第一部分、背景

• 目前NVIDIA和华为软件栈的资料比较容易获取,普通人均可从网上下载到
• 想快速看看寒武纪软件栈对外暴露哪些接口

第二部分、操作步骤

1、寻找突破口 - GitHub搜索技巧

我们可以从开源社区寻找线索。在GitHub上搜索关键词“MLU590”（寒武纪的一款AI加速卡），往往能找到一些使用该硬件的开源项目或容器镜像。

2、经过寻找发现MinerU里的一个Docker镜像

经过搜索，我们在OpenDataLab的MinerU项目中找到了一个宝贵的Docker镜像。这个镜像包含了完整的寒武纪软件栈环境，是我们探索API的绝佳起点。

3、下载镜像,创建容器

docker pull crpi-vofi3w62lkohhxsp.cn-shanghai.personal.cr.aliyuncs.com/opendatalab-mineru/mlu:vllm0.8.3-torch2.6.0-torchmlu1.26.1-ubuntu22.04-py310
docker run -ti -v $PWD:/home \
	-w /home \
	crpi-vofi3w62lkohhxsp.cn-shanghai.personal.cr.aliyuncs.com/opendatalab-mineru/mlu:vllm0.8.3-torch2.6.0-torchmlu1.26.1-ubuntu22.04-py310 /bin/bash

4、提取资料

mkdir mlu
cp /torch/dep_libs_download/* ./mlu/ -rf
cp /torch/src/* ./mlu -rf
cp /workspace/* ./mlu -rf
cp /usr/local/neuware ./mlu -rf

目录结构如下

-rw-r--r--  1 root root  49M Feb  4 09:54 cnanalyzeinsight_0.6.0-1.ubuntu22.04_amd64.deb
-rw-r--r--  1 root root 402K Feb  4 09:54 cnclbenchmark_1.8.0-1.ubuntu22.04_amd64.deb
-rw-r--r--  1 root root  84M Feb  4 09:54 cncl_1.27.2-1.ubuntu22.04_amd64.deb
-rw-r--r--  1 root root  13M Feb  4 09:54 cncv_2.10.1-1.ubuntu22.04_amd64.deb
-rw-r--r--  1 root root 131M Feb  4 09:54 cnnl_2.1.2-1.ubuntu22.04_amd64.deb
-rw-r--r--  1 root root  44M Feb  4 09:54 cnnlextra_2.1.1-1.ubuntu22.04_amd64.deb
-rw-r--r--  1 root root 331M Feb  4 09:54 cntoolkit_4.1.2-1.ubuntu22.04_amd64.deb
-rw-r--r--  1 root root 2.1M Feb  4 09:54 cntopo_1.7.2-1.ubuntu22.04_amd64.deb
-rw-r--r--  1 root root  41M Feb  4 09:54 cnvs_1.1.6-1.ubuntu22.04_amd64.deb
-rw-r--r--  1 root root 1.9M Feb  4 09:54 dcmm_0.8.0-1.ubuntu22.04_amd64.deb
-rw-r--r--  1 root root 2.9M Feb  4 09:54 mluops_1.6.0-1.ubuntu22.04_amd64.deb
drwxr-xr-x 25 root root 4.0K Feb  4 09:54 pytorch
drwxr-xr-x  3 root root 4.0K Feb  4 09:54 pytorch_models
drwxr-xr-x 11 root root 4.0K Feb  4 09:54 torch_mlu
drwxr-xr-x  5 root root 4.0K Feb  4 09:54 torchaudio_mlu
drwxr-xr-x  9 root root 4.0K Feb  4 09:55 Cambricon_PyTorch_Model_Zoo
drwxr-xr-x 18 root root 4.0K Feb  4 09:55 DeepSpeed
drwxr-xr-x 12 root root 4.0K Feb  4 09:55 Megatron-LM
drwxr-xr-x  6 root root 4.0K Feb  4 09:55 ffmpeg-mlu-v4.4.0
drwxr-xr-x 11 root root 4.0K Feb  4 09:55 torch_mlu_ops-v1.5.0
drwxr-xr-x 16 root root 4.0K Feb  4 09:55 vllm-v0.8.3
drwxr-xr-x  9 root root 4.0K Feb  4 09:55 vllm-v0.8.3+mlu0.8.0.pt26
drwxr-xr-x 11 root root 4.0K Feb  4 09:55 neuware

5、从头文件中提取API

软件栈的API通常定义在头文件（.h文件）中。我们可以使用工具从这些头文件中提取所有的函数声明

# 进入头文件目录
cd /home/mlu/neuware/include

# 安装ctags工具（用于代码索引）
sudo apt-get install universal-ctags

# 使用ctags提取所有函数声明
# -x：以人类可读格式输出
# --c-kinds=fp --c++-kinds=fp：只提取C和C++函数
# -o api_list.txt：输出到文件
ctags -x --c-kinds=fp --c++-kinds=fp -o api_list.txt *.h

# 统计API数量
cat api_list.txt | awk '{print $1}' | sort | uniq | wc -l \
# 输出：2171（表示有2171个C/C++ API）

# 查看所有API名称
cat api_list.txt | awk '{print $1}' | sort | uniq          # 输出所有API的名字

6、深度分析 - 利用AI理解软件栈架构

获取API列表后，我们可以让AI模型帮助我们进行分析和理解：

# 准备分析提示词，输入给AI模型（如ChatGPT、DeepSeek等）
以下是寒武纪MLU软件栈暴露的API列表，请深入分析并：
1. 画出软件栈的架构层次图
2. 描述各模块的主要功能
3. 阐述软件栈的设计思路
4. 透过软件栈API推断硬件架构设计理念

要求：每一个观点都要有一个代表性的API作为依据

API列表开始：
apiNotAsync
clearBuffer
cnAtomicOperation
cnAtomicReadOps
cnCacheOperation
cnCreateNotifier
cnCreateQueue
...（以下为完整的API列表）

第三部分：方法论总结

这种探索方法不仅适用于寒武纪，也适用于其他相对封闭的技术栈：

1. 寻找公开的容器镜像或构建环境：这是获取完整软件栈的最佳途径
2. 分析头文件提取API：头文件是软件栈的“接口契约”
3. 通过API推断架构：API设计反映了软件架构和硬件能力
4. 利用AI辅助分析：处理大量API时，AI能快速识别模式和关联

---

以下是DeepSeek的输出

基于完整的API列表，我将重新构建包含cnpx和dcmm模块的软件栈架构分析：

一、完整软件栈架构层次图

┌─────────────────────────────────────────────────────────┐
│                   应用层 (Application Layer)             │
│  • AI框架集成 (PyTorch/TensorFlow/MindSpore)            │
│  • 推理引擎、训练框架                                   │
│  • 视觉处理应用                                         │
├─────────────────────────────────────────────────────────┤
│                   算子库层 (Operator Libraries)         │
│  ┌─────────────┐ ┌─────────────┐ ┌─────────────┐       │
│  │   CNNL      │ │   MLU-OP    │ │    CNCv     │       │
│  │ (计算神经网络库) │ │ (算子库)    │ │ (视觉处理库)  │       │
│  └─────────────┘ └─────────────┘ └─────────────┘       │
│  ┌─────────────┐ ┌─────────────┐ ┌─────────────┐       │
│  │   CNJPEG    │ │   CNCODEC   │ │   CNRTC     │       │
│  │ (JPEG编解码)  │ │ (视频编解码) │ │ (运行时编译)  │       │
│  └─────────────┘ └─────────────┘ └─────────────┘       │
├─────────────────────────────────────────────────────────┤
│                   通信与协同层 (Communication Layer)    │
│  ┌─────────────────────────────────────────────────┐   │
│  │               CNCL (集合通信库)                  │   │
│  │      AllReduce/Broadcast/AllGather等             │   │
│  └─────────────────────────────────────────────────┘   │
├─────────────────────────────────────────────────────────┤
│                   运行时层 (Runtime Layer)              │
│  ┌─────────────┐ ┌─────────────┐ ┌─────────────┐       │
│  │  设备管理    │ │  内存管理    │ │  任务调度    │       │
│  │ (cnDevice*) │ │ (cnMem*)    │ │ (cnQueue*)  │       │
│  └─────────────┘ └─────────────┘ └─────────────┘       │
│  ┌─────────────┐ ┌─────────────┐ ┌─────────────┐       │
│  │  事件同步    │ │  内核执行    │ │  拓扑图      │       │
│  │ (cnNotifier*)│ │(cnInvoke*)  │ │(cnTaskTopo* )│      │
│  └─────────────┘ └─────────────┘ └─────────────┘       │
├─────────────────────────────────────────────────────────┤
│                   性能分析层 (Profiling Layer)         │
│  ┌─────────────┐ ┌─────────────┐ ┌─────────────┐       │
│  │   CNPAPI    │ │   CNPERF    │ │    CNPX     │       │
│  │ (硬件计数器) │ │(性能分析器)  │ │(跟踪与分析)  │       │
│  └─────────────┘ └─────────────┘ └─────────────┘       │
├─────────────────────────────────────────────────────────┤
│                   系统管理层 (System Management)        │
│  ┌─────────────┐ ┌─────────────┐ ┌─────────────┐       │
│  │   CNDEV     │ │    DCMM     │ │   CNTOPO    │       │
│  │ (设备控制)   │ │(设备管理监控) │ │(拓扑发现)    │       │
│  └─────────────┘ └─────────────┘ └─────────────┘       │
├─────────────────────────────────────────────────────────┤
│                   驱动层 (Driver Layer)                 │
│  • 内核驱动接口                                         │
│  • 固件管理层                                           │
│  • PCIe/MLU-Link驱动                                   │
├─────────────────────────────────────────────────────────┤
│                   硬件层 (Hardware Layer)               │
│  • MLU计算核心 (NPU/TPU/VPU)                           │
│  • 存储子系统 (HBM/DDR/SRAM)                           │
│  • 互联网络 (MLU-Link/PCIe/RoCE)                       │
│  • 编解码硬件单元                                       │
└─────────────────────────────────────────────────────────┘

二、各模块主要功能及代表性API

1. CNPX模块 - 性能跟踪与分析系统

代表性API: cnpxInit(), cnpxMark(), cnpxRangeStart(), cnpxRangeEnd()

cnpxInit()                    // 初始化性能跟踪系统
cnpxMark()                    // 记录时间戳标记
cnpxRangeStart("conv1")       // 开始一个性能范围
cnpxRangeEnd("conv1")         // 结束性能范围
cnpxNameCNqueue(queue, "训练队列") // 为队列命名

主要功能：

• 应用级性能跟踪和事件记录
• 代码段范围性能分析
• 队列、上下文、设备等资源命名和标记
• 支持自定义payload和模式分析

API依据分析：

• cnpxMark()：轻量级事件标记，用于记录关键时间点
• cnpxRangeStart()/cnpxRangeEnd()：定义性能分析范围
• cnpxNameCN*()：为各类资源赋予语义化名称，便于性能报告分析
• cnpxPayloadSchemaRegister()：注册自定义数据结构，支持领域特定的性能分析

设计思路：提供精细化的应用性能分析能力，支持从算子级别到整个训练流程的全面性能剖析。

2. DCMM模块 - 分布式设备监控与管理

代表性API: dcmmInit(), dcmmConnect(), dcmmGetAllDevices(), dcmmWatchFields()

dcmmInit()                         // 初始化DCMM系统
dcmmConnect("192.168.1.100:9000")  // 连接到远程管理服务
dcmmGetAllDevices(&devices, &count)// 获取所有设备信息
dcmmWatchFields(field_ids, count)  // 监控指定字段变化
dcmmGetMluLinkLinkStatus(dev_id)   // 获取MLU-Link链路状态

主要功能：

• 多设备集群的统一监控和管理
• 远程设备状态查询和控制
• 实时字段监控和告警
• 健康检查和故障诊断
• 功耗、温度、利用率等指标的集中收集

API依据分析：

• dcmmGetAllDevices()：发现和管理集群中的所有设备
• dcmmWatchFields()：监控设备字段变化，支持实时告警
• dcmmGetMluLinkLinkStatus()：查询高速互联链路状态
• dcmmHealthCheck()：执行设备健康检查
• dcmmInjectFieldValue()：注入测试值，用于故障模拟和测试

设计思路：面向数据中心级别的设备管理系统，支持大规模AI集群的集中监控、管理和维护。

3. 设备管理与监控层（CNDEV、DCMM、CNTOPO）

CNDEV - 本地设备控制库

代表性API: cndevGetDeviceCount(), cndevGetTemperatureInfo()

cndevGetDeviceCount(&count)         // 获取本地设备数量
cndevGetTemperatureInfo(dev_id, &temp_info) // 获取温度信息
cndevGetMLULinkStatusV2(dev_id, &link_status) // 获取MLU-Link状态

