Where does a DOCA service run when it is deployed on a DPU?
Why B is the answer
On the DPU's Arm cores
A DOCA service deployed on a DPU runs on the DPU's Arm cores.
In detail
- DOCA services can run on the DPU's Arm cores.
- DPUs have their own processors.
Use case
An illustrative scenario, not a real incident.
A DOCA service runs on the DPU's Arm cores, away from the host.
See it in different ways
flowchart LR A["Host"] --- B["BlueField Arm cores"] B --> C["DOCA service"] C --> D["Network or telemetry"]
stateDiagram-v2 [*] --> Host Host --> Applications: CPU stays free [*] --> DpuArm DpuArm --> DocaService: infrastructure work DocaService --> [*] Applications --> [*]
mindmap
root(("Server with a DPU"))
n1["Host CPU"]
n2["Applications"]
n3["DPU Arm cores"]
n4["DOCA service"]Key point: DOCA services run on DPU Arm cores.
Why the other options are wrong
- A. On the Slurm controller
- That is a scheduler.
- C. On the BMC of the host
- A BMC is for management.
- D. Inside the GPU's memory
- GPU memory is for models.