cuda.bindings.nvml.device_set_nvlink_bw_mode_async_v1#

cuda.bindings.nvml.device_set_nvlink_bw_mode_async_v1(intptr_t device)#

Set the NvLink Reduced Bandwidth Mode asynchronously for the device. Polling should be done by checking for NVML_GPU_FABRIC_STATE_COMPLETED from device_get_gpu_fabric_info_v().

Parameters:

device (intptr_t) – The identifier of the target device.

Returns:

Reference to: nvmlNvlinkSetBwModeAsync_v1_t.

Return type:

nvmlNvlinkSetBwModeAsync_v1_t