CNTOPO - 拓扑发现库

代表性API: cntopoInitContext(), cntopoFindTopos()

cntopoInitContext(&ctx)              // 初始化拓扑上下文
cntopoFindTopos(ctx, &topos, &count) // 发现设备拓扑
cntopoGetDevInfoFromDevSet(dev_set, index, &info) // 获取设备信息

三模块协同关系：

本地应用/框架 → CNDEV (本地设备控制)
               ↓
数据中心管理 → DCMM (远程集中管理)
               ↓
硬件拓扑感知 → CNTOPO (拓扑发现优化)

4. 性能分析层（CNPAPI、CNPERF、CNPX）

CNPAPI - 硬件性能计数器接口

代表性API: cnpapiActivityEnable(), cnpapiPmuCreateSession()

cnpapiActivityEnable(ACTIVITY_TYPE_KERNEL) // 启用内核活动记录
cnpapiPmuCreateSession(&session)           // 创建PMU性能监控会话
cnpapiPmuEnableCounter(session, counter_id) // 启用特定性能计数器

CNPERF - 性能分析工具库

代表性API: cnperfStart(), cnperfParserCreateFromSession()

cnperfStart(session)                       // 开始性能数据收集
cnperfParserCreateFromSession(session, &parser) // 创建性能数据解析器
cnperfParserGetKernelPmuData(parser, &data) // 获取内核PMU数据

三层性能分析体系：

CNPX（应用跟踪）→ 应用级性能事件和范围
    ↓
CNPAPI（硬件计数器）→ 硬件性能指标采集
    ↓
CNPERF（综合分析）→ 数据收集、解析和报告

三、软件栈设计思路分析

1. 多层解耦与垂直集成

设计理念：分离设备管理（DCMM）、性能分析（CNPX）、计算运行时（CNRT）

• API依据：
  • dcmmGetAllDevices()（管理） vs cnDeviceGetCount()（运行时）
  • cnpxMark()（分析） vs cnInvokeKernel()（执行）
• 优势：各层可独立升级，维护界面清晰

2. 远程管理与本地控制分离

设计理念：DCMM负责集群级管理，CNDEV负责设备级控制

• API依据：
  • dcmmWatchFields()：远程监控字段变化
  • cndevSetPowerManagementLimitation()：本地设备功耗控制
• 优势：支持云原生部署，实现中心化运维

3. 性能分析全栈覆盖

设计理念：从硬件计数器到应用语义的完整性能分析链

• API依据：
  • cnpapiPmuGetCounterValue()：硬件性能计数器
  • cnperfParserGetKernelPmuData()：性能数据解析
  • cnpxRangeStart()：应用语义标记
• 优势：支持端到端性能优化，精准定位瓶颈

4. 拓扑感知的资源调度

设计理念：基于硬件拓扑的智能资源分配

• API依据：
  • cntopoFindTopos()：发现物理拓扑
  • cndevTopologyGetNearestDevices()：获取邻近设备
  • cnclInitComms()：基于拓扑优化通信
• 优势：最大化利用硬件互联带宽，减少通信延迟

四、硬件架构设计理念推断

1. 分层互联架构

API证据：cndevGetMLULinkStatusV2(), dcmmGetMluLinkLinkStatus()

• 推断设计：
  • 芯片级：MLU-Link高速互联（类似NVLink）
  • 节点级：PCIe/InfiniBand连接
  • 集群级：以太网/RoCE网络
• 硬件体现：多级互联支持大规模扩展

2. 硬件虚拟化与分区

API证据：cndevCreateMluInstanceByProfileId(), cndevSetSMLUMode()

• 推断设计：
  • SMLU模式：单卡虚拟化为多个实例
  • 实例化：按算力、内存划分虚拟设备
  • 资源隔离：计算、内存、带宽隔离
• 硬件体现：支持多租户和细粒度资源分配

3. 专用硬件加速单元

API证据：cncodecDecCreate(), cnjpegDecoderCreate()

• 推断设计：
  • 编解码单元：专用视频/图像处理硬件
  • AI核心：矩阵计算专用单元
  • 视觉处理：图像预处理硬件加速
• 硬件体现：异构计算架构，专用单元提升能效比

4. 复杂的存储层次

API证据：cnMallocConstant(), cnMallocFrameBuffer(), cnMemGetNodeInfo()

• 推断设计：
  • HBM：高带宽内存，用于计算数据
  • 常量内存：只读存储，用于权重和参数
  • 帧缓冲：专用图像/视频内存
  • NUMA结构：多节点内存架构
• 硬件体现：针对AI负载优化的存储子系统

5. 健壮的监控与容错

API证据：cndevGetECCInfo(), cndevGetRetiredPages(), dcmmHealthCheck()

• 推断设计：
  • ECC内存：错误检查和纠正
  • 坏页管理：动态重映射和隔离
  • 健康监控：温度、电压、功耗实时监控
  • 链路诊断：互联网络健康检查
• 硬件体现：企业级可靠性和可用性设计

五、架构特点总结

1. 云原生AI计算平台

• 远程管理：DCMM支持大规模集群集中管理
• 资源虚拟化：硬件实例化支持多租户隔离
• 弹性伸缩：基于拓扑的资源动态分配

2. 全栈性能优化

• 硬件计数器：CNPAPI提供底层性能数据
• 应用语义标记：CNPX连接代码和性能指标
• 综合分析：CNPERF提供端到端优化建议

3. 异构计算集成

• AI核心：专用神经网络计算单元
• 视觉处理：CNCv硬件加速
• 编解码：专用视频处理单元
• 通用计算：灵活的可编程核心

4. 大规模扩展能力

• 高速互联：MLU-Link支持多卡协同
• 拓扑感知：CNTOPO优化跨节点通信
• 集合通信：CNCL针对集群训练优化

5. 企业级可靠性

• 健康监控：全面的设备健康检查
• 容错机制：ECC、坏页管理、链路冗余
• 远程维护：DCMM支持远程诊断和修复

该软件栈展现了寒武纪从芯片到集群的全栈AI计算能力，特别强调：

1. 云原生部署：面向数据中心的大规模AI集群管理
2. 性能透明性：从硬件到应用的完整性能分析链
3. 异构集成：多种专用计算单元的统一编程模型
4. 企业级特性：可靠性、可维护性、安全性设计

