a first memory in response to suspending the first and second workloads.

A method comprising:

• suspending first workloads executing on a first subset of a plurality of pipelines implemented on a first subset of compute units of a processing unit, wherein a dependency exists between the workloads;
• storing state information for the first workloads;
• and
• resuming execution of the first workloads on the first subset of the plurality of pipelines implemented on the first subset of compute units on the basis of the stored state information.