This section describes the PDSCH transmit pipeline functions of the cuPHY application programming interface. More...

Data Structures
struct	_cuphyPdschCellDynPrm

struct	_cuphyPdschCellGrpDynPrm

struct	_cuphyPdschCwPrm

struct	_cuphyPdschDataIn

struct	_cuphyPdschDataOut

struct	_cuphyPdschDbgPrms

struct	_cuphyPdschDmrsPrm

struct	_cuphyPdschDynPrms

struct	_cuphyPdschStatPrms

struct	_cuphyPdschUeGrpPrm

struct	_cuphyPdschUePrm

struct	_cuphyPmW_t

Typedefs
typedef struct cuphyPdschBatchPrm *	cuphyPdschBatchPrmHndl_t

typedef struct cuphyPdschTx *	cuphyPdschTxHndl_t

Enumerations
enum	_cuphyPdschProcMode { PDSCH_PROC_MODE_NO_GRAPHS = 0, PDSCH_PROC_MODE_GRAPHS = 1, PDSCH_PROC_MODE_SETUP_ONCE_FALLBACK = 2, PDSCH_INTER_CELL_BATCHING = 4, PDSCH_MAX_PROC_MODES }

Functions
cuphyStatus_t	cuphyBatchPdschTx (cuphyPdschTxHndl_t pdschTxHndl, cuphyPdschDynPrms_t *pDynPrms, cuphyPdschBatchPrmHndl_t batchPrmHndl)
	Batch PDSCH workoad. More...

cuphyStatus_t	cuphyCreatePdschBatchPrm (cuphyPdschBatchPrmHndl_t *pBatchPrmHndl)
	Allocate a container for PDSCH batch parameters. More...

cuphyStatus_t	cuphyCreatePdschTx (cuphyPdschTxHndl_t pPdschTxHndl, cuphyPdschStatPrms_t const pStatPrms)
	Allocate and initialize a cuPHY PDSCH pipeline. More...

cuphyStatus_t	cuphyDestroyPdschBatchPrm (cuphyPdschBatchPrmHndl_t batchPrmHndl)
	Destroy container for PDSCH batch parameters. More...

cuphyStatus_t	cuphyDestroyPdschTx (cuphyPdschTxHndl_t pdschTxHndl)
	Destroy a cuPHY PDSCH transmit pipeline object. More...

cuphyStatus_t	cuphyFallbackBufferSetupPdschTx (cuphyPdschTxHndl_t pdschTxHndl, void *pAddr, cudaStream_t strm)
	Fallback single-cell output buffer setup for cuPHY PDSCH pipeline for slot processing. More...

cuphyStatus_t	cuphyFallbackBuffersSetupPdschTx (cuphyPdschTxHndl_t pdschTxHndl, void **pAddr, int fallback_cells, cudaStream_t strm)
	Fallback output buffers setup for multiple cells in cell group for cuPHY PDSCH pipeline for slot processing. More...

cuphyStatus_t	cuphyRunPdschTx (cuphyPdschTxHndl_t pdschTxHndl, uint64_t procModeBmsk)
	Run cuPHY PDSCH pipeline processing in specified mode. More...

cuphyStatus_t	cuphySetupPdschTx (cuphyPdschTxHndl_t pdschTxHndl, cuphyPdschDynPrms_t *pDynPrms, cuphyPdschBatchPrmHndl_t const batchPrmHndl)
	Setup cuPHY PDSCH pipeline for slot processing. More...

Detailed Description

Typedef Documentation

◆ cuphyPdschBatchPrmHndl_t

typedef struct cuphyPdschBatchPrm* cuphyPdschBatchPrmHndl_t

PDSCH batch configuration handle

◆ cuphyPdschTxHndl_t

typedef struct cuphyPdschTx* cuphyPdschTxHndl_t

cuPHY PDSCH transmitter handle

Enumeration Type Documentation

◆ _cuphyPdschProcMode

enum _cuphyPdschProcMode

PDSCH processing modes.
Think of it as a bitmask [...] B2 B1 B0, where B0 is the least significant bit.
B0: streams (0) or graphs (1) mode
B1: setup once fallback if 1; default 0
B2: inter-cell kernel batching if 1; default 0. Only applicable when there are multiple cells in a cell group.

Enumerator
PDSCH_PROC_MODE_NO_GRAPHS	processing without CUDA GRAPHS
PDSCH_PROC_MODE_GRAPHS	processing with CUDA graphs
PDSCH_PROC_MODE_SETUP_ONCE_FALLBACK	Force reset of TB-CRC buffers when running the same TV back to back with only a single setup; used to guarantee correctness. Use this mode with a bitwise or with any of the other modes (e.g., 0 to 1). Ensure value, currently 2, has only one bit set which does not overlap with any of the other modes.
PDSCH_INTER_CELL_BATCHING	Enable inter-cell kernel batching if set. Mode only takes effect when there are multiple cells in a cell group. Used this mode with a bitwise OR with other modes. Ensure value, currently 4, has only one bit set which does not overlap with any of the other modes.
PDSCH_MAX_PROC_MODES	maximum number of processing modes supported

Function Documentation

◆ cuphyBatchPdschTx()

cuphyStatus_t cuphyBatchPdschTx	(	cuphyPdschTxHndl_t	pdschTxHndl,
		cuphyPdschDynPrms_t *	pDynPrms,
		cuphyPdschBatchPrmHndl_t	batchPrmHndl
	)

Batch PDSCH workload across one or more cells, UE-groups and UEs. The batched configuration is used during slot execution

Returns CUPHY_STATUS_INVALID_ARGUMENT if pdschTxHndl and/or pDynPrms (or its components) and/or batchPrmHndl is NULL

Returns CUPHY_STATUS_SUCCESS if setup is successful.

