| /external/tensorflow/tensorflow/core/distributed_runtime/ |
| D | collective_param_resolver_distributed_test.cc | 138 void DefineWorkers(int num_workers, int num_devices, in DefineWorkers() argument
142 DefineWorker(name, device_type, num_devices, nccl); in DefineWorkers()
147 int num_devices, bool nccl) { in DefineWorker() argument
154 for (int i = 0; i < num_devices; ++i) { in DefineWorker()
178 void DefineCollectiveParams(int num_workers, int num_devices, in DefineCollectiveParams() argument
182 for (int di = 0; di < num_devices; ++di) { in DefineCollectiveParams()
186 CreateCollectiveParams(num_workers, num_devices, device_type); in DefineCollectiveParams()
191 CollectiveParams* CreateCollectiveParams(int num_workers, int num_devices, in CreateCollectiveParams() argument
197 cp->group.group_size = num_workers * num_devices; in CreateCollectiveParams()
208 void IssueRequests(int num_workers, int num_devices) { in IssueRequests() argument
[all …]
|
| /external/OpenCL-CTS/test_conformance/SVM/ |
| D | test_byte_granularity.cpp | 59 cl_uint num_devices = 0; in test_svm_byte_granularity() local
63 …iceID, &byte_manipulation_kernels[0], &context, &program, &queues[0], &num_devices, CL_DEVICE_SVM_… in test_svm_byte_granularity()
67 cl_uint num_devices_plus_host = num_devices + 1; in test_svm_byte_granularity()
77 cl_uint **error_counts = (cl_uint**) malloc(sizeof(void*) * num_devices); in test_svm_byte_granularity()
79 for(cl_uint i=0; i < num_devices; i++) { in test_svm_byte_granularity()
91 for(cl_uint d=0; d < num_devices; d++) // device ids starting at 1. in test_svm_byte_granularity()
99 for(cl_uint d=0; d < num_devices; d++) clFlush(queues[d]); in test_svm_byte_granularity()
101 cl_uint host_id = num_devices; // host code will take the id above the devices. in test_svm_byte_granularity()
102 for(int i = (int)num_devices; i < num_elements; i+= num_devices_plus_host) pA[i] = host_id; in test_svm_byte_granularity()
104 for(cl_uint id = 0; id < num_devices; id++) clFinish(queues[id]); in test_svm_byte_granularity()
[all …]
|
| D | test_fine_grain_memory_consistency.cpp | 68 …ntext, clCommandQueueWrapper* queues, clKernelWrapper &kernel, cl_uint num_devices, cl_uint numBin… in launch_kernels_and_verify() argument
72 …NE_GRAIN_BUFFER | CL_MEM_SVM_ATOMICS, sizeof(BinNode) * (num_pixels * (num_devices + 1) + numBins)… in launch_kernels_and_verify()
91 for(cl_uint d=0; d<num_devices; d++) in launch_kernels_and_verify()
96 for(cl_uint d=0; d<num_devices; d++) clFlush(queues[d]); in launch_kernels_and_verify()
103 for(cl_uint d=0; d<num_devices; d++) clFinish(queues[d]); in launch_kernels_and_verify()
127 if(num_items != num_pixels * (num_devices + 1) ) in launch_kernels_and_verify()
129 …s not correct, num items %d, expected num items: %d\n", num_items, num_pixels * (num_devices + 1)); in launch_kernels_and_verify()
147 cl_uint num_devices = 0; in test_svm_fine_grain_memory_consistency() local
158 …jects(deviceID, (const char**)source, &context, &program, &queues[0], &num_devices, CL_DEVICE_SVM_… in test_svm_fine_grain_memory_consistency()
168 result = launch_kernels_and_verify(context, queues, kernel, num_devices, numBins, num_pixels); in test_svm_fine_grain_memory_consistency()
[all …]
|
| D | test_shared_address_space_fine_grain.cpp | 27 cl_uint num_devices = 0; in test_svm_shared_address_space_fine_grain() local
