CUDA Runtime API. API Reference Manual (vRelease Version, January 2024)

Data Structures

cudaExternalSemaphoreWaitParams

*cudaExternalSemaphoreWaitNodeParams::paramsArray

Array of external semaphore wait parameters.

7.24. cudaExternalSemaphoreWaitNodeParamsV2

Struct Reference

External semaphore wait node parameters

cudaExternalSemaphore_t

*cudaExternalSemaphoreWaitNodeParamsV2::extSemArray

Array of external semaphore handles.

unsigned int

cudaExternalSemaphoreWaitNodeParamsV2::numExtSems

Number of handles and parameters supplied in extSemArray and paramsArray.

cudaExternalSemaphoreWaitParams

*cudaExternalSemaphoreWaitNodeParamsV2::paramsArray

Array of external semaphore wait parameters.

7.25. cudaExternalSemaphoreWaitParams

Struct Reference

External semaphore wait parameters, compatible with driver type

void *cudaExternalSemaphoreWaitParams::fence

Pointer to NvSciSyncFence. Valid if cudaExternalSemaphoreHandleType is of type

cudaExternalSemaphoreHandleTypeNvSciSync.

568

Data Structures

cudaExternalSemaphoreWaitParams::@26::@27

cudaExternalSemaphoreWaitParams::fence

Parameters for fence objects

unsigned int cudaExternalSemaphoreWaitParams::flags

Only when cudaExternalSemaphoreSignalParams is used to signal a cudaExternalSemaphore_t

of type cudaExternalSemaphoreHandleTypeNvSciSync, the valid flag is

cudaExternalSemaphoreSignalSkipNvSciBufMemSync: which indicates that while waiting for the

cudaExternalSemaphore_t, no memory synchronization operations should be performed for any

external memory object imported as cudaExternalMemoryHandleTypeNvSciBuf. For all other types of

cudaExternalSemaphore_t, flags must be zero.

unsigned long long

cudaExternalSemaphoreWaitParams::key

Value of key to acquire the mutex with

cudaExternalSemaphoreWaitParams::@26::@29

cudaExternalSemaphoreWaitParams::keyedMutex

Parameters for keyed mutex objects

unsigned int

cudaExternalSemaphoreWaitParams::timeoutMs

Timeout in milliseconds to wait to acquire the mutex

unsigned long long

cudaExternalSemaphoreWaitParams::value

Value of fence to be waited on

7.26. cudaExternalSemaphoreWaitParams_v1

Struct Reference

External semaphore wait parameters(deprecated)

569

Data Structures

void *cudaExternalSemaphoreWaitParams_v1::fence

Pointer to NvSciSyncFence. Valid if cudaExternalSemaphoreHandleType is of type

cudaExternalSemaphoreHandleTypeNvSciSync.

cudaExternalSemaphoreWaitParams_v1::@18::@19

cudaExternalSemaphoreWaitParams_v1::fence

Parameters for fence objects

unsigned int cudaExternalSemaphoreWaitParams_v1::flags

Only when cudaExternalSemaphoreSignalParams is used to signal a cudaExternalSemaphore_t

of type cudaExternalSemaphoreHandleTypeNvSciSync, the valid flag is

cudaExternalSemaphoreSignalSkipNvSciBufMemSync: which indicates that while waiting for the

cudaExternalSemaphore_t, no memory synchronization operations should be performed for any

external memory object imported as cudaExternalMemoryHandleTypeNvSciBuf. For all other types of

cudaExternalSemaphore_t, flags must be zero.

unsigned long long

cudaExternalSemaphoreWaitParams_v1::key

Value of key to acquire the mutex with

cudaExternalSemaphoreWaitParams_v1::@18::@21

cudaExternalSemaphoreWaitParams_v1::keyedMutex

Parameters for keyed mutex objects

unsigned int

cudaExternalSemaphoreWaitParams_v1::timeoutMs

Timeout in milliseconds to wait to acquire the mutex

unsigned long long

cudaExternalSemaphoreWaitParams_v1::value

Value of fence to be waited on

570

Data Structures

7.27. cudaFuncAttributes Struct Reference

CUDA function attributes

int cudaFuncAttributes::binaryVersion

The binary architecture version for which the function was compiled. This value is the major binary

version * 10 + the minor binary version, so a binary version 1.3 function would return the value 13.

