vkCreateCudaModuleNV - Creates a new CUDA module object

Description

Once a CUDA module has been created, the application may create the function entry point, which must refer to one function in the module.

Valid Usage (Implicit)

• device must be a valid Device handle

• pCreateInfo must be a valid pointer to a valid CudaModuleCreateInfoNV structure
• If pAllocator is not NULL, pAllocator must be a valid pointer to a valid AllocationCallbacks structure
• pModule must be a valid pointer to a CudaModuleNV handle
• The device must have been created with at least 1 queue

Return Codes

Success

• SUCCESS

Failure

• ERROR_INITIALIZATION_FAILED
• ERROR_OUT_OF_HOST_MEMORY
• ERROR_UNKNOWN
• ERROR_VALIDATION_FAILED

See Also

VK_NV_cuda_kernel_launch, AllocationCallbacks, CudaModuleCreateInfoNV, CudaModuleNV, Device

A convenience wrapper to make a compatible pair of calls to createCudaModuleNV and destroyCudaModuleNV

To ensure that destroyCudaModuleNV is always called: pass bracket (or the allocate function from your favourite resource management library) as the last argument. To just extract the pair pass (,) as the last argument.

vkGetCudaModuleCacheNV - Get CUDA module cache

Description

If pCacheData is NULL, then the size of the binary cache, in bytes, is returned in pCacheSize. Otherwise, pCacheSize must point to a variable set by the application to the size of the buffer, in bytes, pointed to by pCacheData, and on return the variable is overwritten with the amount of data actually written to pCacheData. If pCacheSize is less than the size of the binary shader code, nothing is written to pCacheData, and INCOMPLETE will be returned instead of SUCCESS.

The returned cache may then be used later for further initialization of the CUDA module, by sending this cache instead of the PTX code when using createCudaModuleNV.

Using the binary cache instead of the original PTX code should significantly speed up initialization of the CUDA module, given that the whole compilation and validation will not be necessary.

As with PipelineCache, the binary cache depends on the specific implementation. The application must assume the cache upload might fail in many circumstances and thus may have to get ready for falling back to the original PTX code if necessary. Most often, the cache may succeed if the same device driver and architecture is used between the cache generation from PTX and the use of this cache. In the event of a new driver version or if using a different device architecture, this cache may become invalid.

This query does not behave consistently with the behavior described in Opaque Binary Data Results, for historical reasons.

If the amount of data available is larger than the passed pDataSize, the query returns a INCOMPLETE success status instead of a ERROR_NOT_ENOUGH_SPACE_KHR error status.

Valid Usage (Implicit)

• device must be a valid Device handle

• module must be a valid CudaModuleNV handle
• pCacheSize must be a valid pointer to a size_t value
• If the value referenced by pCacheSize is not 0, and pCacheData is not NULL, pCacheData must be a valid pointer to an array of pCacheSize bytes
• module must have been created, allocated, or retrieved from device

Return Codes

Success

• INCOMPLETE
• SUCCESS

Failure

• ERROR_INITIALIZATION_FAILED
• ERROR_UNKNOWN
• ERROR_VALIDATION_FAILED

See Also

VK_NV_cuda_kernel_launch, CudaModuleNV, Device

vkCreateCudaFunctionNV - Creates a new CUDA function object

Valid Usage (Implicit)

• device must be a valid Device handle

• pCreateInfo must be a valid pointer to a valid CudaFunctionCreateInfoNV structure
• If pAllocator is not NULL, pAllocator must be a valid pointer to a valid AllocationCallbacks structure
• pFunction must be a valid pointer to a CudaFunctionNV handle
• The device must have been created with at least 1 queue

Return Codes

Success

• SUCCESS

Failure

• ERROR_INITIALIZATION_FAILED
• ERROR_OUT_OF_HOST_MEMORY
• ERROR_UNKNOWN
• ERROR_VALIDATION_FAILED

See Also

VK_NV_cuda_kernel_launch, AllocationCallbacks, CudaFunctionCreateInfoNV, CudaFunctionNV, Device

A convenience wrapper to make a compatible pair of calls to createCudaFunctionNV and destroyCudaFunctionNV

To ensure that destroyCudaFunctionNV is always called: pass bracket (or the allocate function from your favourite resource management library) as the last argument. To just extract the pair pass (,) as the last argument.

vkDestroyCudaModuleNV - Destroy a CUDA module

Valid Usage (Implicit)

• device must be a valid Device handle

• module must be a valid CudaModuleNV handle
• If pAllocator is not NULL, pAllocator must be a valid pointer to a valid AllocationCallbacks structure
• module must have been created, allocated, or retrieved from device

See Also

VK_NV_cuda_kernel_launch, AllocationCallbacks, CudaModuleNV, Device

vkDestroyCudaFunctionNV - Destroy a CUDA function

Valid Usage (Implicit)

• device must be a valid Device handle

• function must be a valid CudaFunctionNV handle
• If pAllocator is not NULL, pAllocator must be a valid pointer to a valid AllocationCallbacks structure
• function must have been created, allocated, or retrieved from device

See Also

VK_NV_cuda_kernel_launch, AllocationCallbacks, CudaFunctionNV, Device

vkCmdCudaLaunchKernelNV - Dispatch compute work items

Description

When the command is executed, a global workgroup consisting of gridDimX × gridDimY × gridDimZ local workgroups is assembled.