31 …ed_list_create_and_verify_kernels[0], &context, &program, &queues[0], &num_devices, CL_DEVICE_SVM_… in test_svm_shared_address_space_fine_grain()
69 …for (int ci=0; ci<(int)num_devices+1; ci++) // ci is CreationIndex, index of device/q to create l… in test_svm_shared_address_space_fine_grain()
71 …for (int vi=0; vi<(int)num_devices+1; vi++) // vi is VerificationIndex, index of device/q to veri… in test_svm_shared_address_space_fine_grain()
73 if(ci == num_devices) // last device index represents the host, note the num_device+1 above. in test_svm_shared_address_space_fine_grain()
84 if(vi == num_devices) in test_svm_shared_address_space_fine_grain()
|
| D | main.cpp | 171 …text* context, cl_program *program, cl_command_queue *queues, cl_uint *num_devices, cl_device_svm_… in create_cl_objects() argument
180 error = clGetDeviceIDs(platform_id, CL_DEVICE_TYPE_ALL, 0, NULL, num_devices ); in create_cl_objects()
183 std::vector<cl_device_id> devicesTmp(*num_devices), devices, capable_devices; in create_cl_objects()
185 error = clGetDeviceIDs(platform_id, CL_DEVICE_TYPE_ALL, *num_devices, &devicesTmp[0], NULL ); in create_cl_objects()
199 for(cl_uint i = 0; i < *num_devices; i++) in create_cl_objects()
233 *num_devices = num_capable_devices; in create_cl_objects()
242 *context = clCreateContext(context_properties, *num_devices, &devices[0], NULL, NULL, &error); in create_cl_objects()
247 for(cl_uint i = 0; i < *num_devices; i++) in create_cl_objects()
|
| D | test_shared_address_space_fine_grain_buffers.cpp | 68 cl_uint num_devices = 0; in test_svm_shared_address_space_fine_grain_buffers() local
72 …ed_list_create_and_verify_kernels[0], &context, &program, &queues[0], &num_devices, CL_DEVICE_SVM_… in test_svm_shared_address_space_fine_grain_buffers()
105 …for (int ci=0; ci<(int)num_devices+1; ci++) // ci is CreationIndex, index of device/q to create l… in test_svm_shared_address_space_fine_grain_buffers()
107 …for (int vi=0; vi<(int)num_devices+1; vi++) // vi is VerificationIndex, index of device/q to veri… in test_svm_shared_address_space_fine_grain_buffers()
109 if(ci == num_devices) // last device index represents the host, note the num_device+1 above. in test_svm_shared_address_space_fine_grain_buffers()
120 if(vi == num_devices) in test_svm_shared_address_space_fine_grain_buffers()
|
| D | test_cross_buffer_pointers.cpp | 135 cl_uint num_devices = 0; in test_svm_cross_buffer_pointers_coarse_grain() local
139 …VMCrossBufferPointers_test_kernel[0], &context, &program, &queues[0], &num_devices, CL_DEVICE_SVM_… in test_svm_cross_buffer_pointers_coarse_grain()
186 …for (int ci=0; ci<(int)num_devices+1; ci++) // ci is CreationIndex, index of device/q to create l… in test_svm_cross_buffer_pointers_coarse_grain()
188 …for (int vi=0; vi<(int)num_devices+1; vi++) // vi is VerificationIndex, index of device/q to veri… in test_svm_cross_buffer_pointers_coarse_grain()
190 if(ci == num_devices) // last device index represents the host, note the num_device+1 above. in test_svm_cross_buffer_pointers_coarse_grain()
201 if(vi == num_devices) in test_svm_cross_buffer_pointers_coarse_grain()
|
| /external/tensorflow/tensorflow/python/distribute/v1/ |
| D | all_reduce.py | 317 num_devices = len(input_tensors)
318 if num_devices == 0:
320 if num_devices == 1:
325 num_chunks = num_devices * num_subchunks
326 num_ticks = num_devices - 1
330 for d in range(0, num_devices):
339 for d in range(0, num_devices):
343 seg_index = (rank + num_devices - (2 + tick)) % num_devices