Parameters

pdschTxHndl	- Handle of PdschTx instance to be setup
pDynPrms	- Dynamic parameters carrying information needed for slot processing
batchPrmHndl	- Workload batching information

Returns: CUPHY_STATUS_SUCCESS, CUPHY_STATUS_INVALID_ARGUMENT

See also: cuphyStatus_t,cuphyGetErrorName,cuphyGetErrorString,cuphyCreatePdschTx,cuphyRunPdschTx,cuphyDestroyPdschTx

◆ cuphyCreatePdschBatchPrm()

cuphyStatus_t cuphyCreatePdschBatchPrm ( cuphyPdschBatchPrmHndl_t * pBatchPrmHndl )

Allocate storage to hold PDSCH batch parameters and return a handle in the address provided by the caller.

Returns CUPHY_STATUS_INVALID_ARGUMENT if pBatchPrmHndl is NULL.

Returns CUPHY_STATUS_ALLOC_FAILED if a batch parameter container object cannot be allocated

Returns CUPHY_STATUS_SUCCESS if allocation were successful.

Parameters

pBatchPrmHndl - Address to return the container for PDSCH batch parameters

Returns: CUPHY_STATUS_SUCCESS, CUPHY_STATUS_ALLOC_FAILED, CUPHY_STATUS_INVALID_ARGUMENT

See also: cuphyStatus_t,cuphyGetErrorName,cuphyGetErrorString,cuphyDestroyPdschBatchPrm

◆ cuphyCreatePdschTx()

cuphyStatus_t cuphyCreatePdschTx	(	cuphyPdschTxHndl_t *	pPdschTxHndl,
		cuphyPdschStatPrms_t const *	pStatPrms
	)

Allocate a cuPHY PDSCH transmitter pipeline and returns a handle in the address provided by the caller.

Returns CUPHY_STATUS_INVALID_ARGUMENT if pPdschTxHndl and/or pStatPrms is NULL.

Returns CUPHY_STATUS_ALLOC_FAILED if a PdschTx object cannot be allocated

Returns CUPHY_STATUS_SUCCESS if allocation and initialization were successful.

Parameters

pPdschTxHndl	- Address to return the new PdschTx instance
pStatPrms	- Pointer to PDSCH static parameters to be used in pipeline creation

Returns: CUPHY_STATUS_SUCCESS, CUPHY_STATUS_ALLOC_FAILED, CUPHY_STATUS_INVALID_ARGUMENT

See also: cuphyStatus_t,cuphyGetErrorName,cuphyGetErrorString,cuphySetupPdschTx,cuphyRunPdschTx,cuphyDestroyPdschTx

◆ cuphyDestroyPdschBatchPrm()

cuphyStatus_t cuphyDestroyPdschBatchPrm ( cuphyPdschBatchPrmHndl_t batchPrmHndl )

Destroy a cuPHY context object that was previously created by a call to cuphyCreatePdschBatchPrm. The handle provided to this function should not be used for any operations after this function returns.

Returns CUPHY_STATUS_INVALID_ARGUMENT if batchPrmHndl is NULL.

Returns CUPHY_STATUS_SUCCESS if destruction was successful.

Parameters

batchPrmHndl - previously allocated PDSCH batch parameter instance

Returns: CUPHY_STATUS_SUCCESS, CUPHY_STATUS_INVALID_ARGUMENT

See also: cuphyStatus_t,cuphyCreatePdschBatchPrm

◆ cuphyDestroyPdschTx()

cuphyStatus_t cuphyDestroyPdschTx ( cuphyPdschTxHndl_t pdschTxHndl )

Destroy a cuPHY PDSCH transmitter pipeline object that was previously created by cuphyCreatePdschTx. The handle provided to this function should not be used for any operations after this function returns.

Returns CUPHY_STATUS_INVALID_ARGUMENT if pdschTxHndl is NULL.

Returns CUPHY_STATUS_SUCCESS if destruction was successful.

Parameters

pdschTxHndl - handle to previously allocated PdschTx instance

Returns: CUPHY_STATUS_SUCCESS, CUPHY_STATUS_INVALID_ARGUMENT

See also: cuphyStatus_t,cuphyCreatePdschTx

◆ cuphyFallbackBufferSetupPdschTx()

cuphyStatus_t cuphyFallbackBufferSetupPdschTx	(	cuphyPdschTxHndl_t	pdschTxHndl,
		void *	pAddr,
		cudaStream_t	strm
	)

Setup the output buffer address for cuPHY PDSCH transmitter pipeline in preparation towards slot execution. This function should only be used in PDSCH_PROC_MODE_SETUP_ONCE_FALLBACK mode and requires that cuphySetupPdschTx() has been called once before it for each pipeline object. It supports only one cell per pipeline.