int cudaFuncAttributes::cacheModeCA

The attribute to indicate whether the function has been compiled with user specified option "-Xptxas --

dlcm=ca" set.

int cudaFuncAttributes::clusterDimMustBeSet

If this attribute is set, the kernel must launch with a valid cluster dimension specified.

int cudaFuncAttributes::clusterSchedulingPolicyPreference

The block scheduling policy of a function. See cudaFuncSetAttribute

size_t cudaFuncAttributes::constSizeBytes

The size in bytes of user-allocated constant memory required by this function.

size_t cudaFuncAttributes::localSizeBytes

The size in bytes of local memory used by each thread of this function.

int cudaFuncAttributes::maxDynamicSharedSizeBytes

The maximum size in bytes of dynamic shared memory per block for this function. Any launch must

have a dynamic shared memory size smaller than this value.

int cudaFuncAttributes::maxThreadsPerBlock

The maximum number of threads per block, beyond which a launch of the function would fail. This

number depends on both the function and the device on which the function is currently loaded.

571

Data Structures

int cudaFuncAttributes::nonPortableClusterSizeAllowed

Whether the function can be launched with non-portable cluster size. 1 is allowed, 0 is disallowed. A

non-portable cluster size may only function on the specific SKUs the program is tested on. The launch

might fail if the program is run on a different hardware platform.

CUDA API provides cudaOccupancyMaxActiveClusters to assist with checking whether the desired

size can be launched on the current device.

Portable Cluster Size

A portable cluster size is guaranteed to be functional on all compute capabilities higher than the target

compute capability. The portable cluster size for sm_90 is 8 blocks per cluster. This value may increase

for future compute capabilities.

The specific hardware unit may support higher cluster sizes that’s not guaranteed to be portable. See

cudaFuncSetAttribute

int cudaFuncAttributes::numRegs

The number of registers used by each thread of this function.

int cudaFuncAttributes::preferredShmemCarveout

On devices where the L1 cache and shared memory use the same hardware resources, this sets

the shared memory carveout preference, in percent of the maximum shared memory. Refer to

cudaDevAttrMaxSharedMemoryPerMultiprocessor. This is only a hint, and the driver can choose a

different ratio if required to execute the function. See cudaFuncSetAttribute

int cudaFuncAttributes::ptxVersion

The PTX virtual architecture version for which the function was compiled. This value is the major PTX

version * 10 + the minor PTX version, so a PTX version 1.3 function would return the value 13.

int cudaFuncAttributes::requiredClusterWidth

The required cluster width/height/depth in blocks. The values must either all be 0 or all be positive. The

validity of the cluster dimensions is otherwise checked at launch time.

If the value is set during compile time, it cannot be set at runtime. Setting it at runtime should return

cudaErrorNotPermitted. See cudaFuncSetAttribute

int cudaFuncAttributes::reserved

Reserved for future use.

572

Data Structures

size_t cudaFuncAttributes::sharedSizeBytes

The size in bytes of statically-allocated shared memory per block required by this function. This does

not include dynamically-allocated shared memory requested by the user at runtime.

7.28. cudaGraphEdgeData Struct Reference

Optional annotation for edges in a CUDA graph. Note, all edges implicitly have annotations and

default to a zero-initialized value if not specified. A zero-initialized struct indicates a standard full

serialization of two nodes with memory visibility.

unsigned char cudaGraphEdgeData::from_port

This indicates when the dependency is triggered from the upstream node on the edge. The

meaning is specfic to the node type. A value of 0 in all cases means full completion of the

upstream node, with memory visibility to the downstream node or portion thereof (indicated by

to_port). Only kernel nodes define non-zero ports. A kernel node can use the following output

port types: cudaGraphKernelNodePortDefault, cudaGraphKernelNodePortProgrammatic, or

cudaGraphKernelNodePortLaunchCompletion.

unsigned char cudaGraphEdgeData::reserved

These bytes are unused and must be zeroed. This ensures compatibility if additional fields are added in

the future.

unsigned char cudaGraphEdgeData::to_port

This indicates what portion of the downstream node is dependent on the upstream node or portion

thereof (indicated by from_port). The meaning is specific to the node type. A value of 0 in all cases

means the entirety of the downstream node is dependent on the upstream work. Currently no node

types define non-zero ports. Accordingly, this field must be set to zero.

unsigned char cudaGraphEdgeData::type

This should be populated with a value from cudaGraphDependencyType. (It is typed as char due to

compiler-specific layout of bitfields.) See cudaGraphDependencyType.

