[C++ API] Returns a channel descriptor using the specified format

Channel descriptor with format f

Returns a channel descriptor with format f and number of bits of each component x, y, z, and w. The cudaChannelFormatDesc is defined as:

‎  struct cudaChannelFormatDesc {
          int x, y, z, w;
          enum cudaChannelFormatKind 
                  f;
        };

where cudaChannelFormatKind is one of cudaChannelFormatKindSigned, cudaChannelFormatKindUnsigned, cudaChannelFormatKindFloat, cudaChannelFormatKindSignedNormalized8X1, cudaChannelFormatKindSignedNormalized8X2, cudaChannelFormatKindSignedNormalized8X4, cudaChannelFormatKindUnsignedNormalized8X1, cudaChannelFormatKindUnsignedNormalized8X2, cudaChannelFormatKindUnsignedNormalized8X4, cudaChannelFormatKindSignedNormalized16X1, cudaChannelFormatKindSignedNormalized16X2, cudaChannelFormatKindSignedNormalized16X4, cudaChannelFormatKindUnsignedNormalized16X1, cudaChannelFormatKindUnsignedNormalized16X2, cudaChannelFormatKindUnsignedNormalized16X4, cudaChannelFormatKindUnsignedNormalized1010102 or cudaChannelFormatKindNV12.

The format is specified by the template specialization.

The template function specializes for the following scalar types: char, signed char, unsigned char, short, unsigned short, int, unsigned int, long, unsigned long, and float. The template function specializes for the following vector types: char{1|2|4}, uchar{1|2|4}, short{1|2|4}, ushort{1|2|4}, int{1|2|4}, uint{1|2|4}, long{1|2|4}, ulong{1|2|4}, float{1|2|4}. The template function specializes for following cudaChannelFormatKind enum values: cudaChannelFormatKind{Uns|S}ignedNormalized{8|16}X{1|2|4}, cudaChannelFormatKindUnsignedNormalized1010102 and cudaChannelFormatKindNV12.

Invoking the function on a type without a specialization defaults to creating a channel format of kind cudaChannelFormatKindNone

See also:

cudaCreateChannelDesc ( Low level), cudaGetChannelDesc,

[C++ API] Creates an event object with the specified flags

event

- Newly created event

flags

- Flags for new event

[C++ API] Find out attributes for a given function

attr

- Return pointer to function's attributes

entry

- Function to get attributes of

Returns the function name for a device entry function pointer.

name

- The returned name of the function

func

- The function pointer to retrieve name for

Returns in **name the function name associated with the symbol func . The function name is returned as a null-terminated string. This API may return a mangled name if the function is not declared as having C linkage. If **name is NULL, cudaErrorInvalidValue is returned. If func is not a device entry function, cudaErrorInvalidDeviceFunction is returned.

• Note that this function may also return error codes from previous, asynchronous launches.

• Note that this function may also return cudaErrorInitializationError, cudaErrorInsufficientDriver or cudaErrorNoDevice if this call tries to initialize internal CUDA RT state.

• Note that as specified by cudaStreamAddCallback no CUDA function may be called from callback. cudaErrorNotPermitted may, but is not guaranteed to, be returned as a diagnostic in such case.

• The API can also be used with a kernel cudaKernel_t by querying the handle using cudaLibraryGetKernel() or cudaGetKernel and then passing it to the API by casting to void*. The symbol entryFuncAddr passed to cudaGetKernel should be a symbol that is registered with the same CUDA Runtime instance.

• Passing a symbol that belongs that belongs to a different runtime instance will result in undefined behavior. The only type that can be reliably passed to a different runtime instance is cudaKernel_t

cudaFuncGetName ( C API)

[C++ API] Set attributes for a given function

func

attr

- Attribute to set

value

- Value to set

This function sets the attributes of a function specified via entry. The parameter entry must be a pointer to a function that executes on the device. The parameter specified by entry must be declared as a __global__ function. The enumeration defined by attr is set to the value defined by value. If the specified function does not exist, then cudaErrorInvalidDeviceFunction is returned. If the specified attribute cannot be written, or if the value is incorrect, then cudaErrorInvalidValue is returned.

Valid values for attr are:

• cudaFuncAttributeMaxDynamicSharedMemorySize - The requested maximum size in bytes of dynamically-allocated shared memory. The sum of this value and the function attribute sharedSizeBytes cannot exceed the device attribute cudaDevAttrMaxSharedMemoryPerBlockOptin. The maximal size of requestable dynamic shared memory may differ by GPU architecture.

• cudaFuncAttributePreferredSharedMemoryCarveout - 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 total shared memory. See cudaDevAttrMaxSharedMemoryPerMultiprocessor. This is only a hint, and the driver can choose a different ratio if required to execute the function.

• cudaFuncAttributeRequiredClusterWidth: The required cluster width in blocks. The width, height, and depth values must either all be 0 or all be positive. The validity of the cluster dimensions is checked at launch time. If the value is set during compile time, it cannot be set at runtime. Setting it at runtime will return cudaErrorNotPermitted.

• cudaFuncAttributeRequiredClusterHeight: The required cluster height in blocks. The width, height, and depth values must either all be 0 or all be positive. The validity of the cluster dimensions is checked at launch time. If the value is set during compile time, it cannot be set at runtime. Setting it at runtime will return cudaErrorNotPermitted.

• cudaFuncAttributeRequiredClusterDepth: The required cluster depth in blocks. The width, height, and depth values must either all be 0 or all be positive. The validity of the cluster dimensions is checked at launch time. If the value is set during compile time, it cannot be set at runtime. Setting it at runtime will return cudaErrorNotPermitted.

• cudaFuncAttributeNonPortableClusterSizeAllowed: Indicates whether the function can be launched with non-portable cluster size. 1 is allowed, 0 is disallowed.

• cudaFuncAttributeClusterSchedulingPolicyPreference: The block scheduling policy of a function. The value type is cudaClusterSchedulingPolicy.