350 for d in range(0, num_devices):
353 seg_index = (rank + num_devices - (2 + tick)) % num_devices
[all …]
|
| /external/OpenCL-CTS/test_conformance/buffers/ |
| D | test_buffer_migrate.cpp | 51 …ations migrate, cl_command_queue *queues, cl_mem *mem_objects, cl_uint num_devices, cl_mem_migrati… in migrateMemObject() argument
56 for (i=0; i<num_devices; i++) { in migrateMemObject()
57 j = genrand_int32(d) % num_devices; in migrateMemObject()
66 if ((j == i) && (num_devices > 1)) j = (j+1) % num_devices; in migrateMemObject()
81 static cl_int restoreBuffer(cl_command_queue *queues, cl_mem *buffers, cl_uint num_devices, cl_mem_… in restoreBuffer() argument
88 for (i=0; i<num_devices; i++) { in restoreBuffer()
105 cl_uint num_devices, num_devices_limited; in test_buffer_migrate() local
178 …CreateSubDevices(deviceID, (const cl_device_partition_property *)property, -1, NULL, &num_devices); in test_buffer_migrate()
179 if ((err != CL_SUCCESS) || (num_devices == 0)) { in test_buffer_migrate()
185 num_devices = 1; in test_buffer_migrate()
[all …]
|
| D | test_image_migrate.cpp | 62 cl_uint num_devices, cl_mem_migration_flags *flags, MTdata d) in migrateMemObject() argument
67 for (i=0; i<num_devices; i++) { in migrateMemObject()
68 j = genrand_int32(d) % num_devices; in migrateMemObject()
77 if ((j == i) && (num_devices > 1)) j = (j+1) % num_devices; in migrateMemObject()
93 static cl_int restoreImage(cl_command_queue *queues, cl_mem *mem_objects, cl_uint num_devices, in restoreImage() argument
104 for (i=0; i<num_devices; i++) { in restoreImage()
122 cl_uint num_devices, num_devices_limited; in test_image_migrate() local
211 …CreateSubDevices(deviceID, (const cl_device_partition_property *)property, -1, NULL, &num_devices); in test_image_migrate()
212 if ((err != CL_SUCCESS) || (num_devices == 0)) { in test_image_migrate()
218 num_devices = 1; in test_image_migrate()
[all …]
|
| /external/OpenCL-CTS/test_conformance/d3d10/ |
| D | main.cpp | 63 cl_uint num_devices = 0; in main() local
66 TestAdapterEnumeration(platform, pAdapter, pDevice, &num_devices); in main()
69 if (num_devices) in main()
71 TestAdapterDevices(platform, pAdapter, pDevice, num_devices); in main()
73 num_devices_tested += num_devices; in main()
90 …ation(cl_platform_id platform, IDXGIAdapter* pAdapter, ID3D10Device* pDevice, cl_uint* num_devices) in TestAdapterEnumeration() argument
202 *num_devices = num_device_devices; in TestAdapterEnumeration()
210 cl_uint num_devices = 0; in TestAdapterDevices() local
225 &num_devices); in TestAdapterDevices()
230 (num_devices == num_devices_expected), in TestAdapterDevices()
[all …]
|
| /external/tensorflow/tensorflow/core/kernels/ |
| D | nccl_ops.cc | 56 int num_devices() const { return num_devices_; } in num_devices() function in tensorflow::__anon27d3229c0111::NcclAsyncOpBase
116 /*num_local_devices=*/num_devices(), in ComputeAsync()
117 /*num_global_devices=*/num_devices(), in ComputeAsync()
147 /*num_local_devices=*/num_devices(), in ComputeAsync()
148 /*num_global_devices=*/num_devices(), in ComputeAsync()
183 /*num_local_devices=*/num_devices(), in ComputeAsync()
184 /*num_global_devices=*/num_devices(), in ComputeAsync()
216 /*num_local_devices=*/num_devices(), in ComputeAsync()
217 /*num_global_devices=*/num_devices(), in ComputeAsync()
252 /*num_local_devices=*/num_devices(), in ComputeAsync()
[all …]
|
| /external/OpenCL-CTS/test_conformance/d3d11/ |
