NVSHMEM API

• API 概述
  • 不支持的 OpenSHMEM 1.3 API
  • 远程网络传输不支持的 OpenSHMEM 1.3 API
  • 支持的 OpenSHMEM API (OpenSHMEM 1.4 和 1.5)
  • 用于 CPU 线程的 NVSHMEM API 扩展
  • 用于 GPU 线程的 NVSHMEM API 扩展
• 库设置、退出和查询
  • NVSHMEM_INIT
  • NVSHMEMX_INIT_ATTR
  • NVSHMEMX_HOSTLIB_INIT_ATTR
  • NVSHMEMX_HOSTLIB_FINALIZE
  • NVSHMEMX_GET_UNIQUE_ID
  • NVSHMEMX_SET_ATTR_UNIQUEID_ARGS
  • NVSHMEMX_CUMODULE_INIT
  • NVSHMEMX_INIT_STATUS
  • NVSHMEM_MY_PE
  • NVSHMEM_N_PES
  • NVSHMEM_FINALIZE
  • NVSHMEM_GLOBAL_EXIT
  • NVSHMEM_PTR
  • NVSHMEMX_MC_PTR
  • NVSHMEM_INFO_GET_VERSION
  • NVSHMEM_INFO_GET_NAME
  • NVSHMEMX_VENDOR_GET_VERSION_INFO
• 线程支持
  • NVSHMEM_INIT_THREAD
  • NVSHMEM_QUERY_THREAD
• 内核启动例程
  • NVSHMEMX_COLLECTIVE_LAUNCH
  • NVSHMEMX_COLLECTIVE_LAUNCH_QUERY_GRIDSIZE
• 内存管理
  • NVSHMEM_MALLOC, NVSHMEM_FREE, NVSHMEM_ALIGN
  • NVSHMEM_CALLOC
  • 内存注册
    • NVSHMEMX_BUFFER_REGISTER
    • NVSHMEMX_BUFFER_UNREGISTER
    • NVSHMEMX_BUFFER_UNREGISTER_ALL
• 团队管理
  • 预定义和应用程序定义的团队
  • 团队句柄
  • 线程安全
  • 集合操作排序
  • 团队创建
  • NVSHMEM_TEAM_MY_PE
  • NVSHMEM_TEAM_N_PES
  • NVSHMEM_TEAM_CONFIG_T
  • NVSHMEM_TEAM_GET_CONFIG
  • NVSHMEM_TEAM_TRANSLATE_PE
  • NVSHMEM_TEAM_SPLIT_STRIDED
  • NVSHMEM_TEAM_SPLIT_2D
  • NVSHMEM_TEAM_DESTROY
• 远程内存访问
  • 阻塞 RMA
    • NVSHMEM_PUT
    • NVSHMEM_P
    • NVSHMEM_IPUT
    • NVSHMEM_GET
    • NVSHMEM_G
    • NVSHMEM_IGET
  • 非阻塞 RMA
    • NVSHMEM_PUT_NBI
    • NVSHMEM_GET_NBI
• 原子内存操作
  • NVSHMEM_ATOMIC_FETCH
  • NVSHMEM_ATOMIC_SET
  • NVSHMEM_ATOMIC_COMPARE_SWAP
  • NVSHMEM_ATOMIC_SWAP
  • NVSHMEM_ATOMIC_FETCH_INC
  • NVSHMEM_ATOMIC_INC
  • NVSHMEM_ATOMIC_FETCH_ADD
  • NVSHMEM_ATOMIC_ADD
  • NVSHMEM_ATOMIC_FETCH_AND
  • NVSHMEM_ATOMIC_AND
  • NVSHMEM_ATOMIC_FETCH_OR
  • NVSHMEM_ATOMIC_OR
  • NVSHMEM_ATOMIC_FETCH_XOR
  • NVSHMEM_ATOMIC_XOR
• 信号操作
  • 信号操作的原子性保证
  • 可用信号运算符
  • NVSHMEM_PUT_SIGNAL
  • NVSHMEM_PUT_SIGNAL_NBI
  • NVSHMEM_SIGNAL_FETCH
  • NVSHMEMX_SIGNAL
  • NVSHMEMX_SIGNAL_OP
• 集合通信
  • 基于团队的集合操作
  • 隐式团队集合操作
  • 从基于团队的集合操作返回的错误代码
  • NVSHMEM_BARRIER_ALL
  • NVSHMEM_BARRIER
  • NVSHMEM_SYNC
  • NVSHMEM_SYNC_ALL
  • NVSHMEM_ALLTOALL
  • NVSHMEM_BROADCAST
  • NVSHMEM_FCOLLECT
  • NVSHMEM_REDUCTIONS
    • AND
    • OR
    • XOR
    • MAX
    • MIN
    • SUM
    • PROD
• 点对点同步
  • NVSHMEM_WAIT_UNTIL
  • NVSHMEM_WAIT_UNTIL_ALL
  • NVSHMEM_WAIT_UNTIL_ANY
  • NVSHMEM_WAIT_UNTIL_SOME
  • NVSHMEM_WAIT_UNTIL_ALL_VECTOR
  • NVSHMEM_WAIT_UNTIL_ANY_VECTOR
  • NVSHMEM_WAIT_UNTIL_SOME_VECTOR
  • NVSHMEM_TEST
  • NVSHMEM_TEST_ALL
  • NVSHMEM_TEST_ANY
  • NVSHMEM_TEST_SOME
  • NVSHMEM_TEST_ALL_VECTOR
  • NVSHMEM_TEST_ANY_VECTOR
  • NVSHMEM_TEST_SOME_VECTOR
  • NVSHMEM_SIGNAL_WAIT_UNTIL
• 内存排序
  • NVSHMEM_FENCE
  • NVSHMEM_QUIET