• Note that this function may also return error codes from previous, asynchronous launches.

• Note that this function may also return cudaErrorInitializationError, cudaErrorInsufficientDriver or cudaErrorNoDevice if this call tries to initialize internal CUDA RT state.

• Note that as specified by cudaStreamAddCallback no CUDA function may be called from callback. cudaErrorNotPermitted may, but is not guaranteed to, be returned as a diagnostic in such case.

• The API can also be used with a kernel cudaKernel_t by querying the handle using cudaLibraryGetKernel() or cudaGetKernel and then passing it to the API by casting to void*. The symbol entryFuncAddr passed to cudaGetKernel should be a symbol that is registered with the same CUDA Runtime instance.

• Passing a symbol that belongs that belongs to a different runtime instance will result in undefined behavior. The only type that can be reliably passed to a different runtime instance is cudaKernel_t

cudaLaunchKernel ( C++ API), cudaFuncSetCacheConfig ( C++ API), cudaFuncGetAttributes ( C API), cudaSetDoubleForDevice, cudaSetDoubleForHost

[C++ API] Sets the preferred cache configuration for a device function

func

- device function pointer

cacheConfig

- Requested cache configuration

On devices where the L1 cache and shared memory use the same hardware resources, this sets through cacheConfig the preferred cache configuration for the function specified via func. This is only a preference. The runtime will use the requested configuration if possible, but it is free to choose a different configuration if required to execute func.

func must be a pointer to a function that executes on the device. The parameter specified by func must be declared as a __global__ function. If the specified function does not exist, then cudaErrorInvalidDeviceFunction is returned.

This setting does nothing on devices where the size of the L1 cache and shared memory are fixed.

Launching a kernel with a different preference than the most recent preference setting may insert a device-side synchronization point.

The supported cache configurations are:

• cudaFuncCachePreferNone: no preference for shared memory or L1 (default)

• cudaFuncCachePreferShared: prefer larger shared memory and smaller L1 cache

• cudaFuncCachePreferL1: prefer larger L1 cache and smaller shared memory

cudaLaunchKernel ( C++ API), cudaFuncSetCacheConfig ( C API), cudaFuncGetAttributes ( C++ API), cudaSetDoubleForDevice, cudaSetDoubleForHost, cudaThreadGetCacheConfig, cudaThreadSetCacheConfig

Get pointer to device kernel that matches entry function entryFuncAddr.

kernelPtr

- Returns the device kernel

func

Returns in kernelPtr the device kernel corresponding to the entry function entryFuncAddr.

See also:

cudaGetKernel ( C API)

[C++ API] Finds the address associated with a CUDA symbol

devPtr

- Return device pointer associated with symbol

symbol

- Device symbol reference

[C++ API] Finds the size of the object associated with a CUDA symbol

size

- Size of object associated with symbol

symbol

- Device symbol reference

Creates a memcpy node to copy from a symbol on the device and adds it to a graph.

pGraphNode

- Returns newly created node

graph

- Graph to which to add the node

pDependencies

- Dependencies of the node

numDependencies

- Number of dependencies

dst

- Destination memory address

symbol

- Device symbol address

count

- Size in bytes to copy

offset

- Offset from start of symbol in bytes

kind

- Type of transfer

Creates a new memcpy node to copy from symbol and adds it to graph with numDependencies dependencies specified via pDependencies. It is possible for numDependencies to be 0, in which case the node will be placed at the root of the graph. pDependencies may not have any duplicate entries. A handle to the new node will be returned in pGraphNode.

When the graph is launched, the node will copy count bytes from the memory area pointed to by offset bytes from the start of symbol symbol to the memory area pointed to by dst. The memory areas may not overlap. symbol is a variable that resides in global or constant memory space. kind can be either cudaMemcpyDeviceToHost, cudaMemcpyDeviceToDevice, or cudaMemcpyDefault. Passing cudaMemcpyDefault is recommended, in which case the type of transfer is inferred from the pointer values. However, cudaMemcpyDefault is only allowed on systems that support unified virtual addressing.

Memcpy nodes have some additional restrictions with regards to managed memory, if the system contains at least one device which has a zero value for the device attribute cudaDevAttrConcurrentManagedAccess.

See also:

cudaMemcpyFromSymbol, cudaGraphAddMemcpyNode, cudaGraphAddMemcpyNodeToSymbol, cudaGraphMemcpyNodeGetParams, cudaGraphMemcpyNodeSetParams, cudaGraphMemcpyNodeSetParamsFromSymbol, cudaGraphMemcpyNodeSetParamsToSymbol, cudaGraphCreate, cudaGraphDestroyNode, cudaGraphAddChildGraphNode, cudaGraphAddEmptyNode, cudaGraphAddKernelNode, cudaGraphAddHostNode, cudaGraphAddMemsetNode

Creates a memcpy node to copy to a symbol on the device and adds it to a graph.

pGraphNode

- Returns newly created node

graph

- Graph to which to add the node

pDependencies

- Dependencies of the node

numDependencies

- Number of dependencies

symbol

- Device symbol address

src

- Source memory address

count

- Size in bytes to copy

offset

- Offset from start of symbol in bytes

kind

- Type of transfer

Creates a new memcpy node to copy to symbol and adds it to graph with numDependencies dependencies specified via pDependencies. It is possible for numDependencies to be 0, in which case the node will be placed at the root of the graph. pDependencies may not have any duplicate entries. A handle to the new node will be returned in pGraphNode.

When the graph is launched, the node will copy count bytes from the memory area pointed to by src to the memory area pointed to by offset bytes from the start of symbol symbol. The memory areas may not overlap. symbol is a variable that resides in global or constant memory space. kind can be either cudaMemcpyHostToDevice, cudaMemcpyDeviceToDevice, or cudaMemcpyDefault. Passing cudaMemcpyDefault is recommended, in which case the type of transfer is inferred from the pointer values. However, cudaMemcpyDefault is only allowed on systems that support unified virtual addressing.