这反映了寒武纪不仅在AI芯片设计上，也在软件生态建设上的系统化思考，旨在构建完整的AI计算基础设施。

---

API列表：

apiNotAsync
clearBuffer
cnAtomicOperation
cnAtomicReadOps
cnCacheOperation
cnCreateNotifier
cnCreateQueue
cnCreateQueueWithPriority
cnCtxCreate
cnCtxDestroy
cnCtxGetApiVersion
cnCtxGetConfig
cnCtxGetCurrent
cnCtxGetDevice
cnCtxGetFlags
cnCtxGetQueuePriorityRange
cnCtxResetPersistingL2Cache
cnCtxSetConfig
cnCtxSetCurrent
cnCtxSync
cnDestroyNotifier
cnDestroyQueue
cnDeviceCanPeerAble
cnDeviceGet
cnDeviceGetAttribute
cnDeviceGetByPCIBusId
cnDeviceGetByUuidStr
cnDeviceGetCount
cnDeviceGetDefaultMemPool
cnDeviceGetMemPool
cnDeviceGetName
cnDeviceGetPCIBusId
cnDeviceGetUuid
cnDeviceGetUuidStr
cnDeviceSetMemPool
cnDeviceTotalMem
cnDriverGetVersion
cnFree
cnFreeHost
cnGetCtxConfigParam
cnGetCtxMaxParallelUnionTasks
cnGetDriverVersion
cnGetErrorName
cnGetErrorString
cnGetExportFunction
cnGetLibVersion
cnGetMemAttribute
cnGetMemAttributes
cnInit
cnInvokeHostFunc
cnInvokeKernel
cnInvokeKernelEx
cnIpcCloseMemHandle
cnIpcGetMemHandle
cnIpcGetNotifierHandle
cnIpcOpenMemHandle
cnIpcOpenNotifierHandle
cnKernelGetAttribute
cnMalloc
cnMallocConstant
cnMallocFrameBuffer
cnMallocHost
cnMallocNode
cnMallocNodeConstant
cnMallocPeerAble
cnMallocSecurity
cnMemAddressFree
cnMemAddressReserve
cnMemAllocAsync
cnMemAllocFromPoolAsync
cnMemCreate
cnMemExportToShareableHandle
cnMemFreeAsync
cnMemGetAccess
cnMemGetAddressRange
cnMemGetAllocationGranularity
cnMemGetAllocationPropertiesFromHandle
cnMemGetHandleForAddressRange
cnMemGetInfo
cnMemGetNodeInfo
cnMemImportFromShareableHandle
cnMemMap
cnMemMerge
cnMemPoolCreate
cnMemPoolDestroy
cnMemPoolGetAccess
cnMemPoolGetAttribute
cnMemPoolSetAttribute
cnMemPoolTrimTo
cnMemRelease
cnMemRetainAllocationHandle
cnMemSetAccess
cnMemUnmap
cnMemcpy
cnMemcpy2D
cnMemcpy2DAsync
cnMemcpy3D
cnMemcpy3DAsync
cnMemcpyAsync
cnMemcpyAsync_V2
cnMemcpyAsync_V3
cnMemcpyDtoD
cnMemcpyDtoD2D
cnMemcpyDtoD3D
cnMemcpyDtoDAsync
cnMemcpyDtoH
cnMemcpyDtoHAsync
cnMemcpyDtoHAsync_V2
cnMemcpyHtoD
cnMemcpyHtoDAsync
cnMemcpyHtoDAsync_V2
cnMemcpyPeer
cnMemcpyPeerAsync
cnMemsetD16
cnMemsetD16Async
cnMemsetD32
cnMemsetD32Async
cnMemsetD8
cnMemsetD8Async
cnMmap
cnMmapCached
cnModuleGetKernel
cnModuleGetLoadingMode
cnModuleGetSymbol
cnModuleLoad
cnModuleLoadData
cnModuleLoadFatBinary
cnModuleQueryFatBinaryMemoryUsage
cnModuleQueryMemoryUsage
cnModuleUnload
cnMunmap
cnNotifierElapsedExecTime
cnNotifierElapsedTime
cnPlaceNotifier
cnPlaceNotifierWithFlags
cnProfilerStart
cnProfilerStop
cnQueryNotifier
cnQueryQueue
cnQueueAddCallback
cnQueueAtomicOperation
cnQueueBeginCapture
cnQueueCopyAttributes
cnQueueEndCapture
cnQueueGetAttribute
cnQueueGetCaptureInfo
cnQueueGetContext
cnQueueGetId
cnQueueGetPriority
cnQueueIsCapturing
cnQueueSetAttribute
cnQueueSync
cnQueueUpdateCaptureDependencies
cnQueueWaitNotifier
cnQueueWaitNotifierWithFlags
cnSetCtxConfigParam
cnSetMemAttribute
cnSetMemRangeAttribute
cnSharedContextAcquire
cnSharedContextGetState
cnSharedContextRelease
cnSharedContextReset
cnSharedContextSetFlags
cnTaskTopoAcquireUserObject
cnTaskTopoAddChildTopoNode
cnTaskTopoAddDependencies
cnTaskTopoAddEmptyNode
cnTaskTopoAddHostNode
cnTaskTopoAddKernelNode
cnTaskTopoAddMemcpyNode
cnTaskTopoAddMemsetNode
cnTaskTopoAddNotifierPlaceNode
cnTaskTopoAddNotifierWaitNode
cnTaskTopoChildTopoNodeGetTopo
cnTaskTopoClone
cnTaskTopoCreate
cnTaskTopoDebugDotPrint
cnTaskTopoDestroy
cnTaskTopoDestroyNode
cnTaskTopoEntityChildTopoNodeSetParams
cnTaskTopoEntityDestroy
cnTaskTopoEntityHostNodeSetParams
cnTaskTopoEntityInvoke
cnTaskTopoEntityKernelNodeSetParams
cnTaskTopoEntityMemcpyNodeSetParams
cnTaskTopoEntityMemsetNodeSetParams
cnTaskTopoEntityNotifierPlaceNodeSetNotifier
cnTaskTopoEntityNotifierWaitNodeSetNotifier
cnTaskTopoEntityUpdate
cnTaskTopoGetEdges
cnTaskTopoGetNodes
cnTaskTopoGetRootNodes
cnTaskTopoHostNodeGetParams
cnTaskTopoHostNodeSetParams
cnTaskTopoInstantiate
cnTaskTopoKernelNodeCopyAttributes
cnTaskTopoKernelNodeGetAttribute
cnTaskTopoKernelNodeGetParams
cnTaskTopoKernelNodeSetAttribute
cnTaskTopoKernelNodeSetParams
cnTaskTopoMemcpyNodeGetParams
cnTaskTopoMemcpyNodeSetParams
cnTaskTopoMemsetNodeGetParams
cnTaskTopoMemsetNodeSetParams
cnTaskTopoNodeFindInClone
cnTaskTopoNodeGetDependencies
cnTaskTopoNodeGetDependentNodes
cnTaskTopoNodeGetType
cnTaskTopoNotifierPlaceNodeGetNotifier
cnTaskTopoNotifierPlaceNodeSetNotifier
cnTaskTopoNotifierWaitNodeGetNotifier
cnTaskTopoNotifierWaitNodeSetNotifier
cnTaskTopoReleaseUserObject
cnTaskTopoRemoveDependencies
cnTaskTopoUpload
cnThreadExchangeQueueCaptureMode
cnUserObjectAcquire
cnUserObjectCreate
cnUserObjectRelease
cnWaitNotifier
cnZmalloc
cnZmallocNode
cnclAbortComm
cnclAllGather
cnclAllReduce
cnclAlltoAll
cnclAlltoAllv
cnclBcast
cnclBroadcast
cnclDestroyComms
cnclFreeComm
cnclGetCliqueId
cnclGetCommAsyncError
cnclGetCommCount
cnclGetCommDevice
cnclGetCommRank
cnclGetErrorStr
cnclGetLibVersion
cnclGroupEnd
cnclGroupStart
cnclInitComms
cnclRecv
cnclReduce
cnclReduceScatter
cnclSend
cnclSetCommConfig
cncodecDecCreate
cncodecDecDestroy
cncodecDecFrameRef
cncodecDecFrameUnref
cncodecDecGetCaps
cncodecDecGetJpegInfo
cncodecDecJpegSupported
cncodecDecJpegSyncDecode
cncodecDecQueryBufStatus
cncodecDecSendStream
cncodecDecSetEos
cncodecDecSetParams
cncodecEncCreate
cncodecEncDestroy
cncodecEncGetCaps
cncodecEncGetJpegBufSize
cncodecEncGetPresetConfig
cncodecEncJpegSyncEncode
cncodecEncQueryBufStatus
cncodecEncQueryPicStats
cncodecEncSendFrame
cncodecEncSetEos
cncodecEncSetParams
cncodecEncStreamRef
cncodecEncStreamUnref
cncodecEncWaitAvailInputBuf
cncodecGetLibVersion
cncodecInitLogging
cncodecSetMluAffinity
cncvAbsDiff_Basic
cncvAbsDiff_V2
cncvAddWeighted_Basic
cncvAdd_Basic
cncvAdjustContrast_BasicV3
cncvAdjustContrast_V2
cncvAdjustHue
cncvAdjustHue_Basic
cncvAdjustSaturation
cncvAdjustSaturation_Basic
cncvAlphaBlend_BasicROI
cncvAlphaComp_BasicROI
cncvAverageErrorEx
cncvAverageError_Basic
cncvBGSViBeCreate
cncvBGSViBeDestroy
cncvBGSViBeGetHistoryImageSize
cncvBGSViBeInit
cncvBGSViBeUpdate
cncvBgrToLab
cncvBitwiseAnd_Basic
cncvBitwiseAnd_V2
cncvBlendAlphaEx
cncvBlur
cncvBufferListCreate
cncvBufferListDestroy
cncvBufferListWrapDevBuffers
cncvCalcHist_AdvancedROI
cncvCalcHist_AdvancedROI_V2
cncvColorTwist
cncvColorTwist_Advance
cncvColorTwist_Advanced_V2
cncvCompareAndFindMinLoc_ROI
cncvConvertTo
cncvConvertTo_Advanced
cncvConvertTo_Advanced_V2
cncvCopyTo_ROI
cncvCreate
cncvCropMirrorNormalize_AdvancedCuboid
cncvCropMirrorNormalize_AdvancedCuboid_V2
cncvCrop_AdvancedROI
cncvCrop_AdvancedROI_V2
cncvCvtColorEx
cncvDestroy
cncvDilate
cncvDilate_Basic
cncvDivide
cncvDrawLinesEx
cncvErode
cncvErode_Basic
cncvFill_AdvancedROI
cncvFill_ROI
cncvFilter
cncvFlip
cncvGaussianBlur
cncvGenBlurKernel
cncvGenDerivKernel
cncvGenGaussianKernel
cncvGenLaplacianKernel
cncvGetAdjustContrastWorkspaceSize
cncvGetAffineTransform
cncvGetCalcHistWorkSpace
cncvGetColorTwistWorkspaceSize
cncvGetConvertToWorkspaceSize
cncvGetCropMirrorNormalizeWorkspaceSize
cncvGetCropWorkspaceSize
cncvGetErrorString
cncvGetLibVersion
cncvGetMeanStdWorkspaceSize
cncvGetPerspectiveTransform
cncvGetQueue
cncvGetResizeConvertWorkspaceSize
cncvGetResizeGrayWorkspaceSize
cncvGetResizeRgbxAdvancedWorkspaceSize
cncvGetResizeWorkspaceSize
cncvGetResizeYuvWorkspaceSize
cncvGetRotationMatrix2D
cncvGetWarpAffineWorkspaceSize
cncvGetYuvToRgbxAdvancedWorkspaceSize
cncvGetYuvToRgbxWorkspaceSize
cncvHOGCompute
cncvHOGCreate
cncvHOGDestroy
cncvHOGGetDescriptorSize
cncvHOGGetWorkspaceSize
cncvHsvToRgbx
cncvHsvToRgbx_Basic
cncvImageBufferCreate
cncvImageBufferCreateV2
cncvImageBufferDestroy
cncvLabToBgr
cncvLaplacian
cncvLogInit
cncvMeanEx
cncvMeanStd
cncvMeanStd_V2
cncvMean_BasicROI
cncvMerge_BasicROI
cncvMultiply_Basic
cncvPad_BasicROI
cncvPolyLinesYuv_Basic
cncvQueue
cncvQueueCreate
cncvQueueDestroy
cncvRectangleEx
cncvRectangle_Basic
cncvRemap_ROI
cncvResizeBlendEx
cncvResizeConvertApply_AdvancedROI
cncvResizeConvertCreate
cncvResizeConvertDestroy
cncvResizeConvertGetAuxDataSize
cncvResizeConvertInitAuxData
cncvResizeConvertSetOp_AdvancedROI
cncvResizeConvert_AdvancedROI
cncvResizeConvert_AdvancedROIV2
cncvResizeEx
cncvResizeExV2
cncvResizeGetAuxDataSize
cncvResizeGray_AdvancedROI
cncvResizeInitAuxData
cncvResizeRgbx_AdvancedROI
cncvResizeRgbx_AdvancedROIV2
cncvResizeYuv_AdvancedROI
cncvResizeYuv_AdvancedROI_V2
cncvResize_AdvancedROI
cncvResize_AdvancedROI_V2
cncvRgbxToGray_ROI
cncvRgbxToHsv
cncvRgbxToHsv_Basic
cncvRgbxToRgbx_BasicROI
cncvRgbxToRgbx_ROI
cncvRgbxToYuv_BasicROI
cncvRgbxToYuv_BasicROIP2
cncvRotate
cncvRotateEx
cncvRotate_Basic
cncvScaleAdd_Basic
cncvSetClusterNum
cncvSetQueue
cncvSobel
cncvSplit_BasicROI