573

Data Structures

7.29. cudaGraphExecUpdateResultInfo Struct

Reference

Result information returned by cudaGraphExecUpdate

cudaGraphNode_t

cudaGraphExecUpdateResultInfo::errorFromNode

The from node of error edge when the topologies do not match. Otherwise NULL.

cudaGraphNode_t

cudaGraphExecUpdateResultInfo::errorNode

The "to node" of the error edge when the topologies do not match. The error node when the error is

associated with a specific node. NULL when the error is generic.

enumcudaGraphExecUpdateResult

cudaGraphExecUpdateResultInfo::result

Gives more specific detail when a cuda graph update fails.

7.30. cudaGraphInstantiateParams Struct

Reference

Graph instantiation parameters

cudaGraphNode_t

cudaGraphInstantiateParams::errNode_out

The node which caused instantiation to fail, if any

unsigned long long cudaGraphInstantiateParams::flags

Instantiation flags

574

Data Structures

cudaGraphInstantiateResult

cudaGraphInstantiateParams::result_out

Whether instantiation was successful. If it failed, the reason why

cudaStream_t cudaGraphInstantiateParams::uploadStream

Upload stream

7.31. cudaGraphKernelNodeUpdate Struct

Reference

Struct to specify a single node update to pass as part of a larger array to

cudaGraphKernelNodeUpdatesApply

enumcudaGraphKernelNodeField

cudaGraphKernelNodeUpdate::field

Which type of update to apply. Determines how updateData is interpreted

uint3 cudaGraphKernelNodeUpdate::gridDim

Grid dimensions

unsigned int cudaGraphKernelNodeUpdate::isEnabled

Node enable/disable data. Nonzero if the node should be enabled, 0 if it should be disabled

cudaGraphDeviceNode_t

cudaGraphKernelNodeUpdate::node

Node to update

size_t cudaGraphKernelNodeUpdate::offset

Offset into the parameter buffer at which to apply the update

575

Data Structures

cudaGraphKernelNodeUpdate::@32::@33

cudaGraphKernelNodeUpdate::param

Kernel parameter data

const void *cudaGraphKernelNodeUpdate::pValue

Kernel parameter data to write in

size_t cudaGraphKernelNodeUpdate::size

Number of bytes to update

cudaGraphKernelNodeUpdate::@32

cudaGraphKernelNodeUpdate::updateData

Update data to apply. Which field is used depends on field's value

7.32. cudaGraphNodeParams Struct Reference

Graph node parameters. See cudaGraphAddNode.

struct cudaMemAllocNodeParamsV2

cudaGraphNodeParams::alloc

Memory allocation node parameters.

struct cudaConditionalNodeParams

cudaGraphNodeParams::conditional

Conditional node parameters.

struct cudaEventRecordNodeParams

cudaGraphNodeParams::eventRecord

Event record node parameters.

576

Data Structures

struct cudaEventWaitNodeParams

cudaGraphNodeParams::eventWait

Event wait node parameters.

struct cudaExternalSemaphoreSignalNodeParamsV2

cudaGraphNodeParams::extSemSignal

External semaphore signal node parameters.

struct cudaExternalSemaphoreWaitNodeParamsV2

cudaGraphNodeParams::extSemWait

External semaphore wait node parameters.

struct cudaMemFreeNodeParams

cudaGraphNodeParams::free

Memory free node parameters.

struct cudaChildGraphNodeParams

cudaGraphNodeParams::graph

Child graph node parameters.

struct cudaHostNodeParamsV2

cudaGraphNodeParams::host

Host node parameters.

struct cudaKernelNodeParamsV2

cudaGraphNodeParams::kernel

Kernel node parameters.

struct cudaMemcpyNodeParams

cudaGraphNodeParams::memcpy

Memcpy node parameters.