Memcpy nodes have some additional restrictions with regards to managed memory, if the system contains at least one device which has a zero value for the device attribute cudaDevAttrConcurrentManagedAccess.

See also:

cudaMemcpyToSymbol, cudaGraphAddMemcpyNode, cudaGraphAddMemcpyNodeFromSymbol, cudaGraphMemcpyNodeGetParams, cudaGraphMemcpyNodeSetParams, cudaGraphMemcpyNodeSetParamsToSymbol, cudaGraphMemcpyNodeSetParamsFromSymbol, cudaGraphCreate, cudaGraphDestroyNode, cudaGraphAddChildGraphNode, cudaGraphAddEmptyNode, cudaGraphAddKernelNode, cudaGraphAddHostNode, cudaGraphAddMemsetNode

Sets the parameters for a memcpy node in the given graphExec to copy from a symbol on the device.

hGraphExec

- The executable graph in which to set the specified node

node

- Memcpy node from the graph which was used to instantiate graphExec

dst

- Destination memory address

symbol

- Device symbol address

count

- Size in bytes to copy

offset

- Offset from start of symbol in bytes

kind

- Type of transfer

Sets the parameters for a memcpy node in the given graphExec to copy to a symbol on the device.

hGraphExec

- The executable graph in which to set the specified node

node

- Memcpy node from the graph which was used to instantiate graphExec

symbol

- Device symbol address

src

- Source memory address

count

- Size in bytes to copy

offset

- Offset from start of symbol in bytes

kind

- Type of transfer

Creates an executable graph from a graph.

pGraphExec

- Returns instantiated graph

graph

- Graph to instantiate

pErrorNode

- In case of an instantiation error, this may be modified to indicate a node contributing to the error

pLogBuffer

- A character buffer to store diagnostic messages

bufferSize

- Size of the log buffer in bytes

Instantiates graph as an executable graph. The graph is validated for any structural constraints or intra-node constraints which were not previously validated. If instantiation is successful, a handle to the instantiated graph is returned in pGraphExec.

If there are any errors, diagnostic information may be returned in pErrorNode and pLogBuffer. This is the primary way to inspect instantiation errors. The output will be null terminated unless the diagnostics overflow the buffer. In this case, they will be truncated, and the last byte can be inspected to determine if truncation occurred.

See also:

cudaGraphInstantiateWithFlags, cudaGraphCreate, cudaGraphUpload, cudaGraphLaunch, cudaGraphExecDestroy

Sets a memcpy node's parameters to copy from a symbol on the device.

node

- Node to set the parameters for

dst

- Destination memory address

symbol

- Device symbol address

count

- Size in bytes to copy

offset

- Offset from start of symbol in bytes

kind

- Type of transfer

Sets a memcpy node's parameters to copy to a symbol on the device.

node

- Node to set the parameters for

symbol

- Device symbol address

src

- Source memory address

count

- Size in bytes to copy

offset

- Offset from start of symbol in bytes

kind

- Type of transfer

Launches a device function.

func

- Device function symbol

gridDim

- Grid dimentions

blockDim

- Block dimentions

args

- Arguments

sharedMem

- Shared memory (defaults to 0)

stream

- Stream identifier (defaults to NULL)

The function invokes kernel func on gridDim (gridDim.xgridDim.ygridDim.z) grid of blocks. Each block contains blockDim (blockDim.xblockDim.yblockDim.z) threads.

The device on which this kernel is invoked must have a non-zero value for the device attribute cudaDevAttrCooperativeLaunch.

The total number of blocks launched cannot exceed the maximum number of blocks per multiprocessor as returned by cudaOccupancyMaxActiveBlocksPerMultiprocessor (or cudaOccupancyMaxActiveBlocksPerMultiprocessorWithFlags) times the number of multiprocessors as specified by the device attribute cudaDevAttrMultiProcessorCount.

The kernel cannot make use of CUDA dynamic parallelism.

If the kernel has N parameters the args should point to array of N pointers. Each pointer, from args[0] to args[N - 1], point to the region of memory from which the actual parameter will be copied.

sharedMem sets the amount of dynamic shared memory that will be available to each thread block.

stream specifies a stream the invocation is associated to.

• Note that this function may also return error codes from previous, asynchronous launches.

• This function exhibits asynchronous behavior for most use cases.

• This function uses standard default stream semantics.

• Note that this function may also return cudaErrorInitializationError, cudaErrorInsufficientDriver or cudaErrorNoDevice if this call tries to initialize internal CUDA RT state.

• Note that as specified by cudaStreamAddCallback no CUDA function may be called from callback. cudaErrorNotPermitted may, but is not guaranteed to, be returned as a diagnostic in such case.

• The API can also be used with a kernel cudaKernel_t by querying the handle using cudaLibraryGetKernel() or cudaGetKernel and then passing it to the API by casting to void*. The symbol entryFuncAddr passed to cudaGetKernel should be a symbol that is registered with the same CUDA Runtime instance.

• Passing a symbol that belongs that belongs to a different runtime instance will result in undefined behavior. The only type that can be reliably passed to a different runtime instance is cudaKernel_t

cudaLaunchCooperativeKernel ( C API)

Launches a device function.

func

- Device function symbol

gridDim

- Grid dimentions

blockDim

- Block dimentions

args

- Arguments

sharedMem

- Shared memory (defaults to 0)

stream

- Stream identifier (defaults to NULL)

cudaSuccess, cudaErrorInvalidDeviceFunction, cudaErrorInvalidConfiguration, cudaErrorLaunchFailure, cudaErrorLaunchTimeout, cudaErrorLaunchOutOfResources, cudaErrorSharedObjectInitFailed, cudaErrorInvalidPtx, cudaErrorUnsupportedPtxVersion, cudaErrorNoKernelImageForDevice, cudaErrorJitCompilerNotFound, cudaErrorJitCompilationDisabled

The function invokes kernel func on gridDim (gridDim.xgridDim.ygridDim.z) grid of blocks. Each block contains blockDim (blockDim.xblockDim.yblockDim.z) threads.

