Pinned Loading

1. AITemplate AITemplate Public

   Forked from facebookincubator/AITemplate

   AITemplate is a Python framework which renders neural network into high performance CUDA/HIP C++ code. Specialized for FP16 TensorCore (NVIDIA GPU) and MatrixCore (AMD GPU) inference.

   Python

2. c++ multiprocessing Barrier/CondVar/... c++ multiprocessing Barrier/CondVar/SharedMemory using SysV IPC

   1

   #include "ipc.hpp"

   2

   #include <cassert>

   3

   #include <cstring>

   4

   #include <sys/ipc.h>

   5

   #include <sys/sem.h>

3. flat_array_t and flat_tuple_t are au... flat_array_t and flat_tuple_t are automatically hashable type wrapper for std::array<T, N> and std::tuple<Ts...> that can be used as keys in map/unordered_map

   1

   #include <array>

   2

   #include <functional>

   3

   #include <iostream>

   4

   #include <tuple>

   5

   #include <unordered_map>

4. host_vector (a std::vector with a cu... host_vector (a std::vector with a custom allocator based on cudaMallocHost)

   1

   #include "host_vector.hpp"

   2

   #define DEVICE_CHECK(call)                                                                         \

   3

     if ((call) != cudaSuccess) {                                                                     \

   4

       throw std::runtime_error(#call " API call failed: " + GetLastErrorString() + " at " +          \

   5

                                __FILE__ + ", line" + std::to_string(__LINE__));                      \

5. pdf_crop pdf_crop Public

   Python

6. dlp2019 dlp2019 Public archive

   Deep Learning and Practice 2019

   Python