577

Data Structures

struct cudaMemsetParamsV2

cudaGraphNodeParams::memset

Memset node parameters.

int cudaGraphNodeParams::reserved0

Reserved. Must be zero.

long long cudaGraphNodeParams::reserved1

Padding. Unused bytes must be zero.

long long cudaGraphNodeParams::reserved2

Reserved bytes. Must be zero.

enumcudaGraphNodeType cudaGraphNodeParams::type

Type of the node

7.33. cudaHostNodeParams Struct Reference

CUDA host node parameters

cudaHostFn_t cudaHostNodeParams::fn

The function to call when the node executes

void *cudaHostNodeParams::userData

Argument to pass to the function

7.34. cudaHostNodeParamsV2 Struct

Reference

CUDA host node parameters

578

Data Structures

cudaHostFn_t cudaHostNodeParamsV2::fn

The function to call when the node executes

void *cudaHostNodeParamsV2::userData

Argument to pass to the function

7.35. cudaIpcEventHandle_t Struct Reference

CUDA IPC event handle

7.36. cudaIpcMemHandle_t Struct Reference

CUDA IPC memory handle

7.37. cudaKernelNodeParams Struct Reference

CUDA GPU kernel node parameters

dim3 cudaKernelNodeParams::blockDim

Block dimensions

**cudaKernelNodeParams::extra

Pointer to kernel arguments in the "extra" format

void *cudaKernelNodeParams::func

Kernel to launch

dim3 cudaKernelNodeParams::gridDim

Grid dimensions

**cudaKernelNodeParams::kernelParams

Array of pointers to individual kernel arguments

579

Data Structures

unsigned int cudaKernelNodeParams::sharedMemBytes

Dynamic shared-memory size per thread block in bytes

7.38. cudaKernelNodeParamsV2 Struct

Reference

CUDA GPU kernel node parameters

uint3 cudaKernelNodeParamsV2::blockDim

Block dimensions

**cudaKernelNodeParamsV2::extra

Pointer to kernel arguments in the "extra" format

void *cudaKernelNodeParamsV2::func

Kernel to launch

uint3 cudaKernelNodeParamsV2::gridDim

Grid dimensions

**cudaKernelNodeParamsV2::kernelParams

Array of pointers to individual kernel arguments

unsigned int cudaKernelNodeParamsV2::sharedMemBytes

Dynamic shared-memory size per thread block in bytes

7.39. cudaLaunchAttribute Struct Reference

Launch attribute

cudaLaunchAttributeID cudaLaunchAttribute::id

Attribute to set

580

Data Structures

cudaLaunchAttribute::val

Value of the attribute

7.40. cudaLaunchAttributeValue Union

Reference

Launch attributes union; used as value field of cudaLaunchAttribute

struct cudaAccessPolicyWindow

cudaLaunchAttributeValue::accessPolicyWindow

Value of launch attribute cudaLaunchAttributeAccessPolicyWindow.

cudaLaunchAttributeValue::@34

cudaLaunchAttributeValue::clusterDim

Value of launch attribute cudaLaunchAttributeClusterDimension that represents the desired cluster

dimensions for the kernel. Opaque type with the following fields:

‣ x - The X dimension of the cluster, in blocks. Must be a divisor of the grid X dimension.

‣ y - The Y dimension of the cluster, in blocks. Must be a divisor of the grid Y dimension.

‣ z - The Z dimension of the cluster, in blocks. Must be a divisor of the grid Z dimension.

enumcudaClusterSchedulingPolicy

cudaLaunchAttributeValue::clusterSchedulingPolicyPreference

Value of launch attribute cudaLaunchAttributeClusterSchedulingPolicyPreference. Cluster scheduling

policy preference for the kernel.

int cudaLaunchAttributeValue::cooperative

Value of launch attribute cudaLaunchAttributeCooperative. Nonzero indicates a cooperative kernel

(see cudaLaunchCooperativeKernel).

cudaLaunchAttributeValue::@38

cudaLaunchAttributeValue::deviceUpdatableKernelNode

Value of launch attribute cudaLaunchAttributeDeviceUpdatableKernelNode with the following fields:

581

Data Structures

‣ int deviceUpdatable - Whether or not the resulting kernel node should be device-updatable.

‣ cudaGraphDeviceNode_t devNode - Returns a handle to pass to the various device-side

update functions.

cudaLaunchAttributeValue::@37

cudaLaunchAttributeValue::launchCompletionEvent

Value of launch attribute cudaLaunchAttributeLaunchCompletionEvent with the following fields:

‣ cudaEvent_t event - Event to fire when the last block launches.