If the kernel has N parameters the args should point to array of N pointers. Each pointer, from args[0] to args[N - 1], point to the region of memory from which the actual parameter will be copied.

sharedMem sets the amount of dynamic shared memory that will be available to each thread block.

stream specifies a stream the invocation is associated to.

• Note that this function may also return error codes from previous, asynchronous launches.

• This function exhibits asynchronous behavior for most use cases.

• This function uses standard default stream semantics.

• Note that this function may also return cudaErrorInitializationError, cudaErrorInsufficientDriver or cudaErrorNoDevice if this call tries to initialize internal CUDA RT state.

• Note that as specified by cudaStreamAddCallback no CUDA function may be called from callback. cudaErrorNotPermitted may, but is not guaranteed to, be returned as a diagnostic in such case.

• The API can also be used with a kernel cudaKernel_t by querying the handle using cudaLibraryGetKernel() or cudaGetKernel and then passing it to the API by casting to void*. The symbol entryFuncAddr passed to cudaGetKernel should be a symbol that is registered with the same CUDA Runtime instance.

• Passing a symbol that belongs that belongs to a different runtime instance will result in undefined behavior. The only type that can be reliably passed to a different runtime instance is cudaKernel_t

cudaLaunchKernel ( C API)

Launches a CUDA function with launch-time configuration.

config

- Launch configuration

kernel

args

- Parameter pack of kernel parameters

cudaSuccess, cudaErrorInvalidDeviceFunction, cudaErrorInvalidConfiguration, cudaErrorLaunchFailure, cudaErrorLaunchTimeout, cudaErrorLaunchOutOfResources, cudaErrorSharedObjectInitFailed, cudaErrorInvalidPtx, cudaErrorUnsupportedPtxVersion, cudaErrorNoKernelImageForDevice, cudaErrorJitCompilerNotFound, cudaErrorJitCompilationDisabled

Invokes the kernel kernel on config->gridDim (config->gridDim.xconfig->gridDim.yconfig->gridDim.z) grid of blocks. Each block contains config->blockDim (config->blockDim.xconfig->blockDim.yconfig->blockDim.z) threads.

config->dynamicSmemBytes sets the amount of dynamic shared memory that will be available to each thread block.

config->stream specifies a stream the invocation is associated to.

Configuration beyond grid and block dimensions, dynamic shared memory size, and stream can be provided with the following two fields of config:

config->attrs is an array of config->numAttrs contiguous cudaLaunchAttribute elements. The value of this pointer is not considered if config->numAttrs is zero. However, in that case, it is recommended to set the pointer to NULL. config->numAttrs is the number of attributes populating the first config->numAttrs positions of the config->attrs array.

The kernel arguments should be passed as arguments to this function via the args parameter pack.

The C API version of this function, cudaLaunchKernelExC, is also available for pre-C++11 compilers and for use cases where the ability to pass kernel parameters via void* array is preferable.

See also:

cudaLaunchKernelEx ( C API), cuLaunchKernelEx

Launches a CUDA function with launch-time configuration.

config

- Launch configuration

kernel

- Kernel to launch

args

- Parameter pack of kernel parameters

cudaSuccess, cudaErrorInvalidDeviceFunction, cudaErrorInvalidConfiguration, cudaErrorLaunchFailure, cudaErrorLaunchTimeout, cudaErrorLaunchOutOfResources, cudaErrorSharedObjectInitFailed, cudaErrorInvalidPtx, cudaErrorUnsupportedPtxVersion, cudaErrorNoKernelImageForDevice, cudaErrorJitCompilerNotFound, cudaErrorJitCompilationDisabled

Invokes the kernel kernel on config->gridDim (config->gridDim.xconfig->gridDim.yconfig->gridDim.z) grid of blocks. Each block contains config->blockDim (config->blockDim.xconfig->blockDim.yconfig->blockDim.z) threads.

config->dynamicSmemBytes sets the amount of dynamic shared memory that will be available to each thread block.

config->stream specifies a stream the invocation is associated to.

Configuration beyond grid and block dimensions, dynamic shared memory size, and stream can be provided with the following two fields of config:

config->attrs is an array of config->numAttrs contiguous cudaLaunchAttribute elements. The value of this pointer is not considered if config->numAttrs is zero. However, in that case, it is recommended to set the pointer to NULL. config->numAttrs is the number of attributes populating the first config->numAttrs positions of the config->attrs array.

The kernel arguments should be passed as arguments to this function via the args parameter pack.

The C API version of this function, cudaLaunchKernelExC, is also available for pre-C++11 compilers and for use cases where the ability to pass kernel parameters via void* array is preferable.

• This function uses standard default stream semantics.

• Note that this function may also return error codes from previous, asynchronous launches.

• Note that this function may also return cudaErrorInitializationError, cudaErrorInsufficientDriver or cudaErrorNoDevice if this call tries to initialize internal CUDA RT state.

• Note that as specified by cudaStreamAddCallback no CUDA function may be called from callback. cudaErrorNotPermitted may, but is not guaranteed to, be returned as a diagnostic in such case.

• The API can also be used with a kernel cudaKernel_t by querying the handle using cudaLibraryGetKernel() or cudaGetKernel and then passing it to the API by casting to void*. The symbol entryFuncAddr passed to cudaGetKernel should be a symbol that is registered with the same CUDA Runtime instance.

• Passing a symbol that belongs that belongs to a different runtime instance will result in undefined behavior. The only type that can be reliably passed to a different runtime instance is cudaKernel_t

See also:

cudaLaunchKernelEx ( C API), cuLaunchKernelEx

Returns a global device pointer.

dptr

- Returned global device pointer for the requested library

bytes

- Returned global size in bytes

library

- Library to retrieve global from

name

- Name of global to retrieve

Returns a pointer to managed memory.

dptr

- Returned pointer to the managed memory

bytes

- Returned memory size in bytes

library

- Library to retrieve managed memory from

name

- Name of managed memory to retrieve

Returns in *dptr and *bytes the base pointer and size of the managed memory with name name for the requested library library. If no managed memory with the requested name name exists, the call returns cudaErrorSymbolNotFound. One of the parameters dptr or bytes (not both) can be NULL in which case it is ignored. Note that managed memory for library library is shared across devices and is registered when the library is loaded.