cncvSplit_ROI
cncvSubtract_Basic
cncvSyncQueue
cncvThreshold
cncvThreshold_Basic
cncvTranspose
cncvUpdateContextInformation
cncvWarpAffine_AdvancedROI
cncvWarpAffine_AdvancedROI_V2
cncvWarpAffine_BasicROI
cncvWarpAffine_V3
cncvWarpPerspective
cncvYuvToRgbx
cncvYuvToRgbx_AdvancedROI
cncvYuvToRgbx_AdvancedROI_V2
cncvYuvToRgbx_Basic
cncvYuvToRgbx_V2
cncvYuvToYuv
cndevClearCurrentThreadAffinity
cndevCreateMluInstanceByProfileId
cndevCreateMluInstanceByProfileIdWithPlacement
cndevCreateMluInstanceByProfileName
cndevCreateMluInstanceByProfileNameWithPlacement
cndevCreateSMluInstanceByProfileId
cndevCreateSMluInstanceByProfileName
cndevCreateSMluProfileInfo
cndevDestroyMluInstanceByHandle
cndevDestroyMluInstanceByInstanceName
cndevDestroySMluInstanceByHandle
cndevDestroySMluInstanceByInstanceName
cndevDestroySMluProfileInfo
cndevDeviceActiveConfigs
cndevDeviceGetConfigs
cndevDeviceGetFieldValues
cndevDeviceGetMinorNumber
cndevDeviceResetConfigs
cndevDeviceSetConfigs
cndevEventHandleCreate
cndevEventHandleDestroy
cndevEventWait
cndevGetAddressSwaps
cndevGetAddressSwapsV2
cndevGetAllMluInstanceInfo
cndevGetAllSMluInstanceInfo
cndevGetApplicationsClock
cndevGetBAR4MemoryInfo
cndevGetCRCInfo
cndevGetCardHealthState
cndevGetCardHealthStateV2
cndevGetCardHeartbeatCount
cndevGetCardName
cndevGetCardNameString
cndevGetCardNameStringByDevId
cndevGetCardPartNumber
cndevGetCardSN
cndevGetCardVfState
cndevGetChassisInfoV2
cndevGetChipId
cndevGetClusterCount
cndevGetCodecTurbo
cndevGetComputeCapability
cndevGetComputeMode
cndevGetCoreCount
cndevGetCurrentInfo
cndevGetCurrentPCIInfo
cndevGetDDRInfo
cndevGetDevIdByBDF
cndevGetDeviceAffinity
cndevGetDeviceCPUSamplingInterval
cndevGetDeviceCPUUtilizationV2
cndevGetDeviceCanPeerAble
cndevGetDeviceCount
cndevGetDeviceHandleByIndex
cndevGetDeviceHandleByPciBusId
cndevGetDeviceHandleBySerial
cndevGetDeviceHandleByUUID
cndevGetDeviceHandleFromMluInstanceHandle
cndevGetDeviceOsMemoryUsageV2
cndevGetDevicePowerInfo
cndevGetDeviceUtilizationInfo
cndevGetDockerParam
cndevGetDriverVersion
cndevGetECCInfo
cndevGetEccMode
cndevGetErrorString
cndevGetExportFunction
cndevGetFanSpeedInfo
cndevGetFastAlloc
cndevGetFrequencyInfo
cndevGetImageCodecUtilization
cndevGetLastError
cndevGetLibVersion
cndevGetLowestLinkSpeed
cndevGetLowestSupportDriverVersion
cndevGetMLUFrequencyStatus
cndevGetMLULinkBasicCounter
cndevGetMLULinkCapability
cndevGetMLULinkCongestionCtrlCounter
cndevGetMLULinkCounter
cndevGetMLULinkCounterExt
cndevGetMLULinkDevSN
cndevGetMLULinkErrorCounter
cndevGetMLULinkEventCounter
cndevGetMLULinkFlowCtrlCounter
cndevGetMLULinkOPN
cndevGetMLULinkOverRoCECtrl
cndevGetMLULinkPPI
cndevGetMLULinkPortIP
cndevGetMLULinkPortMode
cndevGetMLULinkPortNumber
cndevGetMLULinkRemoteInfo
cndevGetMLULinkSpeedInfo
cndevGetMLULinkState
cndevGetMLULinkStatusV2
cndevGetMLULinkTaskStatsCounter
cndevGetMLULinkVersion
cndevGetMaxMluInstanceCount
cndevGetMaxPCIInfo
cndevGetMemEccCounter
cndevGetMemoryUsageV2
cndevGetMimMode
cndevGetMluInstance
cndevGetMluInstanceById
cndevGetMluInstanceByIndex
cndevGetMluInstanceId
cndevGetMluInstanceInfo
cndevGetMluInstancePossiblePlacements
cndevGetMluInstanceProfileInfo
cndevGetMluInstanceRemainingCapacity
cndevGetNUMANodeIdByDevId
cndevGetNUMANodeIdByTopologyNode
cndevGetNodeByBDF
cndevGetNodeByDevId
cndevGetNodeByDeviceName
cndevGetNodeCapabilityInfo
cndevGetOpticalInfo
cndevGetOverTemperatureInfo
cndevGetPCIeFirmwareVersion
cndevGetPCIeInfoV2
cndevGetPCIethroughputV2
cndevGetParityError
cndevGetPcieReplayCounter
cndevGetPerformanceThrottleReason
cndevGetPowerManagementDefaultLimitation
cndevGetPowerManagementLimitation
cndevGetPowerManagementLimitationRange
cndevGetProcessInfo
cndevGetProcessUtilization
cndevGetProcessUtilizationInfo
cndevGetRemappedRows
cndevGetRemappedRowsV2
cndevGetRemoteHostMgmtAddr
cndevGetRemoteHostSN
cndevGetRepairStatus
cndevGetRetiredPages
cndevGetRetiredPagesOperation
cndevGetRetiredPagesStatus
cndevGetSMLUMode
cndevGetSMluInstanceInfo
cndevGetSMluProfileIdInfo
cndevGetSMluProfileInfo
cndevGetSRIOVMode
cndevGetScalerUtilization
cndevGetSramEccHistogram
cndevGetSupportedEventTypes
cndevGetTemperatureInfo
cndevGetTinyCoreUtilization
cndevGetUUID
cndevGetVersionInfo
cndevGetVideoCodecUtilization
cndevGetVoltageInfo
cndevInit
cndevInjectEventError
cndevRegisterEvents
cndevRelease
cndevResetDevice
cndevResetMLULinkAllCounters
cndevResetMLULinkCounter
cndevSetComputeMode
cndevSetCurrentThreadAffinity
cndevSetDeviceCPUSamplingInterval
cndevSetEccMode
cndevSetMLULinkState
cndevSetMimMode
cndevSetPowerManagementLimitation
cndevSetSMLUMode
cndevSetSRIOVMode
cndevTopologyGetCpuRelatedDevices
cndevTopologyGetNearestDevices
cndevTopologyGetRelationship
cndevTopologyGetRelationshipByNode
cndevTopologyGetVirtualRootNode
cndevTopologyTraverseTree
cndevUnlockMLUFrequency
cndevUpdateSMluInstanceQuotaByHandle
cndevUpdateSMluInstanceQuotaByInstanceName
cnjpegDecode
cnjpegDecodeBatched
cnjpegDecodeBatchedEx
cnjpegDecodeBatchedInitialize
cnjpegDecodeBatchedParseJpegTables
cnjpegDecodeEx
cnjpegDecodeParamsCreate
cnjpegDecodeParamsDestroy
cnjpegDecodeParamsSetOutputFormat
cnjpegDecodeParamsSetROI
cnjpegDecodeParamsSetScaleFactor
cnjpegDecoderCreate
cnjpegDecoderDestroy
cnjpegDecoderJpegSupported
cnjpegEncodeGetBufferSize
cnjpegEncodeImage
cnjpegEncodeParamsCopyHuffmanTables
cnjpegEncodeParamsCopyMetadata
cnjpegEncodeParamsCopyQuantizationTables
cnjpegEncodeParamsCreate
cnjpegEncodeParamsDestroy
cnjpegEncodeParamsSetEncoding
cnjpegEncodeParamsSetOptimizedHuffman
cnjpegEncodeParamsSetQuality
cnjpegEncodeParamsSetSamplingFactors
cnjpegEncodeRetrieveBistream
cnjpegEncodeRetrieveBistreamDevice
cnjpegEncodeYUV
cnjpegEncoderCreate
cnjpegEncoderDestroy
cnjpegGetImageInfo
cnjpegGetLibVersion
cnjpegJpegStreamGetChromaSubsampling
cnjpegJpegStreamGetComponentDimensions
cnjpegJpegStreamGetComponentsNum
cnjpegJpegStreamGetFrameDimensions
cnjpegJpegStreamGetJpegEncoding
cnjpegJpegStreamParse
cnjpegJpegStreamParseHeader
cnjpegJpegStreamParseTables
cnjpegStreamCreate
cnjpegStreamDestroy
cnnlAbs
cnnlActivationBackward
cnnlActivationForward
cnnlAdaptivePoolingBackward
cnnlAdaptivePoolingForward_v2
cnnlAddN_v2
cnnlAddcdiv
cnnlAddcmul
cnnlAdvancedIndex_v2
cnnlAngle
cnnlApplyAdaGrad
cnnlApplyAdaGradV2
cnnlApplyAdaMax
cnnlApplyAdadelta
cnnlApplyAdam
cnnlApplyAddSign
cnnlApplyCenterRMSProp
cnnlApplyFtrlV2
cnnlApplyProximalAdagrad
cnnlArange_v3
cnnlAsStrided
cnnlAsStridedBackward
cnnlAtan2
cnnlBatch2space
cnnlBatch2spaceNd_v2
cnnlBatchGatherV2_v2
cnnlBatchMatMulEx
cnnlBatchNormBackward_v2
cnnlBatchNormForwardInference
cnnlBatchNormForwardInferenceV2
cnnlBatchNormForwardTraining_v2
cnnlBceLoss
cnnlBceLossBackward
cnnlBceWithLogits
cnnlBceWithLogitsBackward
cnnlBertPre_v2
cnnlBiasActivationGluBackward_v2
cnnlBiasActivationGluForward_v2
cnnlBiasAdd
cnnlBiasAddBackward_v2
cnnlBiasDropoutAddFusedTrainBackward
cnnlBiasDropoutAddFusedTrain_v2
cnnlBincount
cnnlBitCompute_v2
cnnlBoxOverlapBev
cnnlBucketize
cnnlCTCLoss
cnnlCastDataType
cnnlCdistBackward
cnnlCdistForward
cnnlCeil
cnnlClipGradNorm_v2
cnnlClip_v3
cnnlCol2Im
cnnlComplexAbs
cnnlConcat
cnnlConj
cnnlConvolutionBackwardData
cnnlConvolutionBackwardFilter
cnnlConvolutionForward
cnnlCopySign
cnnlCopy_v2
cnnlCos
cnnlCos_v2
cnnlCosineSimilarity
cnnlCreate
cnnlCreateActivationDescriptor
cnnlCreateBertPreDescriptor
cnnlCreateBiasActivationGluDescriptor
cnnlCreateBiasDropoutAddFusedTrainBackwardDescriptor
cnnlCreateBiasDropoutAddFusedTrainDescriptor
cnnlCreateCTCLossDescriptor
cnnlCreateConvolutionDescriptor
cnnlCreateDCNDescriptor
cnnlCreateDeconvolutionDescriptor
cnnlCreateDetectionOutputDescriptor
cnnlCreateDivDescriptor
cnnlCreateDlrmInteractDescriptor
cnnlCreateEmbeddingBagDescriptor_v2
cnnlCreateFlashAttentionDescriptor
cnnlCreateFlatSearchDescriptor
cnnlCreateFloatQuantizeDescriptor
cnnlCreateFusedOpsConstParamPack
cnnlCreateFusedOpsPlan
cnnlCreateFusedOpsVariantParamPack
cnnlCreateGrepDescriptor
cnnlCreateGridSampleDescriptor
cnnlCreateGroupGemmAlgo
cnnlCreateGroupGemmDescriptor
cnnlCreateGroupGemmHeuristicResult
cnnlCreateGroupGemmTensorDescriptor
cnnlCreateGroupTensorDescriptors
cnnlCreateGruDescriptor
cnnlCreateHistogramDescriptor
cnnlCreateIndexPutDescriptor
cnnlCreateInterpDescriptor
cnnlCreateLLMQuantMatmulDescriptor
cnnlCreateMatMulAlgo
cnnlCreateMatMulDescriptor
cnnlCreateMatMulExAlgo
cnnlCreateMatMulExDescriptor
cnnlCreateMatMulExHeuristicResult
cnnlCreateMatMulHeuristicResult
cnnlCreateNmsDescriptor
cnnlCreateNormDesc
cnnlCreateNormalizeDescriptor
cnnlCreateOpTensorDescriptor
cnnlCreatePackPaddedSequenceDescriptor
cnnlCreatePadPackedSequenceDescriptor
