You can not select more than 25 topics Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
 
 
 
 
Megvii Engine Team 47fe766310 feat(dnn/cuda): add implicit bmm kernels for large kernel depthwise convolution backward filter opr 4 years ago
..
algo.cpp feat(dnn/cuda): add implicit bmm kernels for large kernel depthwise convolution backward filter opr 3 years ago
algo.h feat(dnn/cuda): add implicit bmm large kernel dwconv2d dgrad kernels 3 years ago
bfloat16.cpp style(all): reformat c++ code 4 years ago
chanwise.cpp feat(dnn): use ref ptr interface for all backends 4 years ago
chanwise_small.cpp feat(dnn): use ref ptr interface for all backends 4 years ago
cudnn.cpp feat(dnn): use ref ptr interface for all backends 4 years ago
deconv_int8_helper.cu style(all): reformat c++ code 4 years ago
deconv_int8_helper.cuh style(all): reformat c++ code 4 years ago
group_conv.cpp feat(dnn): use ref ptr interface for all backends 4 years ago
implicit_batched_gemm_float16_nchw_hmma.cpp feat(dnn/cuda): add implicit bmm kernels for large kernel depthwise convolution backward filter opr 3 years ago
implicit_batched_gemm_float32_nchw_fma.cpp feat(dnn/cuda): add implicit bmm kernels for large kernel depthwise convolution backward filter opr 3 years ago
implicit_gemm_int8_nchw4_dp4a.cpp fix(dnn/cuda): fix ci 3 years ago
implicit_gemm_int8_nchw_dp4a.cpp feat(dnn/cuda): add implicit bmm large kernel dwconv2d fprop impl 3 years ago
implicit_gemm_int8_nhwc_imma.cpp feat(dnn/cuda): add implicit bmm large kernel dwconv2d fprop impl 3 years ago
matmul.cpp feat(dnn): use ref ptr interface for all backends 4 years ago

MegEngine 安装包中集成了使用 GPU 运行代码所需的 CUDA 环境,不用区分 CPU 和 GPU 版。 如果想要运行 GPU 程序,请确保机器本身配有 GPU 硬件设备并安装好驱动。 如果你想体验在云端 GPU 算力平台进行深度学习开发的感觉,欢迎访问 MegStudio 平台

C++ Cuda Python C SVG other