See also:

cudaLibraryLoadData, cudaLibraryLoadFromFile, cudaLibraryUnload, cudaLibraryGetGlobal

Returns a pointer to a unified function.

fptr

- Returned pointer to a unified function

library

- Library to retrieve function pointer memory from

symbol

- Name of function pointer to retrieve

Allocate from a pool.

[C++ API] Allocates page-locked memory on the host

ptr

- Device pointer to allocated memory

size

- Requested allocation size in bytes

flags

- Requested properties of allocated memory

Allocates size bytes of host memory that is page-locked and accessible to the device. The driver tracks the virtual memory ranges allocated with this function and automatically accelerates calls to functions such as cudaMemcpy(). Since the memory can be accessed directly by the device, it can be read or written with much higher bandwidth than pageable memory obtained with functions such as malloc(). Allocating excessive amounts of pinned memory may degrade system performance, since it reduces the amount of memory available to the system for paging. As a result, this function is best used sparingly to allocate staging areas for data exchange between host and device.

The flags parameter enables different options to be specified that affect the allocation, as follows.

• cudaHostAllocDefault: This flag's value is defined to be 0.

• cudaHostAllocPortable: The memory returned by this call will be considered as pinned memory by all CUDA contexts, not just the one that performed the allocation.

• cudaHostAllocMapped: Maps the allocation into the CUDA address space. The device pointer to the memory may be obtained by calling cudaHostGetDevicePointer().

• cudaHostAllocWriteCombined: Allocates the memory as write-combined (WC). WC memory can be transferred across the PCI Express bus more quickly on some system configurations, but cannot be read efficiently by most CPUs. WC memory is a good option for buffers that will be written by the CPU and read by the device via mapped pinned memory or host->device transfers.

All of these flags are orthogonal to one another: a developer may allocate memory that is portable, mapped and/or write-combined with no restrictions.

cudaSetDeviceFlags() must have been called with the cudaDeviceMapHost flag in order for the cudaHostAllocMapped flag to have any effect.

The cudaHostAllocMapped flag may be specified on CUDA contexts for devices that do not support mapped pinned memory. The failure is deferred to cudaHostGetDevicePointer() because the memory may be mapped into other CUDA contexts via the cudaHostAllocPortable flag.

Memory allocated by this function must be freed with cudaFreeHost().

See also:

cudaSetDeviceFlags, cudaMallocHost ( C API), cudaFreeHost, cudaHostAlloc

Allocates memory that will be automatically managed by the Unified Memory system.

Allocates size bytes of managed memory on the device and returns in *devPtr a pointer to the allocated memory. If the device doesn't support allocating managed memory, cudaErrorNotSupported is returned. Support for managed memory can be queried using the device attribute cudaDevAttrManagedMemory. The allocated memory is suitably aligned for any kind of variable. The memory is not cleared. If size is 0, cudaMallocManaged returns cudaErrorInvalidValue. The pointer is valid on the CPU and on all GPUs in the system that support managed memory. All accesses to this pointer must obey the Unified Memory programming model.

flags specifies the default stream association for this allocation. flags must be one of cudaMemAttachGlobal or cudaMemAttachHost. The default value for flags is cudaMemAttachGlobal. If cudaMemAttachGlobal is specified, then this memory is accessible from any stream on any device. If cudaMemAttachHost is specified, then the allocation should not be accessed from devices that have a zero value for the device attribute cudaDevAttrConcurrentManagedAccess; an explicit call to cudaStreamAttachMemAsync will be required to enable access on such devices.

If the association is later changed via cudaStreamAttachMemAsync to a single stream, the default association, as specifed during cudaMallocManaged, is restored when that stream is destroyed. For __managed__ variables, the default association is always cudaMemAttachGlobal. Note that destroying a stream is an asynchronous operation, and as a result, the change to default association won't happen until all work in the stream has completed.

Memory allocated with cudaMallocManaged should be released with cudaFree.

Device memory oversubscription is possible for GPUs that have a non-zero value for the device attribute cudaDevAttrConcurrentManagedAccess. Managed memory on such GPUs may be evicted from device memory to host memory at any time by the Unified Memory driver in order to make room for other allocations.

In a multi-GPU system where all GPUs have a non-zero value for the device attribute cudaDevAttrConcurrentManagedAccess, managed memory may not be populated when this API returns and instead may be populated on access. In such systems, managed memory can migrate to any processor's memory at any time. The Unified Memory driver will employ heuristics to maintain data locality and prevent excessive page faults to the extent possible. The application can also guide the driver about memory usage patterns via cudaMemAdvise. The application can also explicitly migrate memory to a desired processor's memory via cudaMemPrefetchAsync.

In a multi-GPU system where all of the GPUs have a zero value for the device attribute cudaDevAttrConcurrentManagedAccess and all the GPUs have peer-to-peer support with each other, the physical storage for managed memory is created on the GPU which is active at the time cudaMallocManaged is called. All other GPUs will reference the data at reduced bandwidth via peer mappings over the PCIe bus. The Unified Memory driver does not migrate memory among such GPUs.

In a multi-GPU system where not all GPUs have peer-to-peer support with each other and where the value of the device attribute cudaDevAttrConcurrentManagedAccess is zero for at least one of those GPUs, the location chosen for physical storage of managed memory is system-dependent.