cnnlCreatePoolingDescriptor
cnnlCreateProposalDescriptor
cnnlCreateProposalFpnDescriptor
cnnlCreateQuantizeExDescriptor
cnnlCreateRNNDescriptor
cnnlCreateRNNTLossDescriptor
cnnlCreateReduceDescriptor
cnnlCreateRelPositionMultiHeadAttentionDescriptor
cnnlCreateReorgDescriptor
cnnlCreateRoiAlignDescriptor
cnnlCreateRoialignDescriptor
cnnlCreateRotaryEmbeddingDescriptor
cnnlCreateSeqDataDescriptor
cnnlCreateSpaceBatchNdDescriptor
cnnlCreateSparseDenseMatMulDescriptor
cnnlCreateSparseTensorDescriptor
cnnlCreateStdVarMeanDescriptor
cnnlCreateStrideBatchMatMulAlgo
cnnlCreateStrideBatchMatMulDescriptor
cnnlCreateStrideBatchMatMulHeuristicResult
cnnlCreateTensorDescriptor
cnnlCreateTensorSetDescriptor
cnnlCreateTransformerAttentionDescriptor
cnnlCreateTransformerAttentionQuantizeDescriptor
cnnlCreateTransformerAttnProjDescriptor
cnnlCreateTransformerAttnProjQuantifyDescriptor
cnnlCreateTransformerBeamSearchDescriptor
cnnlCreateTransformerFFNDescriptor
cnnlCreateTransformerFeedForwardDescriptor
cnnlCreateTransformerFeedForwardQuantizeDescriptor
cnnlCreateTransformerSelfAttnDescriptor
cnnlCreateTransposeDescriptor
cnnlCreateTrigonDescriptor
cnnlCreateUniqueConsecutiveDescriptor
cnnlCreateUniqueDescriptor
cnnlCreateWindowAttentionDescriptor
cnnlCropAndResize
cnnlCropAndResizeBackwardBoxes
cnnlCropAndResizeBackwardImage
cnnlCross
cnnlCrossEntropyBackward_v2
cnnlCrossEntropyForward_v2
cnnlCummax
cnnlCummin
cnnlCumprod_v2
cnnlCumsum_v2
cnnlDCNBackwardData
cnnlDCNBackwardWeight
cnnlDCNForward
cnnlDeconvolution
cnnlDestroy
cnnlDestroyActivationDescriptor
cnnlDestroyBertPreDescriptor
cnnlDestroyBiasActivationGluDescriptor
cnnlDestroyBiasDropoutAddFusedTrainBackwardDescriptor
cnnlDestroyBiasDropoutAddFusedTrainDescriptor
cnnlDestroyCTCLossDescriptor
cnnlDestroyConvolutionDescriptor
cnnlDestroyDCNDescriptor
cnnlDestroyDeconvolutionDescriptor
cnnlDestroyDetectionOutputDescriptor
cnnlDestroyDivDescriptor
cnnlDestroyDlrmInteractDescriptor
cnnlDestroyEmbeddingBagDescriptor_v2
cnnlDestroyFlashAttentionDescriptor
cnnlDestroyFlatSearchDescriptor
cnnlDestroyFloatQuantizeDescriptor
cnnlDestroyFusedOpsConstParamPack
cnnlDestroyFusedOpsPlan
cnnlDestroyFusedOpsVariantParamPack
cnnlDestroyGrepDescriptor
cnnlDestroyGridSampleDescriptor
cnnlDestroyGroupGemmAlgo
cnnlDestroyGroupGemmDescriptor
cnnlDestroyGroupGemmHeuristicResult
cnnlDestroyGroupGemmTensorDescriptor
cnnlDestroyGroupTensorDescriptors
cnnlDestroyGruDescriptor
cnnlDestroyHistogramDescriptor
cnnlDestroyIndexPutDescriptor
cnnlDestroyInterpDescriptor
cnnlDestroyLLMQuantMatmulDescriptor
cnnlDestroyMatMulAlgo
cnnlDestroyMatMulDescriptor
cnnlDestroyMatMulExAlgo
cnnlDestroyMatMulExDescriptor
cnnlDestroyMatMulExHeuristicResult
cnnlDestroyMatMulHeuristicResult
cnnlDestroyNmsDescriptor
cnnlDestroyNormDesc
cnnlDestroyNormalizeDescriptor
cnnlDestroyOpTensorDescriptor
cnnlDestroyPackPaddedSequenceDescriptor
cnnlDestroyPadPackedSequenceDescriptor
cnnlDestroyPoolingDescriptor
cnnlDestroyProposalDescriptor
cnnlDestroyProposalFpnDescriptor
cnnlDestroyQuantizeExDescriptor
cnnlDestroyRNNDescriptor
cnnlDestroyRNNTLossDescriptor
cnnlDestroyReduceDescriptor
cnnlDestroyRelPositionMultiHeadAttentionDescriptor
cnnlDestroyReorgDescriptor
cnnlDestroyRoiAlignDescriptor
cnnlDestroyRoialignDescriptor
cnnlDestroyRotaryEmbeddingDescriptor
cnnlDestroySeqDataDescriptor
cnnlDestroySpaceBatchNdDescriptor
cnnlDestroySparseDenseMatMulDescriptor
cnnlDestroySparseTensorDescriptor
cnnlDestroyStdVarMeanDescriptor
cnnlDestroyStrideBatchMatMulAlgo
cnnlDestroyStrideBatchMatMulDescriptor
cnnlDestroyStrideBatchMatMulHeuristicResult
cnnlDestroyTensorDescriptor
cnnlDestroyTensorSetDescriptor
cnnlDestroyTransformerAttentionDescriptor
cnnlDestroyTransformerAttentionQuantizeDescriptor
cnnlDestroyTransformerAttnProjDescriptor
cnnlDestroyTransformerAttnProjQuantifyDescriptor
cnnlDestroyTransformerBeamSearchDescriptor
cnnlDestroyTransformerFFNDescriptor
cnnlDestroyTransformerFeedForwardDescriptor
cnnlDestroyTransformerFeedForwardQuantizeDescriptor
cnnlDestroyTransformerSelfAttnDescriptor
cnnlDestroyTransposeDescriptor
cnnlDestroyTrigonDescriptor
cnnlDestroyUniqueConsecutiveDescriptor
cnnlDestroyUniqueDescriptor
cnnlDestroyWindowAttentionDescriptor
cnnlDet_v2
cnnlDetectionOutput_v2
cnnlDiag
cnnlDiagPart
cnnlDiagonal
cnnlDivNoNan
cnnlDiv_v3
cnnlDlrmInteract
cnnlDynamicFloatQuantize
cnnlDynamicPartition
cnnlDynamicStitch_v2
cnnlEmbeddingBackward
cnnlEmbeddingBagBackward_v2
cnnlEmbeddingBag_v3
cnnlEmbeddingForward_v2
cnnlErf_v2
cnnlErfinv
cnnlExp
cnnlExpand
cnnlExpm1_v2
cnnlExtraGetGenCaseDirectory
cnnlExtraGetLibVersion
cnnlExtraSetGenCaseDirectory
cnnlExtraSetGenCaseMode
cnnlFill_v4
cnnlFindConvolutionForwardAlgo
cnnlFlashAttentionBackward_v2
cnnlFlashAttentionForward_v2
cnnlFlatSearch_v2
cnnlFlip
cnnlFloor
cnnlFloorDiv
cnnlFloorMod
cnnlFloorModTrunc
cnnlForeachBinaryOp
cnnlForeachCopy
cnnlForeachLerp
cnnlForeachNorm
cnnlForeachPointwiseOp
cnnlForeachUnaryOp
cnnlForeachUnaryOp_v2
cnnlFractionalMaxPoolForward
cnnlFrozenBatchNormBackward_v2
cnnlFuseNorm_v4
cnnlFusedDropout_v3
cnnlFusedOpsExecute
cnnlFwFFMBackward
cnnlFwFFMForward
cnnlGRUCellBackward
cnnlGRUCellForward
cnnlGatherNd_v2
cnnlGather_v2
cnnlGenerateRandDiscreteUniform
cnnlGenerateRandExponential
cnnlGenerateRandMultinomial
cnnlGenerateRandNormal
cnnlGenerateRandNormal_v2
cnnlGenerateRandPoisson
cnnlGenerateRandTruncatedNormal
cnnlGenerateRandTruncatedNormal_v2
cnnlGenerateRandUniform
cnnlGetActivationDescAttr
cnnlGetAdaptivePoolingForwardWorkspaceSize_v2
cnnlGetAddNWorkspaceSize
cnnlGetAddcdivWorkspaceSize_v2
cnnlGetAddcmulWorkspaceSize_v2
cnnlGetAdvancedIndexOutputDim_v2
cnnlGetAdvancedIndexWorkspaceSize
cnnlGetAsStridedBackwardWorkspaceSize
cnnlGetAtan2WorkspaceSize
cnnlGetAtomicsMode
cnnlGetBatch2spaceNdExtraInputSize
cnnlGetBatch2spaceWorkspaceSize
cnnlGetBatchMatMulExAlgoHeuristic
cnnlGetBatchMatMulExHeuristicResult
cnnlGetBatchNormBackwardWorkspaceSize
cnnlGetBatchNormForwardWorkspaceSize
cnnlGetBceLossBackwardWorkspaceSize
cnnlGetBceLossWorkspaceSize
cnnlGetBceWithLogitsBackwardWorkspaceSize
cnnlGetBceWithLogitsWorkspaceSize
cnnlGetBiasActivationGluBackwardWorkspaceSize
cnnlGetBiasAddBackwardWorkspaceSize
cnnlGetBiasAddWorkspaceSize
cnnlGetBiasDropoutAddFusedTrainBackwardWorkspaceSize
cnnlGetBiasDropoutAddFusedTrainWorkspaceSize
cnnlGetBitComputeWorkspaceSize
cnnlGetCTCLossDescriptor
cnnlGetCTCLossWorkspaceSize
cnnlGetClipGradNormExtraInputSize
cnnlGetClipGradNormWorkspaceSize_v2
cnnlGetClipWorkspaceSize
cnnlGetCol2ImWorkspaceSize
cnnlGetConcatWorkspaceSize_v2
cnnlGetContextParam
cnnlGetConvolutionBackwardDataAlgo
cnnlGetConvolutionBackwardDataWorkspaceSize
cnnlGetConvolutionBackwardFilterAlgo
cnnlGetConvolutionBackwardFilterWorkspaceSize
cnnlGetConvolutionForwardOutputDim
cnnlGetConvolutionForwardWorkspaceSize
cnnlGetCopySignWorkspaceSize
cnnlGetCopyWorkspaceSize
cnnlGetCosineSimilarityWorkspaceSize
cnnlGetCrossEntropyBackwardWorkspaceSize
cnnlGetCrossEntropyForwardWorkspaceSize
cnnlGetCrossWorkspaceSize
cnnlGetCumprodWorkspaceSize
cnnlGetCumsumWorkspaceSize
cnnlGetDCNBackwardWeightWorkspaceSize
cnnlGetDCNBakcwardDataWorkspaceSize
cnnlGetDCNForwardWorkspaceSize
cnnlGetDeconvolutionAlgo
cnnlGetDeconvolutionWorkspaceSize_v2
cnnlGetDetWorkspaceSize
cnnlGetDetectionOutputWorkspaceSize
cnnlGetDevice
cnnlGetDivNoNanWorkspaceSize
cnnlGetDivWorkspaceSize_v2
cnnlGetDlrmInteractWorkspaceSize
cnnlGetDynamicFloatQuantizeWorkspaceSize
cnnlGetDynamicPartitionWorkspaceSize
cnnlGetDynamicStitchWorkspaceSize_v2
cnnlGetEmbeddingBackwardWorkspaceSize
cnnlGetErrorString
cnnlGetFlashAttentionBackwardWorkspaceSize_v2
cnnlGetFlashAttentionForwardWorkspaceSize_v2
cnnlGetFlashAttentionGeneratedRandomNumbers
cnnlGetFlatSearchWorkspaceSize_v2
cnnlGetFloorDivWorkspaceSize
cnnlGetFloorModTruncWorkspaceSize
cnnlGetFloorModWorkspaceSize
cnnlGetForeachBinaryOpWorkspaceSize
cnnlGetForeachNormWorkspaceSize
cnnlGetForeachUnaryOpWorkspaceSize
cnnlGetFractionalMaxPoolForwardWorkspaceSize
cnnlGetFrozenBatchNormBackwardWorkspaceSize
cnnlGetFusedOpsConstParamPackAttribute
cnnlGetFusedOpsVariantParamPackAttribute
cnnlGetFwFFMBackwardWorkspaceSize
cnnlGetGatherWorkspaceSize
cnnlGetGenCaseDirectory
cnnlGetGridSampleBackwardWorkspaceSize
cnnlGetGridSampleForwardWorkspaceSize
cnnlGetGroupGemmAlgo
cnnlGetGroupGemmDescAttr
cnnlGetGroupGemmTensorDescriptor
cnnlGetGroupGemmWorkspaceSize
cnnlGetGroupNormBackwardWorkspaceSize_v2
cnnlGetGroupNormForwardWorkspaceSize
cnnlGetGruExtraInputSize
cnnlGetGruWorkspaceSize_v3
cnnlGetHistogramWorkspaceSize
cnnlGetIm2ColWorkspaceSize
cnnlGetIndexAddWorkspaceSize
cnnlGetIndexCopyWorkspaceSize_v2