‣ int flags - Event record flags, see cudaEventRecordWithFlags. Does not accept

cudaEventRecordExternal.

cudaLaunchMemSyncDomain

cudaLaunchAttributeValue::memSyncDomain

Value of launch attribute cudaLaunchAttributeMemSyncDomain. See cudaLaunchMemSyncDomain.

struct cudaLaunchMemSyncDomainMap

cudaLaunchAttributeValue::memSyncDomainMap

Value of launch attribute cudaLaunchAttributeMemSyncDomainMap. See

cudaLaunchMemSyncDomainMap.

cudaLaunchAttributeValue::@36

cudaLaunchAttributeValue::preferredClusterDim

Value of launch attribute cudaLaunchAttributePreferredClusterDimension that represents the desired

preferred cluster dimensions for the kernel. Opaque type with the following fields:

‣ x - The X dimension of the preferred cluster, in blocks. Must be a divisor of the grid X dimension,

and must be a multiple of the x field of cudaLaunchAttributeValue::clusterDim.

‣ y - The Y dimension of the preferred cluster, in blocks. Must be a divisor of the grid Y dimension,

and must be a multiple of the y field of cudaLaunchAttributeValue::clusterDim.

‣ z - The Z dimension of the preferred cluster, in blocks. Must be equal to the z field of

cudaLaunchAttributeValue::clusterDim.

int cudaLaunchAttributeValue::priority

Value of launch attribute cudaLaunchAttributePriority. Execution priority of the kernel.

582

Data Structures

cudaLaunchAttributeValue::@35

cudaLaunchAttributeValue::programmaticEvent

Value of launch attribute cudaLaunchAttributeProgrammaticEvent with the following fields:

‣ cudaEvent_t event - Event to fire when all blocks trigger it.

‣ int flags; - Event record flags, see cudaEventRecordWithFlags. Does not accept

cudaEventRecordExternal.

‣ int triggerAtBlockStart - If this is set to non-0, each block launch will automatically trigger the

event.

int

cudaLaunchAttributeValue::programmaticStreamSerializationAllowe

Value of launch attribute cudaLaunchAttributeProgrammaticStreamSerialization.

unsigned int

cudaLaunchAttributeValue::sharedMemCarveout

Value of launch attribute cudaLaunchAttributePreferredSharedMemoryCarveout.

enum cudaSynchronizationPolicy

cudaLaunchAttributeValue::syncPolicy

Value of launch attribute cudaLaunchAttributeSynchronizationPolicy. cudaSynchronizationPolicy for

work queued up in this stream.

7.41. cudaLaunchConfig_t Struct Reference

CUDA extensible launch configuration

cudaLaunchAttribute *cudaLaunchConfig_t::attrs

List of attributes; nullable if cudaLaunchConfig_t::numAttrs == 0

dim3 cudaLaunchConfig_t::blockDim

Block dimensions

583

Data Structures

size_t cudaLaunchConfig_t::dynamicSmemBytes

Dynamic shared-memory size per thread block in bytes

dim3 cudaLaunchConfig_t::gridDim

Grid dimensions

unsigned int cudaLaunchConfig_t::numAttrs

Number of attributes populated in cudaLaunchConfig_t::attrs

cudaStream_t cudaLaunchConfig_t::stream

Stream identifier

7.42. cudaLaunchMemSyncDomainMap Struct

Reference

Memory Synchronization Domain map

See cudaLaunchMemSyncDomain.

By default, kernels are launched in domain 0. Kernel launched with

cudaLaunchMemSyncDomainRemote will have a different domain ID. User may also alter the domain

ID with cudaLaunchMemSyncDomainMap for a specific stream / graph node / kernel launch. See

cudaLaunchAttributeMemSyncDomainMap.

Domain ID range is available through cudaDevAttrMemSyncDomainCount.

unsigned char cudaLaunchMemSyncDomainMap::default_

The default domain ID to use for designated kernels

unsigned char cudaLaunchMemSyncDomainMap::remote

The remote domain ID to use for designated kernels

7.43. cudaLaunchParams Struct Reference

CUDA launch parameters

584

Data Structures

**cudaLaunchParams::args

Arguments

dim3 cudaLaunchParams::blockDim

Block dimentions

void *cudaLaunchParams::func

Device function symbol

dim3 cudaLaunchParams::gridDim

Grid dimentions

size_t cudaLaunchParams::sharedMem

Shared memory

cudaStream_t cudaLaunchParams::stream

Stream identifier

7.44. cudaMemAccessDesc Struct Reference

Memory access descriptor

enumcudaMemAccessFlags cudaMemAccessDesc::flags

CUmemProt accessibility flags to set on the request

struct cudaMemLocation cudaMemAccessDesc::location

Location on which the request is to change it's accessibility

7.45. cudaMemAllocNodeParams Struct

Reference

Memory allocation node parameters

585

Data Structures

size_t cudaMemAllocNodeParams::accessDescCount

in: Number of `accessDescs`s

cudaMemAccessDesc

*cudaMemAllocNodeParams::accessDescs

in: number of memory access descriptors. Must not exceed the number of GPUs.

