Home
last modified time | relevance | path

Searched refs:SHARED_MEM_PER_BLOCK (Results 1 – 3 of 3) sorted by relevance

/third_party/mindspore/mindspore/ccsrc/runtime/device/gpu/
Dcuda_common.h69 #define SHARED_MEM_PER_BLOCK mindspore::device::gpu::CudaCommon::GetInstance().share_memory_size() macro
/third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/gpu/cuda_impl/
Dtranspose_impl_opt.cu190 …s_large_matrix = short_side >= kMinDimensionToUseTiles && ShmemReqLargeMat <= SHARED_MEM_PER_BLOCK; in Swap3DTensorLast2Dim()
198 ShmemReqNarrowMat <= SHARED_MEM_PER_BLOCK; in Swap3DTensorLast2Dim()
/third_party/mindspore/mindspore/ccsrc/backend/kernel_compiler/gpu/nn/
Dbias_add_grad_gpu_kenel.h153 if (required_sharedmem_size > SHARED_MEM_PER_BLOCK || bias_size_ <= 6) { in MethodSelection()