cnnlGetIndexPutWorkspaceSize_v2
cnnlGetInstanceNormBackwardWorkspaceSize
cnnlGetInstanceNormForwardWorkspaceSize
cnnlGetInverseWorkspaceSize
cnnlGetIsinfWorkspaceSize
cnnlGetKthValueWorkspaceSize
cnnlGetL2LossWorkspaceSize
cnnlGetLLMQuantMatmulWorkspaceSize
cnnlGetLSTMGatesTempSize
cnnlGetLayerNormBackwardWorkspaceSize_v2
cnnlGetLayerNormOpWorkspaceSize
cnnlGetLerpWorkspaceSize
cnnlGetLibVersion
cnnlGetLogAddExp2WorkspaceSize
cnnlGetLogAddExpWorkspaceSize
cnnlGetLogicOpWorkspaceSize
cnnlGetLrnExtraInputSize_v2
cnnlGetLrnWorkspaceSize_v2
cnnlGetMSELossWorkspaceSize
cnnlGetMaskedWorkspaceSize
cnnlGetMatMulAlgoHeuristic
cnnlGetMatMulAlgoIds
cnnlGetMatMulDescAttr
cnnlGetMatMulExAlgo
cnnlGetMatMulExAlgoHeuristic
cnnlGetMatMulExAlgoIds
cnnlGetMatMulExHeuristicResult
cnnlGetMatMulExWorkspaceSize
cnnlGetMatMulHeuristicResult
cnnlGetMaximumWorkspaceSize
cnnlGetMedianWorkspaceSize
cnnlGetMinimumWorkspaceSize
cnnlGetNlllossWorkspaceSize
cnnlGetNmsWorkspaceSize_v4
cnnlGetNormalizeWorkspaceSize
cnnlGetNumTrueWorkspaceSize
cnnlGetNumTrueWorkspaceSize_v2
cnnlGetOpTensorDescriptor
cnnlGetOpTensorWorkspaceSize
cnnlGetOpTensorWorkspaceSize_v3
cnnlGetOrgqrWorkspaceSize
cnnlGetPackPaddedSequenceWorkspaceSize
cnnlGetPadPackedSequenceWorkspaceSize
cnnlGetPdistBackwardWorkspaceSize
cnnlGetPdistForwardExtraInputSize
cnnlGetPdistForwardWorkspaceSize
cnnlGetPolarWorkspaceSize
cnnlGetPolygammaWorkspaceSize
cnnlGetPooling2dDescriptor
cnnlGetPoolingBackwardWorkspaceSize
cnnlGetPoolingExtraInputSize
cnnlGetPoolingIndexWorkspaceSize
cnnlGetPoolingWithIndexWorkspaceSize_v2
cnnlGetPoolingWorkspaceSize_v2
cnnlGetPowWorkspaceSize
cnnlGetPqSearchWorkspaceSize
cnnlGetPreluBackwardV2WorkspaceSize
cnnlGetProposalFpnWorkspaceSize_v2
cnnlGetProposalWorkspaceSize
cnnlGetQRWorkspaceSize
cnnlGetQuantizeExDescriptor
cnnlGetQuantizeExDescriptorQuantSchemeAndDtype
cnnlGetQuantizeExDescriptorScalarQuant
cnnlGetQuantizeExDescriptor_v2
cnnlGetQuantizeParamWorkspaceSize
cnnlGetQuantizeRoundMode
cnnlGetQueue
cnnlGetRNNBiasMode
cnnlGetRNNClip
cnnlGetRNNComputationPreference
cnnlGetRNNDescriptor
cnnlGetRNNDescriptor_v2
cnnlGetRNNExtraInputSize
cnnlGetRNNMaskMode
cnnlGetRNNMathPrec
cnnlGetRNNOutputMode
cnnlGetRNNPaddingMode
cnnlGetRNNPeepholeMode
cnnlGetRNNProjectionLayers
cnnlGetRNNTLossDescriptor
cnnlGetRNNTLossWorkspaceSize
cnnlGetRNNTempSizes
cnnlGetRNNWeightOrder
cnnlGetRNNWeightParams
cnnlGetRNNWeightSpaceSize
cnnlGetRNNWorkspaceSize
cnnlGetRandGenerateMultinomialWorkspaceSize
cnnlGetRandSimulateThreadNum_v2
cnnlGetReduceOpWorkspaceSize_v2
cnnlGetReorgExtraInputSize
cnnlGetReorgWorkspaceSize
cnnlGetReservedMemSize
cnnlGetRmsNormBackwardWorkspaceSize
cnnlGetRmsNormOpWorkspaceSize
cnnlGetRoialignWorkspaceSize
cnnlGetRollWorkspaceSize
cnnlGetRreluWithNoiseWorkspaceSize
cnnlGetScaledDotProductAttnWorkspaceSize_v3
cnnlGetScatterWorkspaceSize
cnnlGetSelectV2WorkspaceSize
cnnlGetSeqDataDescriptor
cnnlGetSeqDataDescriptorOnchipDataType
cnnlGetSeqDataDescriptorPositionAndScale
cnnlGetSeqDataDescriptor_v2
cnnlGetShufflechannelWorkspaceSize
cnnlGetSingleQueryCachedKVAttnWorkspaceSize_v2
cnnlGetSizeOfDataType
cnnlGetSmoothL1LossBackwardWorkspaceSize
cnnlGetSmoothL1LossForwardWorkspaceSize
cnnlGetSortPairsWorkspaceSize
cnnlGetSpace2batchNdExtraInputSize
cnnlGetSpace2batchWorkspaceSize
cnnlGetSparseDenseMatMulDescAttr
cnnlGetSparseDenseMatMulWorkspaceSize
cnnlGetSparseTensorDescAttr
cnnlGetSparseTensorDescriptor
cnnlGetSplitWorkspaceSize
cnnlGetSquaredDifferenceWorkspaceSize
cnnlGetStaticFloatQuantizeWorkspaceSize
cnnlGetStdVarMeanWorkspaceSize
cnnlGetStrideBatchMatMulAlgoHeuristic_v2
cnnlGetStrideBatchMatMulDescAttr
cnnlGetStrideBatchMatMulHeuristicResult
cnnlGetSvdWorkspaceSize
cnnlGetSyncBatchNormStatsWorkspaceSize
cnnlGetSyncBatchnormBackwardReduceWorkspaceSize
cnnlGetTensorAndDataFromTensorSet
cnnlGetTensorDescriptor
cnnlGetTensorDescriptorEx
cnnlGetTensorDescriptorEx_v2
cnnlGetTensorDescriptorPointerMode
cnnlGetTensorDescriptor_v2
cnnlGetTensorElementNum
cnnlGetTensorSetDescriptor
cnnlGetTensorSetDescriptorSize
cnnlGetTopKTensorWorkspaceSize
cnnlGetTraceWorkspaceSize
cnnlGetTransformWorkspaceSize
cnnlGetTransformerAttentionCacheStrategy
cnnlGetTransformerAttentionWorkspaceSize_v2
cnnlGetTransformerAttnProjDescriptor
cnnlGetTransformerAttnProjQuantifyDescriptor
cnnlGetTransformerAttnProjWorkspaceSize
cnnlGetTransformerBeamRearrangeWorkspaceSize
cnnlGetTransformerFFNFilterDescriptor
cnnlGetTransformerFFNWorkspaceSize
cnnlGetTransformerFcTopkWorkspaceSize
cnnlGetTransformerFeedForwardWorkspaceSize_v3
cnnlGetTransformerSelfAttnWorkspaceSize
cnnlGetTransposeWorkspaceSize
cnnlGetTreeEnsembleWorkspaceSize
cnnlGetUniqueConsecutiveWorkspaceSize
cnnlGetUniqueWorkspaceSize
cnnlGetUnsortedSegmentSumWorkspaceSize_v2
cnnlGetWhereWorkspaceSize
cnnlGetWindowAttentionWorkspaceSize
cnnlGetXlogyWorkspaceSize
cnnlGradientDescent
cnnlGrep
cnnlGridSampleBackward
cnnlGridSampleForward
cnnlGroupGemm
cnnlGroupNormBackward
cnnlGroupNormForward_v3
cnnlGroupQuant_v2
cnnlGru_v2
cnnlHardtanh
cnnlHardtanhBackward
cnnlHistc
cnnlIm2Col
cnnlInTopK
cnnlIndexAdd_v2
cnnlIndexCopy
cnnlIndexFill_v3
cnnlIndexPut_v2
cnnlIndexSelect
cnnlInitBatch2spaceNdExtraInput
cnnlInitClipGradNormExtraInput
cnnlInitGRUWeightPositionAndScale
cnnlInitLrnExtraInput
cnnlInitPdistForwardExtraInput
cnnlInitPoolingExtraInput
cnnlInitRNNWeightPositionAndScale
cnnlInitReorgExtraInput
cnnlInitSpace2batchNdExtraInput
cnnlInitTensorSetMemberDescriptor
cnnlInitTensorSetMemberDescriptor_v2
cnnlInitialMatMulAlgo
cnnlInitialMatMulExAlgo
cnnlInstanceNormBackward
cnnlInstanceNormForward
cnnlInterpBackward_v3
cnnlInterp_v3
cnnlInverse_v2
cnnlInvertPermutation
cnnlIsFinite
cnnlIsInf
cnnlIsNan
cnnlKerasMomentum
cnnlKthValue_v2
cnnlL1LossBackward
cnnlL2Loss_v2
cnnlLLMQuantMatmul
cnnlLLMQuantMatmul_v2
cnnlLSTMGatesBackward
cnnlLSTMGatesForward
cnnlLayerNormBackward_v3
cnnlLayerNormForward_v2
cnnlLerp
cnnlLinspace
cnnlLog
cnnlLog1p_v2
cnnlLogAddExp
cnnlLogAddExp2
cnnlLogicOp
cnnlLogicOpNot
cnnlLrnGrad
cnnlLrn_v2
cnnlMSELossBackward
cnnlMSELoss_v2
cnnlMakeFusedOpsPlan
cnnlMaskedScaleSoftmaxBackward
cnnlMaskedSoftmax
cnnlMasked_v5
cnnlMatMulEx_v2
cnnlMatMul_v2
cnnlMatrixBandPart
cnnlMatrixDiag
cnnlMaximum
cnnlMedian_v2
cnnlMinimum
cnnlMomentum
cnnlMulN
cnnlMultiTensorScale
cnnlNanInf
cnnlNanToNum
cnnlNegTensor
cnnlNlllossBackward_v2
cnnlNlllossForward_v2
cnnlNms_v2
cnnlNormalize_v3
cnnlNumTrue_v3
cnnlOneHot
cnnlOpTensor
cnnlOrgqr
cnnlPackPaddedSequence
cnnlPad
cnnlPadPackedSequence
cnnlPdistBackward
cnnlPdistForward
cnnlPolar
cnnlPolygamma
cnnlPoolingBackward_v2
cnnlPoolingForwardWithIndex
cnnlPoolingForward_v2
cnnlPoolingIndex_v2
cnnlPow_v2
cnnlPow_v3
cnnlPqAdd
cnnlPqRemove
cnnlPqSearch_v2
cnnlPrelu
cnnlPreluBackwardV2
cnnlProposalFpn
cnnlProposal_v2
cnnlQR
cnnlQuantize
cnnlQuantizeParam_v2
cnnlRMSProp
cnnlRNNBackward
cnnlRNNBackwardData
cnnlRNNBackwardWeights
cnnlRNNForwardInference_v2
cnnlRNNForwardTraining
cnnlRNNTLoss
cnnlReciprocal
cnnlReduce_v2
cnnlReflectionPad2d
cnnlReflectionPadBackward
cnnlReformatTransformerFFNFilter
cnnlRelPositionMultiHeadAttention_v2
cnnlReorg_v2
cnnlRepeatInterleave
cnnlReplicationPad2d
cnnlReplicationPadBackward
cnnlResetTensorDescriptor
cnnlRmsNormBackward_v2
cnnlRmsNormForward_v2
cnnlRoiAlignBackward_v2
cnnlRoiAlign_v2
cnnlRoiPoolingBackward
cnnlRoiPoolingForward
cnnlRoialign
cnnlRoll_v2
cnnlRotaryEmbedding_v2
cnnlRound
cnnlRreluWithNoise
cnnlRsqrt
cnnlRsqrtBackward
cnnlScaledDotProductAttn_v5
cnnlScaledMaskedSoftmaxBackward
cnnlScatterNd_v2
cnnlScatterRef
cnnlScatter_v2
cnnlSearchSorted
cnnlSelect
cnnlSelectV2
cnnlSetActivationDescAttr
cnnlSetAtomicsMode
cnnlSetBatchMatMulExBiasActive
cnnlSetBertPreDescriptor
cnnlSetBiasActivationGluDescriptor
cnnlSetBiasDropoutAddFusedTrainBackwardDescriptor
cnnlSetBiasDropoutAddFusedTrainDescriptor
cnnlSetCTCLossDescriptor
cnnlSetConvolutionDescriptor
cnnlSetConvolutionDescriptorAlgoSearchMode
cnnlSetConvolutionDescriptorAllowTF32
cnnlSetConvolutionDescriptorQuant
cnnlSetDCNDescriptor
cnnlSetDeconvolutionDescriptor
cnnlSetDeconvolutionDescriptorAllowTF32
cnnlSetDeconvolutionDescriptorQuant
cnnlSetDetectionOutputComputePreference
cnnlSetDivDescAttr
cnnlSetDlrmInteractDescriptor
cnnlSetEmbeddingBagDescriptor_v2
cnnlSetFasterRCNNDetectionOutputDescriptor
cnnlSetFlashAttentionBackwardDescriptor_v2
cnnlSetFlashAttentionBackwardDeterminismMode
cnnlSetFlashAttentionDescriptor_v2
cnnlSetFlashAttentionDropoutMaskMode
cnnlSetFlashAttentionGlobalAttentionSize
cnnlSetFlashAttentionSlidingWindowSize
cnnlSetFlatSearchDescriptor_v2
cnnlSetFloatQuantizeDescriptorBlockSize
cnnlSetFusedOpsConstParamPackAttribute
