Searched refs:CUDA (Results 1 – 25 of 39) sorted by relevance
12
6 == CUDA Modules10 === Creating a CUDA Module12 [open,refpage='VkCudaModuleNV',desc='Opaque handle to a CUDA module object',type='handles']14 CUDA modules must: contain some kernel code and must: expose at least one17 CUDA modules are represented by sname:VkCudaModuleNV handles:22 [open,refpage='vkCreateCudaModuleNV',desc='Creates a new CUDA module object',type='protos']24 To create a CUDA module, call:34 resulting CUDA module object is returned.36 Once a CUDA module has been created, the application may: create the42 [open,refpage='VkCudaModuleCreateInfoNV',desc='Structure specifying the parameters to create a CUDA…[all …]
6 == Dispatch Command for CUDA PTX Kernels14 For mode information, please refer to the CUDA toolkit documentation at17 Prior to using this command, you must: initialize a CUDA module, and create20 See <<cuda-modules, CUDA Modules>>.22 The dispatching of a CUDA kernel is recorded into a command buffer, and when30 To record a CUDA kernel launch, call:51 [open,refpage='VkCudaLaunchInfoNV',desc='Structure specifying the parameters to launch a CUDA kerne…54 the CUDA-Driver function58 Execution Control] of CUDA Driver API.67 * pname:function is the CUDA-Driver handle to the function being launched.[all …]
43 EGL and CUDA both have the capability to drive multiple devices,46 This extension defines a mapping from an EGL device to a CUDA device67 Add a new section 2.1.3 (CUDA Devices) after 2.1.2 (Devices)69 "Somewhat analogous to an EGL device, a CUDA device establishes a70 namespace for CUDA operations. In the CUDA API, such a device is71 represented by a C int. For more details, see the CUDA80 success, EGL_TRUE is returned, and a valid CUDA device handle82 is compatible with CUDA API functions."
46 This extension allows creating an EGL sync object linked to a CUDA56 A pointer to type cudaEvent_t, defined in the CUDA header files, may86 OpenCL object to which the sync object is linked. Likewise, a <CUDA87 event sync object> reflects the status of a corresponding CUDA89 results between EGL and the CUDA and OpenCL client APIs (see93 completion of the corresponding linked CUDA or OpenCL event object."98 "Section 3.8.1.X Creating CUDA Event Sync Objects100 If <type> is EGL_SYNC_CUDA_EVENT_NV, a CUDA event sync object is103 The object must be properly initialized and recorded by the CUDA API104 (using cudaCreateEvent and cudaEventRecord), and the CUDA device[all …]
51 This extension allows creating an EGL sync object linked to a CUDA61 A pointer to type cudaEvent_t, defined in the CUDA header files, may88 Likewise, a <CUDA event sync object> reflects the status of a89 corresponding CUDA object. Waiting on this type of sync object is91 CUDA event object.96 Section 3.8.1.X Creating and Signaling CUDA Event Sync Objects98 If <type> is EGL_SYNC_CUDA_EVENT_NV, a CUDA event sync object is108 must be properly initialized and recorded by the CUDA API (using109 cudaEventCreate and cudaEventRecord), and the CUDA device115 CUDA device used to create the CUDA event matches the[all …]
18 This extension targets deployment of existing CUDA kernels via Vulkan, with21 and CUDA contexts.22 However, we do encourage actual development using the native CUDA runtime25 The application will first have to create a CUDA module using26 flink:vkCreateCudaModuleNV then create the CUDA function entry point with34 the CUDA module handle with flink:vkDestroyCudaFunctionNV and46 initialization of the CUDA module.
72 HW_CONFIG_HWACCEL(1, 1, 0, CUDA, CUDA, ff_ ## codec ## _nvdec_hwaccel)
69 HW_CONFIG_ENCODER_FRAMES(CUDA, CUDA),70 HW_CONFIG_ENCODER_DEVICE(NONE, CUDA),
209 if (TM && TM->getDrvInterface() == NVPTX::CUDA) { in runOnKernelFunction()233 else if (TM && TM->getDrvInterface() == NVPTX::CUDA) in runOnKernelFunction()
58 if (TM.getDrvInterface() == NVPTX::CUDA) in hasImageHandles()
54 CUDA enumerator
148 if (TM.getDrvInterface() == NVPTX::CUDA) { in findIndexForHandle()
124 drvInterface = NVPTX::CUDA; in NVPTXTargetMachine()
294 NVPTX::CUDA) {} in NVPTXAsmPrinter()
966 if (static_cast<NVPTXTargetMachine &>(TM).getDrvInterface() == NVPTX::CUDA) { in emitLinkageDirective()1494 NVPTX::CUDA) { in emitFunctionParamList()
2 project(json_cuda LANGUAGES CUDA)
308 If EGL_NV_cuda_event is present, then for CUDA event sync objects, the310 as eglCreateSync. The current CUDA event handle is evaluated at the time312 the CUDA object with cudaEventRecord has no effect on the sync object314 unsignaled will cause the CUDA object to be reevaluated.
25 # ifdef __CUDACC__ // Wordaround for a CUDA compiler bug up to CUDA6
181 CUDA, // NVIDIA CUDA enumerator
82 | GPU CUDA 10.1 | Ubuntu-x86 | ✔️ |
85 | GPU CUDA 10.1 | Ubuntu-x86 | ✔️ |
187 case CUDA: return "cuda"; in getOSTypeName()509 .StartsWith("cuda", Triple::CUDA) in parseOS()
529 - add support for using clang to compile CUDA kernels641 - scale_cuda CUDA based video scale filter679 - CUDA thumbnail filter742 - CUDA CUVID H.263/VP8/VP9/10 bit HEVC (Dithered) Decoding789 - libnpp/CUDA-accelerated format conversion and scaling802 - CUDA CUVID H264/HEVC decoder