• On Linux, the location chosen will be device memory as long as the current set of active contexts are on devices that either have peer-to-peer support with each other or have a non-zero value for the device attribute cudaDevAttrConcurrentManagedAccess. If there is an active context on a GPU that does not have a non-zero value for that device attribute and it does not have peer-to-peer support with the other devices that have active contexts on them, then the location for physical storage will be 'zero-copy' or host memory. Note that this means that managed memory that is located in device memory is migrated to host memory if a new context is created on a GPU that doesn't have a non-zero value for the device attribute and does not support peer-to-peer with at least one of the other devices that has an active context. This in turn implies that context creation may fail if there is insufficient host memory to migrate all managed allocations.

• On Windows, the physical storage is always created in 'zero-copy' or host memory. All GPUs will reference the data at reduced bandwidth over the PCIe bus. In these circumstances, use of the environment variable CUDA_VISIBLE_DEVICES is recommended to restrict CUDA to only use those GPUs that have peer-to-peer support. Alternatively, users can also set CUDA_MANAGED_FORCE_DEVICE_ALLOC to a non-zero value to force the driver to always use device memory for physical storage. When this environment variable is set to a non-zero value, all devices used in that process that support managed memory have to be peer-to-peer compatible with each other. The error cudaErrorInvalidDevice will be returned if a device that supports managed memory is used and it is not peer-to-peer compatible with any of the other managed memory supporting devices that were previously used in that process, even if cudaDeviceReset has been called on those devices. These environment variables are described in the CUDA programming guide under the "CUDA environment variables" section.

• On ARM, managed memory is not available on discrete gpu with Drive PX-2.

See also:

cudaMallocPitch, cudaFree, cudaMallocArray, cudaFreeArray, cudaMalloc3D, cudaMalloc3DArray, cudaMallocHost ( C API), cudaFreeHost, cudaHostAlloc, cudaDeviceGetAttribute, cudaStreamAttachMemAsync

Performs a batch of memory discard and prefetches asynchronously.

This is an alternate spelling for cudaMemDiscardAndPrefetchBatchAsync made available through function overloading.

The cudaMemLocation specified by prefetchLocs are applicable for all the operations in the batch.

See also:

cudaMemDiscardAndPrefetchBatchAsync

Performs a batch of memory discard and prefetches asynchronously.

Performs a batch of memory prefetches asynchronously.

This is an alternate spelling for cudaMemPrefetchBatchAsync made available through function overloading.

The cudaMemLocation specified by prefetchLocs are applicable for all the prefetches specified in the batch.

See also:

cudaMemPrefetchBatchAsync

Performs a batch of memory prefetches asynchronously.

This is an alternate spelling for cudaMemPrefetchBatchAsync made available through function overloading.

See also:

cudaMemPrefetchBatchAsync

Performs a batch of memory copies asynchronously.

This is an alternate spelling for cudaMemcpyBatchAsync made available through function overloading.

The cudaMemcpyAttributes specified by attr are applicable for all the copies specified in the batch.

See also:

cudaMemcpyBatchAsync

Performs a batch of memory copies asynchronously.

This is an alternate spelling for cudaMemcpyBatchAsync made available through function overloading.

See also:

cudaMemcpyBatchAsync

[C++ API] Copies data from the given symbol on the device

dst

- Destination memory address

symbol

- Device symbol reference

count

- Size in bytes to copy

offset

- Offset from start of symbol in bytes

kind

- Type of transfer

Copies count bytes from the memory area offset bytes from the start of symbol symbol to the memory area pointed to by dst. The memory areas may not overlap. symbol is a variable that resides in global or constant memory space. kind can be either cudaMemcpyDeviceToHost or cudaMemcpyDeviceToDevice.

• Note that this function may also return error codes from previous, asynchronous launches.

• This function exhibits synchronous behavior for most use cases.

• Use of a string naming a variable as the symbol parameter was deprecated in CUDA 4.1 and removed in CUDA 5.0.

• Note that this function may also return cudaErrorInitializationError, cudaErrorInsufficientDriver or cudaErrorNoDevice if this call tries to initialize internal CUDA RT state.

• Note that as specified by cudaStreamAddCallback no CUDA function may be called from callback. cudaErrorNotPermitted may, but is not guaranteed to, be returned as a diagnostic in such case.

See also:

cudaMemcpy, cudaMemcpy2D, cudaMemcpy2DToArray, cudaMemcpy2DFromArray, cudaMemcpy2DArrayToArray, cudaMemcpyToSymbol, cudaMemcpyAsync, cudaMemcpy2DAsync, cudaMemcpy2DToArrayAsync, cudaMemcpy2DFromArrayAsync, cudaMemcpyToSymbolAsync, cudaMemcpyFromSymbolAsync

[C++ API] Copies data from the given symbol on the device

dst

- Destination memory address

symbol

- Device symbol reference

count

- Size in bytes to copy

offset

- Offset from start of symbol in bytes

kind

- Type of transfer

stream

- Stream identifier

Copies count bytes from the memory area offset bytes from the start of symbol symbol to the memory area pointed to by dst. The memory areas may not overlap. symbol is a variable that resides in global or constant memory space. kind can be either cudaMemcpyDeviceToHost or cudaMemcpyDeviceToDevice.

cudaMemcpyFromSymbolAsync() is asynchronous with respect to the host, so the call may return before the copy is complete. The copy can optionally be associated to a stream by passing a non-zero stream argument. If kind is cudaMemcpyDeviceToHost and stream is non-zero, the copy may overlap with operations in other streams.

• Note that this function may also return error codes from previous, asynchronous launches.

• This function exhibits asynchronous behavior for most use cases.

• Use of a string naming a variable as the symbol parameter was deprecated in CUDA 4.1 and removed in CUDA 5.0.

• Note that this function may also return cudaErrorInitializationError, cudaErrorInsufficientDriver or cudaErrorNoDevice if this call tries to initialize internal CUDA RT state.

• Note that as specified by cudaStreamAddCallback no CUDA function may be called from callback. cudaErrorNotPermitted may, but is not guaranteed to, be returned as a diagnostic in such case.