cnnlSetFusedOpsVariantParamPackAttribute
cnnlSetGenCaseDirectory
cnnlSetGenCaseMode
cnnlSetGrepDescriptor_v4
cnnlSetGridSampleDescriptor
cnnlSetGroupGemmAlgo
cnnlSetGroupGemmDescAttr
cnnlSetGroupGemmGroupMixedQuantBitFlag
cnnlSetGroupGemmGroupwiseScale
cnnlSetGroupGemmPerRowColScaleBiasAct
cnnlSetGroupGemmTensorDescriptor
cnnlSetGroupGemmTensorGatherIdxDesc
cnnlSetGroupTensorDescriptors
cnnlSetGroupTensorDescriptors_v2
cnnlSetGruDescriptor_v2
cnnlSetHistogramDescriptorBinCountMode
cnnlSetHistogramDescriptorHistoCountMode
cnnlSetIndexPutDescriptor
cnnlSetInterpDescriptorEx
cnnlSetInterpDescriptor_v2
cnnlSetLLMQuantMatmulDescriptor_v2
cnnlSetMatMulDescAttr
cnnlSetMatMulExAlgo
cnnlSetMatMulExBias
cnnlSetMatMulExBiasScaleBNActive
cnnlSetMatMulExDescAttr
cnnlSetMatMulExDescAttrBase
cnnlSetMultiLevelRoialignDescriptor
cnnlSetNmsDescAttr
cnnlSetNormDescAttr
cnnlSetNormalizeDescriptor_v2
cnnlSetOpTensorDescriptor
cnnlSetPooling2dDescriptor_v2
cnnlSetPoolingDescriptorQuant
cnnlSetPoolingNdDescriptor_v2
cnnlSetPpyoloDetectionOutputDescriptor
cnnlSetProposalDescriptor
cnnlSetProposalFpnDescriptor
cnnlSetQuantizeExDescriptor
cnnlSetQuantizeExDescriptorQuantSchemeAndDtype
cnnlSetQuantizeExDescriptorScalarQuant
cnnlSetQuantizeExDescriptor_v2
cnnlSetQuantizeRoundMode
cnnlSetQueue
cnnlSetRNNBiasMode
cnnlSetRNNClip
cnnlSetRNNComputationPreference
cnnlSetRNNDescriptor
cnnlSetRNNDescriptorQuant
cnnlSetRNNDescriptorWeightQuant
cnnlSetRNNDescriptor_v2
cnnlSetRNNInferenceDescriptor
cnnlSetRNNMaskMode
cnnlSetRNNMathPrec
cnnlSetRNNOutputMode
cnnlSetRNNPaddingMode
cnnlSetRNNPeepholeMode
cnnlSetRNNProjectionLayers
cnnlSetRNNTLossDescriptor
cnnlSetRNNWeightOrder
cnnlSetReduceDescriptor_v2
cnnlSetReduceDescriptor_v3
cnnlSetRefinedetDetectionOutputDescriptor_v2
cnnlSetRelPositionMultiHeadAttentionDescriptor
cnnlSetRelPositionMultiHeadAttentionDescriptorAllowTF32
cnnlSetReorgDescriptor
cnnlSetRetinaDetectionOutputDescriptor
cnnlSetRoiAlignDescriptor_v2
cnnlSetRotaryEmbeddingDescriptor_v2
cnnlSetRotaryEmbeddingDescriptor_v3
cnnlSetSeqDataDescriptor
cnnlSetSeqDataDescriptorOnchipDataType
cnnlSetSeqDataDescriptorPositionAndScale
cnnlSetSeqDataDescriptor_v2
cnnlSetSpaceBatchNdDescriptor
cnnlSetSparseDenseMatMulDescAttr
cnnlSetSparseTensorDescAttr
cnnlSetSsdDetectionOutputDescriptor_v2
cnnlSetStdVarMeanDescriptor_v2
cnnlSetStrideBatchMatMulDescAttr
cnnlSetTensorDescriptor
cnnlSetTensorDescriptorDim
cnnlSetTensorDescriptorDim_v2
cnnlSetTensorDescriptorEx
cnnlSetTensorDescriptorEx_v2
cnnlSetTensorDescriptorPointerMode
cnnlSetTensorDescriptor_v2
cnnlSetTransformerAttentionDescriptorAllowTF32
cnnlSetTransformerAttentionDescriptor_v2
cnnlSetTransformerAttentionQuantizeDescriptor
cnnlSetTransformerAttentionSeqDataLayout
cnnlSetTransformerAttnProjDescriptor
cnnlSetTransformerAttnProjDescriptorAllowTF32
cnnlSetTransformerAttnProjQuantizeDescriptor_v3
cnnlSetTransformerBeamSearchDescriptor_v2
cnnlSetTransformerFFNDescriptorAllowTF32
cnnlSetTransformerFFNDescriptorInnerLayernormalMode
cnnlSetTransformerFFNDescriptorNormType
cnnlSetTransformerFFNDescriptor_v2
cnnlSetTransformerFFNPrecisionMode
cnnlSetTransformerFFNQATDescriptor
cnnlSetTransformerFFNQATQuantifyParams_v2
cnnlSetTransformerFeedForwardDescriptorAllowTF32
cnnlSetTransformerFeedForwardDescriptor_v2
cnnlSetTransformerFeedForwardQuantizeDescriptor_v4
cnnlSetTransformerSelfAttnComputeType
cnnlSetTransformerSelfAttnDescriptor_v2
cnnlSetTransformerSelfAttnEncoderKeyValueOutputMode
cnnlSetTransformerSelfAttnFactorPosition
cnnlSetTransformerSelfAttnPackModeMaxToken
cnnlSetTransformerSelfAttnQATParameters
cnnlSetTransformerSelfAttnQuantizeParams
cnnlSetTransformerSelfAttnSoftmaxOutPosAndScale
cnnlSetTransposeDescriptor
cnnlSetTrigonDescriptor_v2
cnnlSetUniqueConsecutiveDescriptor
cnnlSetUniqueDescriptor
cnnlSetWindowAttentionDescriptor_v2
cnnlSetYolov2DetectionOutputDescriptor
cnnlSetYolov3DetectionOutputDescriptor_v2
cnnlSetYolov4DetectionOutputDescriptor
cnnlSetYolov5DetectionOutputDescriptor
cnnlSetYolov8DetectionOutputDescriptor
cnnlSetYoloxDetectionOutputDescriptor
cnnlShuffleChannel
cnnlSign
cnnlSin
cnnlSin_v2
cnnlSingleQueryCachedKVAttn_v3
cnnlSingleQueryCachedKVAttn_v4
cnnlSingleQueryCachedKVAttn_v5
cnnlSmoothL1LossBackward_v2
cnnlSmoothL1LossForward_v2
cnnlSoftmaxBackward
cnnlSoftmaxCrossEntropyWithLogits
cnnlSoftmaxForward
cnnlSoftplusBackward_v2
cnnlSoftplusForward_v2
cnnlSoftsignForward
cnnlSoftsignGrad
cnnlSortPairs
cnnlSortedSegmentReduce
cnnlSpace2batch
cnnlSpace2batchNd_v2
cnnlSparseDenseMatMul
cnnlSparseSoftmaxCrossEntropyWithLogits
cnnlSplit
cnnlSqrt
cnnlSqrtBackward
cnnlSqrt_v2
cnnlSquare
cnnlSquaredDifference
cnnlStaticFloatQuantize
cnnlStdVarMean
cnnlStrideBatchMatMul_v3
cnnlStridedSliceBackward
cnnlStridedSlice_v2
cnnlSvd
cnnlSyncBatchNormBackwardElemt
cnnlSyncBatchNormBackwardElemtV2
cnnlSyncBatchNormElemt
cnnlSyncBatchNormGatherStatsWithCounts
cnnlSyncBatchNormStats_v2
cnnlSyncBatchnormBackwardReduce_v2
cnnlThreshold
cnnlThresholdBackward
cnnlTile
cnnlTopKTensor_v3
cnnlTrace
cnnlTransform_v2
cnnlTransform_v3
cnnlTransformerAttention
cnnlTransformerAttnProj
cnnlTransformerBeamRearrange
cnnlTransformerBeamRearrangeWithSplitedCache
cnnlTransformerBeamSearch
cnnlTransformerFFN
cnnlTransformerFeedForward_v2
cnnlTransformerSelfAttn
cnnlTranspose_v2
cnnlTreeEnsemble
cnnlTriIndices
cnnlTri_v2
cnnlTrigonForward
cnnlTrunc
cnnlUnfold
cnnlUniqueConsecutive
cnnlUnique_v2
cnnlUnpoolBackward
cnnlUnpoolForward
cnnlUnsortedSegmentSum
cnnlUpdateContextInformation
cnnlWeightNorm
cnnlWeightNormBackward
cnnlWhere_v2
cnnlWindowAttention
cnnlXlogy
cnnlextraGetSupportCluster
cnnlextraGetSupportDevice
cnpapiActivityDisable
cnpapiActivityEnable
cnpapiActivityEnableLatencyTimestamps
cnpapiActivityFlushAll
cnpapiActivityFlushPeriod
cnpapiActivityGetNextRecord
cnpapiActivityPopExternalCorrelationId
cnpapiActivityPushExternalCorrelationId
cnpapiActivityRegisterCallbacks
cnpapiCheckpointCreate
cnpapiCheckpointFree
cnpapiCheckpointRestore
cnpapiCheckpointSave
cnpapiEnableAllDomains
cnpapiEnableCallback
cnpapiEnableDomain
cnpapiEndPass
cnpapiGetCNtaskTopoEntityId
cnpapiGetCNtaskTopoId
cnpapiGetCNtaskTopoNodeId
cnpapiGetCallbackName
cnpapiGetCallbackState
cnpapiGetContextId
cnpapiGetDeviceCount
cnpapiGetDeviceId
cnpapiGetKernelTypeFromCNkernel
cnpapiGetLastError
cnpapiGetLibVersion
cnpapiGetNameFromCnpxDomainHandle
cnpapiGetPciBusId
cnpapiGetQueueId
cnpapiGetResultString
cnpapiGetSymbolNameFromCNkernel
cnpapiGetTimestamp
cnpapiInit
cnpapiPmuBeginPass
cnpapiPmuConvertCounterNameToCounterEvalRequest
cnpapiPmuCounterDataImageReaderCreate
cnpapiPmuCounterDataImageReaderDestroy
cnpapiPmuCounterEvaluatorCreate
cnpapiPmuCounterEvaluatorDestroy
cnpapiPmuCounterReaderCreate
cnpapiPmuCounterReaderDestroy
cnpapiPmuCounterReaderGetName
cnpapiPmuCreateConfigImage
cnpapiPmuCreateSession
cnpapiPmuCreateSessionWithDeviceId
cnpapiPmuDestroySession
cnpapiPmuDisableProfiling
cnpapiPmuEnableCounter
cnpapiPmuEnableProfiling
cnpapiPmuFlushData
cnpapiPmuGetCorrelationId
cnpapiPmuGetCounterIdByName
cnpapiPmuGetCounterName
cnpapiPmuGetCounterNums
cnpapiPmuGetCounterSupported
cnpapiPmuGetCounterValue
cnpapiPmuGetCounterValueFromImage
cnpapiPmuGetNumPasses
cnpapiPmuGetPciBusIdFromImage
cnpapiPmuGetRangeNums
cnpapiPmuGetTaskTopoId
cnpapiPmuInit
cnpapiPmuPeriodicSamplerGetTimestamp
cnpapiPmuPopRange
cnpapiPmuPushRange
cnpapiPmuRangeReaderCreate
cnpapiPmuRangeReaderDestroy
cnpapiPmuRangeReaderGetName
cnpapiPmuRelease
cnpapiPmuSetConfig
cnpapiPmuSetCurrentSession
cnpapiPmuSetFlushMode
cnpapiPmuStartSampling
cnpapiPmuStopSampling
cnpapiPmuUnsetConfig
cnpapiRelease
cnpapiSubscribe
cnpapiSupportedDomains
cnpapiUnsubscribe
cnperfConfigCreate
cnperfConfigDestroy
cnperfConfigEnable
cnperfConfigGet
cnperfConfigSet
cnperfGetDeviceTaskLaunchFunctions
cnperfGetLibVersion
cnperfGetTimestamp
cnperfInit
cnperfParserCreateFromPath
cnperfParserCreateFromSession
cnperfParserDestroy
cnperfParserGetCommDataInOpRange
cnperfParserGetData
cnperfParserGetKernelHighRatePmuData
cnperfParserGetKernelPmuData
cnperfParserGetOpRanges
cnperfParserGetSamplingPmuData
cnperfParserGetSamplingPmuDataByName
cnperfParserGetStartTimestamp
cnperfParserGetTaskTopoNodeOpRanges
cnperfRelease
cnperfSessionDestroy
cnperfSessionGetResultPath
cnperfSessionRecordData
cnperfSetBaseDir
cnperfSetLogLevel
cnperfStart
cnperfStop
cnpxDomainCreate
cnpxDomainDestroy
cnpxDomainMark
cnpxDomainRangeEnd
cnpxDomainRangePop
cnpxDomainRangePush
cnpxDomainRangeStart
cnpxDomainRegisterString
cnpxInit
cnpxInitApis
cnpxMark
cnpxMarkPayload
cnpxNameCNcontext
cnpxNameCNdev
cnpxNameCNnotifier
cnpxNameCNqueue
