5.28. PRM Access
This chapter describes NVML operations that are associated with PRM register reads
Classes
Functions
- nvmlReturn_t nvmlDeviceReadWritePRM_v1 ( nvmlDevice_t device, nvmlPRMTLV_v1_t* buffer )
Functions
- nvmlReturn_t nvmlDeviceReadWritePRM_v1 ( nvmlDevice_t device, nvmlPRMTLV_v1_t* buffer )
-
Parameters
- device
- Identifer of target GPU device
- buffer
- Structure holding the input data in TLV format as well as the PRM register contents in TLV format (in the case of a successful read operation). Note: the input data and any returned data shall be in network byte order.
Returns
- NVML_SUCCESS on success
- NVML_ERROR_INVALID_ARGUMENT if device or buffer are invalid
- NVML_ERROR_NO_PERMISSION if user does not have permission to perform this operation
- NVML_ERROR_NOT_SUPPORTED if this feature is not supported by the device
- NVML_ERROR_ARGUMENT_VERSION_MISMATCH if the version specified in buffer is not supported
Description
Read or write a GPU PRM register. The input is assumed to be in TLV format in network byte order.
For Blackwell or newer fully supported devices.
Supported on Linux only.