NVIDIA enables GPU System Processor (GSP) on select Tesla/Data Center accelerators

NVIDIA GSP is set to offload tasks traditionally performed by CPUs

With the latest driver, NVIDIA is enabling a new feature within the GPU called GPU System Processor (GSP). 

Starting with 510.39 drivers, NVIDIA is enabling GSP to select Tesla/Data Center GPUs, primarily based on Ampere and Turing architectures. The GSP will offload tasks that were traditionally performed by CPUs, such as GPU initialization and management tasks.

Users can still force-disable GSP, but this means that some display related features or power management related features will not work properly. However, those features are yet to be added to GSP tasks.

Some GPUs include a GPU System Processor (GSP) which can be used to offload GPU initialization and management tasks. This processor is driven by the firmware file /lib/firmware/nvidia/510.39.01/gsp.bin. A few select products currently use GSP by default, and more products will take advantage of GSP in future driver releases.

Offloading tasks which were traditionally performed by the driver on the CPU can improve performance due to lower latency access to GPU hardware internals.

— NVIDIA

There is no indication that GSP might be enabled for consumers products, however, any tasks that can be offloaded from the CPU is probably a good way to go.

One would guess that GSP is based on the RISC-V Falcon microcontroller which was introduced by NVIDIA back in 2016. Assuming that this controller is being used by current-gen NVIDIA GPUS.

* If three IDs are listed, the first is the PCI Device ID, the second is the PCI Subsystem Vendor ID, and the third is the PCI Subsystem Device ID.

Source: NVIDIA