size_t cudaMemAllocNodeParams::bytesize

in: size in bytes of the requested allocation

void *cudaMemAllocNodeParams::dptr

out: address of the allocation returned by CUDA

struct cudaMemPoolProps

cudaMemAllocNodeParams::poolProps

in: location where the allocation should reside (specified in location). handleTypes must be

cudaMemHandleTypeNone. IPC is not supported. in: array of memory access descriptors. Used to

describe peer GPU access

7.46. cudaMemAllocNodeParamsV2 Struct

Reference

Memory allocation node parameters

size_t cudaMemAllocNodeParamsV2::accessDescCount

in: Number of `accessDescs`s

cudaMemAccessDesc

*cudaMemAllocNodeParamsV2::accessDescs

in: number of memory access descriptors. Must not exceed the number of GPUs.

586

Data Structures

size_t cudaMemAllocNodeParamsV2::bytesize

in: size in bytes of the requested allocation

void *cudaMemAllocNodeParamsV2::dptr

out: address of the allocation returned by CUDA

struct cudaMemPoolProps

cudaMemAllocNodeParamsV2::poolProps

in: location where the allocation should reside (specified in location). handleTypes must be

cudaMemHandleTypeNone. IPC is not supported. in: array of memory access descriptors. Used to

describe peer GPU access

7.47. cudaMemcpy3DOperand Struct

Reference

Struct representing an operand for copy with cudaMemcpy3DBatchAsync

cudaMemcpy3DOperand::@7::@9

cudaMemcpy3DOperand::array

Struct representing an operand when cudaMemcpy3DOperand::type is

cudaMemcpyOperandTypeArray

size_t cudaMemcpy3DOperand::layerHeight

Height of each layer in elements.

struct cudaMemLocation cudaMemcpy3DOperand::locHint

Hint location for the operand. Ignored when the pointers are not managed memory or memory allocated

outside CUDA.

587

Data Structures

cudaMemcpy3DOperand::@7::@8

cudaMemcpy3DOperand::ptr

Struct representing an operand when cudaMemcpy3DOperand::type is

cudaMemcpyOperandTypePointer

size_t cudaMemcpy3DOperand::rowLength

Length of each row in elements.

7.48. cudaMemcpy3DParms Struct Reference

CUDA 3D memory copying parameters

cudaArray_t cudaMemcpy3DParms::dstArray

Destination memory address

struct cudaPos cudaMemcpy3DParms::dstPos

Destination position offset

struct cudaPitchedPtr cudaMemcpy3DParms::dstPtr

Pitched destination memory address

struct cudaExtent cudaMemcpy3DParms::extent

Requested memory copy size

enumcudaMemcpyKind cudaMemcpy3DParms::kind

Type of transfer

cudaArray_t cudaMemcpy3DParms::srcArray

Source memory address

struct cudaPos cudaMemcpy3DParms::srcPos

Source position offset

588

Data Structures

struct cudaPitchedPtr cudaMemcpy3DParms::srcPtr

Pitched source memory address

7.49. cudaMemcpy3DPeerParms Struct

Reference

CUDA 3D cross-device memory copying parameters

cudaArray_t cudaMemcpy3DPeerParms::dstArray

Destination memory address

int cudaMemcpy3DPeerParms::dstDevice

Destination device

struct cudaPos cudaMemcpy3DPeerParms::dstPos

Destination position offset

struct cudaPitchedPtr cudaMemcpy3DPeerParms::dstPtr

Pitched destination memory address

struct cudaExtent cudaMemcpy3DPeerParms::extent

Requested memory copy size

cudaArray_t cudaMemcpy3DPeerParms::srcArray

Source memory address

int cudaMemcpy3DPeerParms::srcDevice

Source device

struct cudaPos cudaMemcpy3DPeerParms::srcPos

Source position offset

589

Data Structures

struct cudaPitchedPtr cudaMemcpy3DPeerParms::srcPtr

Pitched source memory address

7.50. cudaMemcpyAttributes Struct Reference

Attributes specific to copies within a batch. For more details on usage see cudaMemcpyBatchAsync.