See also:

cudaMemcpy, cudaMemcpy2D, cudaMemcpy2DToArray, cudaMemcpy2DFromArray, cudaMemcpy2DArrayToArray, cudaMemcpyToSymbol, cudaMemcpyFromSymbol, cudaMemcpyAsync, cudaMemcpy2DAsync, cudaMemcpy2DToArrayAsync, cudaMemcpy2DFromArrayAsync, cudaMemcpyToSymbolAsync

[C++ API] Copies data to the given symbol on the device

symbol

- Device symbol reference

src

- Source memory address

count

- Size in bytes to copy

offset

- Offset from start of symbol in bytes

kind

- Type of transfer

Copies count bytes from the memory area pointed to by src to the memory area offset bytes from the start of symbol symbol. The memory areas may not overlap. symbol is a variable that resides in global or constant memory space. kind can be either cudaMemcpyHostToDevice or cudaMemcpyDeviceToDevice.

• Note that this function may also return error codes from previous, asynchronous launches.

• This function exhibits synchronous behavior for most use cases.

• Use of a string naming a variable as the symbol parameter was deprecated in CUDA 4.1 and removed in CUDA 5.0.

• Note that this function may also return cudaErrorInitializationError, cudaErrorInsufficientDriver or cudaErrorNoDevice if this call tries to initialize internal CUDA RT state.

• Note that as specified by cudaStreamAddCallback no CUDA function may be called from callback. cudaErrorNotPermitted may, but is not guaranteed to, be returned as a diagnostic in such case.

See also:

cudaMemcpy, cudaMemcpy2D, cudaMemcpy2DToArray, cudaMemcpy2DFromArray, cudaMemcpy2DArrayToArray, cudaMemcpyFromSymbol, cudaMemcpyAsync, cudaMemcpy2DAsync, cudaMemcpy2DToArrayAsync, cudaMemcpy2DFromArrayAsync, cudaMemcpyToSymbolAsync, cudaMemcpyFromSymbolAsync

[C++ API] Copies data to the given symbol on the device

symbol

- Device symbol reference

src

- Source memory address

count

- Size in bytes to copy

offset

- Offset from start of symbol in bytes

kind

- Type of transfer

stream

- Stream identifier

Copies count bytes from the memory area pointed to by src to the memory area offset bytes from the start of symbol symbol. The memory areas may not overlap. symbol is a variable that resides in global or constant memory space. kind can be either cudaMemcpyHostToDevice or cudaMemcpyDeviceToDevice.

cudaMemcpyToSymbolAsync() is asynchronous with respect to the host, so the call may return before the copy is complete. The copy can optionally be associated to a stream by passing a non-zero stream argument. If kind is cudaMemcpyHostToDevice and stream is non-zero, the copy may overlap with operations in other streams.

• Note that this function may also return error codes from previous, asynchronous launches.

• This function exhibits asynchronous behavior for most use cases.

• Use of a string naming a variable as the symbol parameter was deprecated in CUDA 4.1 and removed in CUDA 5.0.

• Note that this function may also return cudaErrorInitializationError, cudaErrorInsufficientDriver or cudaErrorNoDevice if this call tries to initialize internal CUDA RT state.

• Note that as specified by cudaStreamAddCallback no CUDA function may be called from callback. cudaErrorNotPermitted may, but is not guaranteed to, be returned as a diagnostic in such case.

See also:

cudaMemcpy, cudaMemcpy2D, cudaMemcpy2DToArray, cudaMemcpy2DFromArray, cudaMemcpy2DArrayToArray, cudaMemcpyToSymbol, cudaMemcpyFromSymbol, cudaMemcpyAsync, cudaMemcpy2DAsync, cudaMemcpy2DToArrayAsync, cudaMemcpy2DFromArrayAsync, cudaMemcpyFromSymbolAsync

Returns dynamic shared memory available per block when launching numBlocks blocks on SM.

dynamicSmemSize

- Returned maximum dynamic shared memory

func

- Kernel function for which occupancy is calculated

numBlocks

- Number of blocks to fit on SM

blockSize

- Size of the block

Returns occupancy for a device function.

numBlocks

- Returned occupancy

func

- Kernel function for which occupancy is calulated

blockSize

- Block size the kernel is intended to be launched with

dynamicSMemSize

- Per-block dynamic shared memory usage intended, in bytes

Returns occupancy for a device function with the specified flags.

numBlocks

- Returned occupancy

func

- Kernel function for which occupancy is calulated

blockSize

- Block size the kernel is intended to be launched with

dynamicSMemSize

- Per-block dynamic shared memory usage intended, in bytes

flags

- Requested behavior for the occupancy calculator

Given the kernel function (func) and launch configuration (config), return the maximum number of clusters that could co-exist on the target device in *numClusters.

numClusters

- Returned maximum number of clusters that could co-exist on the target device

func

- Kernel function for which maximum number of clusters are calculated

config

- Launch configuration for the given kernel function

If the function has required cluster size already set (see cudaFuncGetAttributes), the cluster size from config must either be unspecified or match the required size. Without required sizes, the cluster size must be specified in config, else the function will return an error.

Note that various attributes of the kernel function may affect occupancy calculation. Runtime environment may affect how the hardware schedules the clusters, so the calculated occupancy is not guaranteed to be achievable.

• Note that this function may also return error codes from previous, asynchronous launches.

• Note that this function may also return cudaErrorInitializationError, cudaErrorInsufficientDriver or cudaErrorNoDevice if this call tries to initialize internal CUDA RT state.

• Note that as specified by cudaStreamAddCallback no CUDA function may be called from callback. cudaErrorNotPermitted may, but is not guaranteed to, be returned as a diagnostic in such case.

• The API can also be used with a kernel cudaKernel_t by querying the handle using cudaLibraryGetKernel() or cudaGetKernel and then passing it to the API by casting to void*. The symbol entryFuncAddr passed to cudaGetKernel should be a symbol that is registered with the same CUDA Runtime instance.