Returns CUPHY_STATUS_INVALID_ARGUMENT if pdschTxHndl and/or pAddr is NULL

Returns CUPHY_STATUS_SUCCESS if buffer setup was successful.

Parameters

pdschTxHndl	- Handle of PdschTx instance to be setup
pAddr	- New address for pipeline's output tensor buffer (single cell only)
strm	- CUDA stream for async. memory copy

Returns: CUPHY_STATUS_SUCCESS, CUPHY_STATUS_INVALID_ARGUMENT

◆ cuphyFallbackBuffersSetupPdschTx()

cuphyStatus_t cuphyFallbackBuffersSetupPdschTx	(	cuphyPdschTxHndl_t	pdschTxHndl,
		void **	pAddr,
		int	fallback_cells,
		cudaStream_t	strm
	)

Setup the output buffer addresses for cuPHY PDSCH transmitter pipeline in preparation towards slot execution. This function should only be used in PDSCH_PROC_MODE_SETUP_ONCE_FALLBACK mode and requires that cuphySetupPdschTx() has been called once before it for each pipeline object. It supports multiple cells per pipeline.

Returns CUPHY_STATUS_INVALID_ARGUMENT if pdschTxHndl and/or pAddr or its elements are NULL

Returns CUPHY_STATUS_SUCCESS if buffers setup was successful.

Parameters

pdschTxHndl	- Handle of PdschTx instance to be setup
pAddr	- Array of addresses for pipeline's output tensor buffers (one buffer per cell). Order of addresses should match the order of dynamic cells in initial, non fallback, setup
fallback_cells	- Number of cells (same as valid elements in pAddr array). Should match number of cells during initial, non fallback, setup.
strm	- CUDA stream for async. memory copy

Returns: CUPHY_STATUS_SUCCESS, CUPHY_STATUS_INVALID_ARGUMENT

◆ cuphyRunPdschTx()

cuphyStatus_t cuphyRunPdschTx	(	cuphyPdschTxHndl_t	pdschTxHndl,
		uint64_t	procModeBmsk
	)

Call triggers cuPHY PDSCH transmitter pipeline execution in mode specified by procModeBmsk

Returns CUPHY_STATUS_INVALID_ARGUMENT if pdschTxHndl is NULL and/or procModeBmsk is not supported.

Returns CUPHY_STATUS_SUCCESS if PdschTx execution is successful.

Parameters

pdschTxHndl	- Handle of PdschTx instance which is to be triggered
procModeBmsk	- Processing mode bitmask containing one or more processing modes applicable during this execution

Returns: CUPHY_STATUS_SUCCESS, CUPHY_STATUS_INVALID_ARGUMENT

See also: cuphyStatus_t,cuphyGetErrorName,cuphyGetErrorString,cuphyCreatePdschTx,cuphySetupPdschTx,cuphyDestroyPdschTx

◆ cuphySetupPdschTx()

cuphyStatus_t cuphySetupPdschTx	(	cuphyPdschTxHndl_t	pdschTxHndl,
		cuphyPdschDynPrms_t *	pDynPrms,
		cuphyPdschBatchPrmHndl_t const	batchPrmHndl
	)

Setup cuPHY PDSCH transmitter pipeline (and its components) state in preparation towards slot execution

Returns CUPHY_STATUS_INVALID_ARGUMENT if pdschTxHndl and/or pDynPrms and/or batchPrmHndl is NULL

Returns CUPHY_STATUS_SUCCESS if allocation and initialization were successful.

Parameters

pdschTxHndl	- Handle of PdschTx instance to be setup
pDynPrms	- Dynamic parameters carrying information needed for slot processing
batchPrmHndl	- Workload batching information

Returns: CUPHY_STATUS_SUCCESS, CUPHY_STATUS_INVALID_ARGUMENT

See also: cuphyStatus_t,cuphyGetErrorName,cuphyGetErrorString,cuphyCreatePdschTx,cuphyRunPdschTx,cuphyDestroyPdschTx

Data Structures

Typedefs

Enumerations

Functions

Detailed Description

Typedef Documentation

◆ cuphyPdschBatchPrmHndl_t

◆ cuphyPdschTxHndl_t

Enumeration Type Documentation

◆ _cuphyPdschProcMode

Function Documentation

◆ cuphyBatchPdschTx()

◆ cuphyCreatePdschBatchPrm()

◆ cuphyCreatePdschTx()

◆ cuphyDestroyPdschBatchPrm()

◆ cuphyDestroyPdschTx()

◆ cuphyFallbackBufferSetupPdschTx()

◆ cuphyFallbackBuffersSetupPdschTx()

◆ cuphyRunPdschTx()

◆ cuphySetupPdschTx()