Valid Usage (Implicit)

• commandBuffer must be a valid CommandBuffer handle

• pLaunchInfo must be a valid pointer to a valid CudaLaunchInfoNV structure
• commandBuffer must be in the recording state
• The CommandPool that commandBuffer was allocated from must support QUEUE_COMPUTE_BIT, or QUEUE_GRAPHICS_BIT operations
• This command must not be called between suspended render pass instances
• This command must only be called outside of a video coding scope

Host Synchronization

• Host access to the CommandPool that commandBuffer was allocated from must be externally synchronized

Command Properties

'

Conditional Rendering

vkCmdCudaLaunchKernelNV is not affected by conditional rendering

See Also

VK_NV_cuda_kernel_launch, CommandBuffer, CudaLaunchInfoNV

VkCudaModuleCreateInfoNV - Structure specifying the parameters to create a CUDA Module

Valid Usage (Implicit)

See Also

VK_NV_cuda_kernel_launch, StructureType, createCudaModuleNV

VkCudaFunctionCreateInfoNV - Structure specifying the parameters to create a CUDA Function

Valid Usage (Implicit)

See Also

VK_NV_cuda_kernel_launch, CudaModuleNV, StructureType, createCudaFunctionNV

VkCudaLaunchInfoNV - Structure specifying the parameters to launch a CUDA kernel

Description

Kernel parameters of function are specified via pParams, very much the same way as described in cuLaunchKernel

If function has N parameters, then pParams must be an array of N pointers and paramCount must be N. Each of kernelParams[0] through kernelParams[N-1] must point to a region of memory from which the actual kernel parameter will be copied. The number of kernel parameters and their offsets and sizes are not specified here as that information is stored in the CudaFunctionNV object.

The application-owned memory pointed to by pParams and kernelParams[0] through kernelParams[N-1] are consumed immediately, and may be altered or freed after cmdCudaLaunchKernelNV has returned.

Valid Usage (Implicit)

See Also

VK_NV_cuda_kernel_launch, CudaFunctionNV, StructureType, cmdCudaLaunchKernelNV

VkPhysicalDeviceCudaKernelLaunchFeaturesNV - Structure describing whether cuda kernel launch is supported by the implementation

Members

This structure describes the following features:

Description

If the PhysicalDeviceCudaKernelLaunchFeaturesNV structure is included in the pNext chain of the PhysicalDeviceFeatures2 structure passed to getPhysicalDeviceFeatures2, it is filled in to indicate whether each corresponding feature is supported. If the application wishes to use a Device with any features described by PhysicalDeviceCudaKernelLaunchFeaturesNV, it must add an instance of the structure, with the desired feature members set to TRUE, to the pNext chain of DeviceCreateInfo when creating the Device.

Valid Usage (Implicit)

See Also

VK_NV_cuda_kernel_launch, Bool32, StructureType

VkPhysicalDeviceCudaKernelLaunchPropertiesNV - Structure describing the compute capability version available

Members

The members of the PhysicalDeviceCudaKernelLaunchPropertiesNV structure describe the following features:

Description

If the PhysicalDeviceCudaKernelLaunchPropertiesNV structure is included in the pNext chain of the PhysicalDeviceProperties2 structure passed to getPhysicalDeviceProperties2, it is filled in with each corresponding implementation-dependent property.

Valid Usage (Implicit)

See Also

VK_NV_cuda_kernel_launch, StructureType

VkCudaModuleNV - Opaque handle to a CUDA module object

See Also

VK_DEFINE_NON_DISPATCHABLE_HANDLE, VK_NV_cuda_kernel_launch, CudaFunctionCreateInfoNV, createCudaModuleNV, destroyCudaModuleNV, getCudaModuleCacheNV

VkCudaFunctionNV - Opaque handle to a CUDA function object

See Also

VK_DEFINE_NON_DISPATCHABLE_HANDLE, VK_NV_cuda_kernel_launch, CudaLaunchInfoNV, createCudaFunctionNV, destroyCudaFunctionNV

VkDebugReportObjectTypeEXT - Specify the type of an object handle

Description

'

DebugReportObjectTypeEXT and Vulkan Handle Relationship

The primary expected use of ERROR_VALIDATION_FAILED_EXT is for validation layer testing to prevent invalid commands from reaching the ICD. It is not expected that an application would see this error code during normal use of the validation layers. If an application returns TRUE in DebugUtilsMessengerCallbackDataEXT, the validation layers will return this error code instead of passing the command down the dispatch chain.

See Also

PFN_vkDebugReportCallbackEXT, VK_EXT_debug_marker, VK_EXT_debug_report, DebugMarkerObjectNameInfoEXT, DebugMarkerObjectTagInfoEXT, debugReportMessageEXT