• Passing a symbol that belongs that belongs to a different runtime instance will result in undefined behavior. The only type that can be reliably passed to a different runtime instance is cudaKernel_t

See also:

cudaFuncGetAttributes

Returns grid and block size that achieves maximum potential occupancy for a device function.

minGridSize

- Returned minimum grid size needed to achieve the best potential occupancy

blockSize

- Returned block size

func

- Device function symbol

dynamicSMemSize

- Per-block dynamic shared memory usage intended, in bytes

blockSizeLimit

- The maximum block size func is designed to work with. 0 means no limit.

Returns grid and block size that achieves maximum potential occupancy for a device function.

minGridSize

- Returned minimum grid size needed to achieve the best potential occupancy

blockSize

- Returned block size

func

- Device function symbol

blockSizeToDynamicSMemSize

- A unary function / functor that takes block size, and returns the size, in bytes, of dynamic shared memory needed for a block

blockSizeLimit

- The maximum block size func is designed to work with. 0 means no limit.

Returns grid and block size that achieves maximum potential occupancy for a device function.

minGridSize

- Returned minimum grid size needed to achieve the best potential occupancy

blockSize

- Returned block size

func

- Device function symbol

blockSizeToDynamicSMemSize

- A unary function / functor that takes block size, and returns the size, in bytes, of dynamic shared memory needed for a block

blockSizeLimit

- The maximum block size func is designed to work with. 0 means no limit.

flags

- Requested behavior for the occupancy calculator

Returns grid and block size that achived maximum potential occupancy for a device function with the specified flags.

minGridSize

- Returned minimum grid size needed to achieve the best potential occupancy

blockSize

- Returned block size

func

- Device function symbol

dynamicSMemSize

- Per-block dynamic shared memory usage intended, in bytes

blockSizeLimit

- The maximum block size func is designed to work with. 0 means no limit.

flags

- Requested behavior for the occupancy calculator

Given the kernel function (func) and launch configuration (config), return the maximum cluster size in *clusterSize.

clusterSize

- Returned maximum cluster size that can be launched for the given kernel function and launch configuration

func

- Kernel function for which maximum cluster size is calculated

config

- Launch configuration for the given kernel function

The cluster dimensions in config are ignored. If func has a required cluster size set (see cudaFuncGetAttributes),*clusterSize will reflect the required cluster size.

By default this function will always return a value that's portable on future hardware. A higher value may be returned if the kernel function allows non-portable cluster sizes.

This function will respect the compile time launch bounds.

• Note that this function may also return error codes from previous, asynchronous launches.

• Note that this function may also return cudaErrorInitializationError, cudaErrorInsufficientDriver or cudaErrorNoDevice if this call tries to initialize internal CUDA RT state.

• Note that as specified by cudaStreamAddCallback no CUDA function may be called from callback. cudaErrorNotPermitted may, but is not guaranteed to, be returned as a diagnostic in such case.

• The API can also be used with a kernel cudaKernel_t by querying the handle using cudaLibraryGetKernel() or cudaGetKernel and then passing it to the API by casting to void*. The symbol entryFuncAddr passed to cudaGetKernel should be a symbol that is registered with the same CUDA Runtime instance.

• Passing a symbol that belongs that belongs to a different runtime instance will result in undefined behavior. The only type that can be reliably passed to a different runtime instance is cudaKernel_t

See also:

cudaFuncGetAttributes

Attach memory to a stream asynchronously.

Enqueues an operation in stream to specify stream association of length bytes of memory starting from devPtr. This function is a stream-ordered operation, meaning that it is dependent on, and will only take effect when, previous work in stream has completed. Any previous association is automatically replaced.

devPtr must point to an one of the following types of memories:

• managed memory declared using the __managed__ keyword or allocated with cudaMallocManaged.

• a valid host-accessible region of system-allocated pageable memory. This type of memory may only be specified if the device associated with the stream reports a non-zero value for the device attribute cudaDevAttrPageableMemoryAccess.

For managed allocations, length must be either zero or the entire allocation's size. Both indicate that the entire allocation's stream association is being changed. Currently, it is not possible to change stream association for a portion of a managed allocation.

For pageable allocations, length must be non-zero.

The stream association is specified using flags which must be one of cudaMemAttachGlobal, cudaMemAttachHost or cudaMemAttachSingle. The default value for flags is cudaMemAttachSingle If the cudaMemAttachGlobal flag is specified, the memory can be accessed by any stream on any device. If the cudaMemAttachHost flag is specified, the program makes a guarantee that it won't access the memory on the device from any stream on a device that has a zero value for the device attribute cudaDevAttrConcurrentManagedAccess. If the cudaMemAttachSingle flag is specified and stream is associated with a device that has a zero value for the device attribute cudaDevAttrConcurrentManagedAccess, the program makes a guarantee that it will only access the memory on the device from stream. It is illegal to attach singly to the NULL stream, because the NULL stream is a virtual global stream and not a specific stream. An error will be returned in this case.

When memory is associated with a single stream, the Unified Memory system will allow CPU access to this memory region so long as all operations in stream have completed, regardless of whether other streams are active. In effect, this constrains exclusive ownership of the managed memory region by an active GPU to per-stream activity instead of whole-GPU activity.

Accessing memory on the device from streams that are not associated with it will produce undefined results. No error checking is performed by the Unified Memory system to ensure that kernels launched into other streams do not access this region.

It is a program's responsibility to order calls to cudaStreamAttachMemAsync via events, synchronization or other means to ensure legal access to memory at all times. Data visibility and coherency will be changed appropriately for all kernels which follow a stream-association change.

If stream is destroyed while data is associated with it, the association is removed and the association reverts to the default visibility of the allocation as specified at cudaMallocManaged. For __managed__ variables, the default association is always cudaMemAttachGlobal. Note that destroying a stream is an asynchronous operation, and as a result, the change to default association won't happen until all work in the stream has completed.

See also:

cudaStreamCreate, cudaStreamCreateWithFlags, cudaStreamWaitEvent, cudaStreamSynchronize, cudaStreamAddCallback, cudaStreamDestroy, cudaMallocManaged