struct cudaMemLocation

cudaMemcpyAttributes::dstLocHint

Hint location for the destination operand. Ignored when the pointers are not managed memory or

memory allocated outside CUDA.

unsigned int cudaMemcpyAttributes::flags

Additional flags for copies with this attribute. See cudaMemcpyFlags.

enumcudaMemcpySrcAccessOrder

cudaMemcpyAttributes::srcAccessOrder

Source access ordering to be observed for copies with this attribute.

struct cudaMemLocation

cudaMemcpyAttributes::srcLocHint

Hint location for the source operand. Ignored when the pointers are not managed memory or memory

allocated outside CUDA.

7.51. cudaMemcpyNodeParams Struct

Reference

Memcpy node parameters

struct cudaMemcpy3DParms

cudaMemcpyNodeParams::copyParams

Parameters for the memory copy

590

Data Structures

int cudaMemcpyNodeParams::flags

Must be zero

int cudaMemcpyNodeParams::reserved

Must be zero

7.52. cudaMemFreeNodeParams Struct

Reference

Memory free node parameters

void *cudaMemFreeNodeParams::dptr

in: the pointer to free

7.53. cudaMemLocation Struct Reference

Specifies a memory location.

To specify a gpu, set type = cudaMemLocationTypeDevice and set id = the gpu's device ordinal. To

specify a cpu NUMA node, set type = cudaMemLocationTypeHostNuma and set id = host NUMA

node id.

int cudaMemLocation::id

identifier for a given this location's CUmemLocationType.

enumcudaMemLocationType cudaMemLocation::type

Specifies the location type, which modifies the meaning of id.

7.54. cudaMemPoolProps Struct Reference

Specifies the properties of allocations made from the pool.

591

Data Structures

enumcudaMemAllocationType

cudaMemPoolProps::allocType

Allocation type. Currently must be specified as cudaMemAllocationTypePinned

enumcudaMemAllocationHandleType

cudaMemPoolProps::handleTypes

Handle types that will be supported by allocations from the pool.

struct cudaMemLocation cudaMemPoolProps::location

Location allocations should reside.

size_t cudaMemPoolProps::maxSize

Maximum pool size. When set to 0, defaults to a system dependent value.

unsigned char cudaMemPoolProps::reserved

reserved for future use, must be 0

unsigned short cudaMemPoolProps::usage

Bitmask indicating intended usage for the pool.

void *cudaMemPoolProps::win32SecurityAttributes

Windows-specific LPSECURITYATTRIBUTES required when cudaMemHandleTypeWin32 is

specified. This security attribute defines the scope of which exported allocations may be tranferred to

other processes. In all other cases, this field is required to be zero.

7.55. cudaMemPoolPtrExportData Struct

Reference

Opaque data for exporting a pool allocation

592

Data Structures

7.56. cudaMemsetParams Struct Reference

CUDA Memset node parameters

void *cudaMemsetParams::dst

Destination device pointer

unsigned int cudaMemsetParams::elementSize

Size of each element in bytes. Must be 1, 2, or 4.

size_t cudaMemsetParams::height

Number of rows

size_t cudaMemsetParams::pitch

Pitch of destination device pointer. Unused if height is 1

unsigned int cudaMemsetParams::value

Value to be set

size_t cudaMemsetParams::width

Width of the row in elements

7.57. cudaMemsetParamsV2 Struct Reference

CUDA Memset node parameters

void *cudaMemsetParamsV2::dst

Destination device pointer

unsigned int cudaMemsetParamsV2::elementSize

Size of each element in bytes. Must be 1, 2, or 4.

593

Data Structures

size_t cudaMemsetParamsV2::height

Number of rows

size_t cudaMemsetParamsV2::pitch

Pitch of destination device pointer. Unused if height is 1

unsigned int cudaMemsetParamsV2::value

Value to be set

size_t cudaMemsetParamsV2::width

Width of the row in elements

7.58. cudaOffset3D Struct Reference

Struct representing offset into a cudaArray_t in elements

7.59. cudaPitchedPtr Struct Reference

CUDA Pitched memory pointer

CUDA Runtime API. API Reference Manual (vRelease Version, January 2024) - page 9