| D | main.cpp | 66 cl_uint num_devices = 0; in main() local
69 TestAdapterEnumeration(platform, pAdapter, pDevice, &num_devices); in main()
72 if (num_devices) in main()
74 TestAdapterDevices(platform, pAdapter, pDevice, pDC, num_devices); in main()
76 num_devices_tested += num_devices; in main()
98 cl_uint* num_devices) in TestAdapterEnumeration() argument
195 *num_devices = num_device_devices; in TestAdapterEnumeration()
208 cl_uint num_devices = 0; in TestAdapterDevices() local
223 &num_devices); in TestAdapterDevices()
228 (num_devices == num_devices_expected), in TestAdapterDevices()
[all …]
|
| /external/tensorflow/tensorflow/python/data/experimental/ops/ |
| D | distribute_options.py | 139 num_devices = options.create_option( variable in DistributeOptions
149 if self.num_devices is not None:
150 pb.num_devices = self.num_devices
156 self.num_devices = pb.num_devices
|
| /external/webrtc/modules/audio_device/ |
| D | audio_device_unittest.cc | 723 int num_devices = audio_device()->PlayoutDevices(); in TEST_P() local
725 num_devices += 2; in TEST_P()
727 EXPECT_GT(num_devices, 0); in TEST_P()
728 for (int i = 0; i < num_devices; ++i) { in TEST_P()
731 EXPECT_EQ(-1, audio_device()->PlayoutDeviceName(num_devices, device_name, in TEST_P()
740 int num_devices = audio_device()->RecordingDevices(); in TEST_P() local
742 num_devices += 2; in TEST_P()
744 EXPECT_GT(num_devices, 0); in TEST_P()
745 for (int i = 0; i < num_devices; ++i) { in TEST_P()
749 EXPECT_EQ(-1, audio_device()->RecordingDeviceName(num_devices, device_name, in TEST_P()
[all …]
|
| /external/mesa3d/src/loader/ |
| D | loader.c | 139 int i, num_devices, fd = -1; in loader_open_render_node() local
141 num_devices = drmGetDevices2(0, devices, MAX_DRM_DEVICES); in loader_open_render_node()
142 if (num_devices <= 0) in loader_open_render_node()
145 for (i = 0; i < num_devices; i++) { in loader_open_render_node()
172 drmFreeDevices(devices, num_devices); in loader_open_render_node()
174 if (i == num_devices) in loader_open_render_node()
306 int i, num_devices, fd = -1; in loader_get_user_preferred_fd() local
324 num_devices = drmGetDevices2(0, devices, MAX_DRM_DEVICES); in loader_get_user_preferred_fd()
325 if (num_devices <= 0) in loader_get_user_preferred_fd()
328 for (i = 0; i < num_devices; i++) { in loader_get_user_preferred_fd()
[all …]
|
| /external/OpenCL-CTS/test_conformance/api/ |
| D | test_platform.cpp | 181 cl_uint num_devices; in test_get_platform_ids() local
232 err = clGetDeviceIDs(platforms[p], CL_DEVICE_TYPE_ALL, 0, NULL, &num_devices); in test_get_platform_ids()
234 devices = (cl_device_id *)malloc(num_devices*sizeof(cl_device_id)); in test_get_platform_ids()
235 memset(devices, 0, sizeof(cl_device_id)*num_devices); in test_get_platform_ids()
236 err = clGetDeviceIDs(platforms[p], CL_DEVICE_TYPE_ALL, num_devices, devices, NULL); in test_get_platform_ids()
239 log_info("\tPlatform has %d devices.\n", (int)num_devices); in test_get_platform_ids()
240 for (int d=0; d<(int)num_devices; d++) { in test_get_platform_ids()
|
| /external/llvm-project/openmp/libomptarget/test/api/ |
| D | omp_get_num_devices.c | 12 int num_devices = omp_get_num_devices(); in test_omp_get_num_devices() local
13 printf("num_devices = %d\n", num_devices); in test_omp_get_num_devices()
18 return (num_devices > 0); in test_omp_get_num_devices()
|
| /external/mesa3d/src/gallium/frontends/clover/api/ |
