cuda.bindings._v2.driver.device_get_nv_sci_sync_attributes#

cuda.bindings._v2.driver.device_get_nv_sci_sync_attributes( intptr_t nv_sci_sync_attr_list, int dev, int flags, ) → intptr_t#

Fill in NvSciSync attributes for the given CUDA device.

Returns in nv_sci_sync_attr_list the properties of NvSciSync that dev can support. The list can be used to create an NvSciSync object that matches this device’s capabilities.

The flags controls how the application intends to use the NvSciSync:

CUDA_NVSCISYNC_ATTR_SIGNAL: the application intends to signal an NvSciSync on this device.
CUDA_NVSCISYNC_ATTR_WAIT: the application intends to wait on an NvSciSync on this device.

At least one flag must be set. Both flags are orthogonal and may be combined. If NvSciSyncAttrKey_RequiredPerm is already set in nv_sci_sync_attr_list, this API returns CUDA_ERROR_INVALID_VALUE.

Parameters:

nv_sci_sync_attr_list (intptr_t) – Allocated NvSciSyncAttrList handle to fill.
dev (int) – CUDA device ordinal.
flags (int) – Flags controlling which attributes are set.

Returns:

The same nv_sci_sync_attr_list handle after being populated.

Return type:

intptr_t