| D | context.cpp | 71 cl_uint num_devices; in clCreateContextFromType() local
77 ret = clGetDeviceIDs(d_platform, type, 0, NULL, &num_devices); in clCreateContextFromType()
80 devs.resize(num_devices); in clCreateContextFromType()
81 ret = clGetDeviceIDs(d_platform, type, num_devices, devs.data(), 0); in clCreateContextFromType()
85 return clCreateContext(d_props, num_devices, devs.data(), pfn_notify, in clCreateContextFromType()
|
| /external/eigen/unsupported/Eigen/CXX11/src/Tensor/ |
| D | TensorDeviceCuda.h | 60 int num_devices; in initializeDeviceProp() local
61 cudaError_t status = cudaGetDeviceCount(&num_devices); in initializeDeviceProp()
68 m_deviceProperties = new cudaDeviceProp[num_devices]; in initializeDeviceProp()
69 for (int i = 0; i < num_devices; ++i) { in initializeDeviceProp()
119 int num_devices; in initializeDeviceProp() local
120 cudaError_t err = cudaGetDeviceCount(&num_devices); in initializeDeviceProp()
123 assert(device < num_devices); in initializeDeviceProp()
|
| /external/tensorflow/tensorflow/core/api_def/base_api/ |
| D | api_def_NcclReduce.pbtxt | 3 summary: "Reduces `input` from `num_devices` using `reduction` to a single device."
5 Reduces `input` from `num_devices` using `reduction` to a single device.
11 data: the value of the reduction across all `num_devices` devices.
|
| /external/tensorflow/tensorflow/lite/nnapi/ |
| D | nnapi_util.cc | 51 uint32_t num_devices = 0; in GetDeviceNamesList() local
53 NnApiImplementation()->ANeuralNetworks_getDeviceCount(&num_devices); in GetDeviceNamesList()
55 for (uint32_t i = 0; i < num_devices; i++) { in GetDeviceNamesList()
|
| /external/tensorflow/tensorflow/core/common_runtime/ |
| D | ring_gatherer_test.cc | 139 void Init(int num_workers, int num_devices, DataType dtype, in Init() argument
151 for (int di = 0; di < num_devices; ++di) { in Init()
158 int dev_idx = (wi * num_devices) + di; in Init()
190 col_params_->group.group_size = num_workers * num_devices; in Init()
199 int subdiv_stride = num_devices / num_subdivs; in Init()
208 for (int di = 0; di < num_devices; ++di) { in Init()
211 for (int di = 0; di < num_devices; ++di) { in Init()
213 int other = (di + (is_odd ? 7 : 3)) % num_devices; in Init()
224 for (int di = 0; di < num_devices; ++di) { in Init()
239 num_devices; in Init()
[all …]
|
| D | ring_reducer_test.cc | 162 void Init(int num_workers, int num_devices, DataType dtype, in Init() argument
174 for (int di = 0; di < num_devices; ++di) { in Init()
181 int dev_idx = (wi * num_devices) + di; in Init()
213 col_params_->group.group_size = num_workers * num_devices; in Init()
222 int subdiv_stride = num_devices / num_subdivs; in Init()
231 for (int di = 0; di < num_devices; ++di) { in Init()
234 for (int di = 0; di < num_devices; ++di) { in Init()
236 int other = (di + (is_odd ? 7 : 3)) % num_devices; in Init()
248 col_params_->group.num_devices_per_task[task_name] = num_devices; in Init()
249 for (int di = 0; di < num_devices; ++di) { in Init()
[all …]
|
| /external/tensorflow/tensorflow/compiler/xla/experimental/xla_sharding/ |
| D | xla_sharding.py | 129 def split(cls, tensor, split_dimension, num_devices, input_shape=None): argument
150 shape[split_dimension] < num_devices):
153 (shape, split_dimension, num_devices))
156 tile_assignment_dims[split_dimension] = num_devices
162 tile_assignment_devices=range(num_devices)))
300 num_devices, argument
314 return Sharding.split(tensor, split_dimension, num_devices,
|