cnpxNameOsThread
cnpxNameQueue
cnpxPayloadEnumRegister
cnpxPayloadSchemaRegister
cnpxRangeEnd
cnpxRangeEndPayload
cnpxRangePop
cnpxRangePopPayload
cnpxRangePush
cnpxRangePushPayload
cnpxRangeStart
cnpxRangeStartPayload
cnrtAcquireMemHandle
cnrtCastDataType_V2
cnrtCreateQuantizedParam
cnrtCreateQuantizedParamByChannel
cnrtDestroyQuantizedParam
cnrtDeviceGetAttribute
cnrtDeviceGetByPCIBusId
cnrtDeviceGetConfig
cnrtDeviceGetDefaultMemPool
cnrtDeviceGetMemPool
cnrtDeviceGetPCIBusId
cnrtDeviceGetQueuePriorityRange
cnrtDeviceQueryKernelMemoryUsage
cnrtDeviceReset
cnrtDeviceResetPersistingL2Cache
cnrtDeviceSetConfig
cnrtDeviceSetMemPool
cnrtDriverGetVersion
cnrtFree
cnrtFreeHost
cnrtGetDevice
cnrtGetDeviceCount
cnrtGetDeviceFlag
cnrtGetDeviceProperties
cnrtGetDeviceProperties_V2
cnrtGetDeviceProperties_V3
cnrtGetErrorName
cnrtGetErrorStr
cnrtGetLastError
cnrtGetLibVersion
cnrtGetPeerAccessibility
cnrtGetSymbolAddress
cnrtGetSymbolSize
cnrtHostMalloc
cnrtInvokeHostFunc
cnrtInvokeKernel
cnrtIpcGetNotifierHandle
cnrtIpcOpenNotifierHandle
cnrtMalloc
cnrtMallocConstant
cnrtMapMemHandle
cnrtMcacheOperation
cnrtMemAllocAsync
cnrtMemAllocFromPoolAsync
cnrtMemFreeAsync
cnrtMemGetInfo
cnrtMemPoolCreate
cnrtMemPoolDestroy
cnrtMemPoolGetAccess
cnrtMemPoolGetAttribute
cnrtMemPoolSetAttribute
cnrtMemPoolTrimTo
cnrtMemcpy
cnrtMemcpy2D
cnrtMemcpy2DAsync
cnrtMemcpy3D
cnrtMemcpy3DAsync
cnrtMemcpyAsync
cnrtMemcpyAsync_V2
cnrtMemcpyAsync_V3
cnrtMemcpyFromSymbol
cnrtMemcpyFromSymbolAsync
cnrtMemcpyFromSymbolAsync_V2
cnrtMemcpyPeer
cnrtMemcpyPeerAsync
cnrtMemcpyToSymbol
cnrtMemcpyToSymbolAsync
cnrtMemcpyToSymbolAsync_V2
cnrtMemset
cnrtMemsetAsync
cnrtMmap
cnrtMmapCached
cnrtMunmap
cnrtNotifierCreate
cnrtNotifierCreateWithFlags
cnrtNotifierDestroy
cnrtNotifierDuration
cnrtNotifierElapsedTime
cnrtPeekAtLastError
cnrtPlaceNotifier
cnrtPlaceNotifierWithFlags
cnrtPointerGetAttributes
cnrtProfilerStart
cnrtProfilerStop
cnrtQueryNotifier
cnrtQueueBeginCapture
cnrtQueueCopyAttributes
cnrtQueueCreate
cnrtQueueCreateWithPriority
cnrtQueueDestroy
cnrtQueueEndCapture
cnrtQueueGetAttribute
cnrtQueueGetCaptureInfo
cnrtQueueGetId
cnrtQueueGetPriority
cnrtQueueIsCapturing
cnrtQueueQuery
cnrtQueueSetAttribute
cnrtQueueSync
cnrtQueueUpdateCaptureDependencies
cnrtQueueWaitNotifier
cnrtSetDevice
cnrtSetDeviceFlag
cnrtSyncDevice
cnrtTaskTopoAcquireUserObject
cnrtTaskTopoAddChildTopoNode
cnrtTaskTopoAddDependencies
cnrtTaskTopoAddEmptyNode
cnrtTaskTopoAddHostNode
cnrtTaskTopoAddKernelNode
cnrtTaskTopoAddMemcpyNode
cnrtTaskTopoAddMemsetNode
cnrtTaskTopoAddNotifierPlaceNode
cnrtTaskTopoAddNotifierWaitNode
cnrtTaskTopoChildTopoNodeGetTopo
cnrtTaskTopoClone
cnrtTaskTopoCreate
cnrtTaskTopoDebugDotPrint
cnrtTaskTopoDestroy
cnrtTaskTopoDestroyNode
cnrtTaskTopoEntityChildTopoNodeSetParams
cnrtTaskTopoEntityDestroy
cnrtTaskTopoEntityHostNodeSetParams
cnrtTaskTopoEntityInvoke
cnrtTaskTopoEntityKernelNodeSetParams
cnrtTaskTopoEntityMemcpyNodeSetParams
cnrtTaskTopoEntityMemsetNodeSetParams
cnrtTaskTopoEntityNotifierPlaceNodeSetNotifier
cnrtTaskTopoEntityNotifierWaitNodeSetNotifier
cnrtTaskTopoEntityUpdate
cnrtTaskTopoGetEdges
cnrtTaskTopoGetNodes
cnrtTaskTopoGetRootNodes
cnrtTaskTopoHostNodeGetParams
cnrtTaskTopoHostNodeSetParams
cnrtTaskTopoInstantiate
cnrtTaskTopoKernelNodeCopyAttributes
cnrtTaskTopoKernelNodeGetAttribute
cnrtTaskTopoKernelNodeGetParams
cnrtTaskTopoKernelNodeSetAttribute
cnrtTaskTopoKernelNodeSetParams
cnrtTaskTopoMemcpyNodeGetParams
cnrtTaskTopoMemcpyNodeSetParams
cnrtTaskTopoMemsetNodeGetParams
cnrtTaskTopoMemsetNodeSetParams
cnrtTaskTopoNodeFindInClone
cnrtTaskTopoNodeGetDependencies
cnrtTaskTopoNodeGetDependentNodes
cnrtTaskTopoNodeGetType
cnrtTaskTopoNotifierPlaceNodeGetNotifier
cnrtTaskTopoNotifierPlaceNodeSetNotifier
cnrtTaskTopoNotifierWaitNodeGetNotifier
cnrtTaskTopoNotifierWaitNodeSetNotifier
cnrtTaskTopoReleaseUserObject
cnrtTaskTopoRemoveDependencies
cnrtTaskTopoUpload
cnrtThreadExchangeQueueCaptureMode
cnrtUnMapMemHandle
cnrtUserObjectAcquire
cnrtUserObjectCreate
cnrtUserObjectRelease
cnrtWaitNotifier
cnrtcCompileCode
cnrtcCreateCode
cnrtcCreateCodeV2
cnrtcDestroyCode
cnrtcGetCompilationLog
cnrtcGetCompilationLogSize
cnrtcGetCompilationOutput
cnrtcGetCompilationOutputSize
cnrtcGetFatBinary
cnrtcGetFatBinarySize
cnrtcTransStatusToString
cnrtcVersion
cntopoAddMachineInfo
cntopoClearMachineInfo
cntopoCreateQuery
cntopoDestroyContext
cntopoDestroyQuery
cntopoFindDevSets
cntopoFindTopos
cntopoGetDevInfoFromDevSet
cntopoGetDevSetSize
cntopoGetErrorStr
cntopoGetLibVersion
cntopoGetLocalMachineInfo
cntopoGetNodeFromTopo
cntopoInitContext
cntopoLoadMachineInfoFromFile
cntopoSaveMachineInfoToFile
cntopoSetBlacklistDevOrdinal
cntopoSetBlacklistUUID
cntopoSetDevNumFilter
cntopoSetWhitelistDevOrdinal
cntopoSetWhitelistUUID
dcmmActionValidate
dcmmConfigGet
dcmmConfigSet
dcmmConnect
dcmmDisconnect
dcmmEntitiesGetLatestValues
dcmmEntityGetLatestValues
dcmmFieldGetById
dcmmFieldGroupCreate
dcmmFieldGroupDestroy
dcmmFieldGroupGetAll
dcmmFieldGroupGetInfo
dcmmGetAllDevices
dcmmGetAllSupportedDevices
dcmmGetDeviceTopology
dcmmGetEntityGroupEntities
dcmmGetErrorString
dcmmGetFieldValuesSince
dcmmGetGroupTopology
dcmmGetLatestValues
dcmmGetMluLinkLinkStatus
dcmmGetServerVersionInfo
dcmmGetValuesSince
dcmmGetVersionInfo
dcmmGroupAddEntity
dcmmGroupCreate
dcmmGroupDestroy
dcmmGroupGetAllIds
dcmmGroupGetInfo
dcmmGroupRemoveEntity
dcmmHealthCheck
dcmmHealthGet
dcmmHealthSet
dcmmInit
dcmmInjectFieldValue
dcmmJobGetStats
dcmmJobRemove
dcmmJobRemoveAll
dcmmJobStartStats
dcmmJobStopStats
dcmmPolicyGet
dcmmPolicyRegister
dcmmPolicySet
dcmmPolicyUnregister
dcmmRelease
dcmmRunDiagnostic
dcmmSelectMlusByTopology
dcmmSetEntityMluLinkLinkState
dcmmStartServer
dcmmStopDiagnostic
dcmmStopServer
dcmmUnwatchFields
dcmmUpdateAllFields
dcmmWatchFields
dcmmWatchJobFields
getBuffer
getBufferList
getCncvHandle
getCnrtQueue
getCounterSupported
mluOpAbs
mluOpActiveRotatedFilterForward
mluOpAdamW
mluOpBallQuery
mluOpBboxOverlaps
mluOpBorderAlignBackward
mluOpBorderAlignForward
mluOpBoxIouRotated
mluOpCarafeBackward
mluOpCarafeForward
mluOpCreate
mluOpCreateAdamWDescriptor
mluOpCreateCarafeDescriptor
mluOpCreateDCNDescriptor
mluOpCreateFFTPlan
mluOpCreateGroupTensorDescriptors
mluOpCreateNmsDescriptor
mluOpCreateRoiAlignForwardDescriptor
mluOpCreateSeqDataDescriptor
mluOpCreateSparseConvolutionDescriptor
mluOpCreateTensorDescriptor
mluOpCreateTensorSetDescriptor
mluOpDCNBackwardData
mluOpDCNBackwardWeight
mluOpDCNForward
mluOpDeformRoiPoolBackward
mluOpDeformRoiPoolForward
mluOpDestroy
mluOpDestroyAdamWDescriptor
mluOpDestroyCarafeDescriptor
mluOpDestroyDCNDescriptor
mluOpDestroyFFTPlan
mluOpDestroyGroupTensorDescriptors
mluOpDestroyNmsDescriptor
mluOpDestroyRoiAlignForwardDescriptor
mluOpDestroySeqDataDescriptor
mluOpDestroySparseConvolutionDescriptor
mluOpDestroyTensorDescriptor
mluOpDestroyTensorSetDescriptor
mluOpDiffIouRotatedSortVerticesForward
mluOpDiv
mluOpDynamicPointToVoxelBackward
mluOpDynamicPointToVoxelForward
mluOpExecFFT
mluOpFocalLossSigmoidBackward
mluOpFocalLossSigmoidForward
mluOpGenerateProposalsV2
mluOpGetActiveRotatedFilterForwardWorkspaceSize
mluOpGetAtomicsMode
mluOpGetDCNBackwardWeightWorkspaceSize
mluOpGetDCNBakcwardDataWorkspaceSize
mluOpGetDCNForwardWorkspaceSize
mluOpGetDynamicPointToVoxelBackwardWorkspaceSize
mluOpGetDynamicPointToVoxelForwardWorkspaceSize
mluOpGetErrorString
mluOpGetGenCaseDirectory
mluOpGetGenerateProposalsV2WorkspaceSize
mluOpGetGenerateProposalsV2WorkspaceSize_v2
mluOpGetIndiceConvolutionBackwardDataWorkspaceSize
mluOpGetIndiceConvolutionBackwardFilterWorkspaceSize
mluOpGetIndiceConvolutionForwardWorkspaceSize
mluOpGetIndicePairs
mluOpGetIndicePairsWorkspaceSize
mluOpGetLibVersion
mluOpGetMaskedCol2imForwardWorkspaceSize
mluOpGetMaskedIm2colForwardWorkspaceSize
mluOpGetMoeDispatchBackwardGateWorkspaceSize
mluOpGetMutualInformationBackwardWorkspaceSize
mluOpGetMutualInformationForwardWorkspaceSize
mluOpGetNmsRotatedWorkspaceSize
mluOpGetNmsWorkspaceSize
mluOpGetPolyNmsWorkspaceSize
mluOpGetQuantizeRoundMode
mluOpGetQueue
mluOpGetRoiAwarePool3dForwardWorkspaceSize
mluOpGetRoiPointPool3dWorkspaceSize
mluOpGetRoiawarePool3dForwardWorkspaceSize
mluOpGetSeqDataDescriptor_v2
mluOpGetSizeOfDataType
mluOpGetSparseConvolutionNumActOut
mluOpGetSyncBatchNormBackwardReduceWorkspaceSize
mluOpGetSyncBatchNormStatsWorkspaceSize
mluOpGetSyncBatchnormBackwardReduceWorkspaceSize
mluOpGetTensorAndDataFromTensorSet
mluOpGetTensorDescriptor
mluOpGetTensorDescriptorEx
mluOpGetTensorDescriptorEx_v2
mluOpGetTensorDescriptorOnchipDataType
mluOpGetTensorDescriptorPointerMode
mluOpGetTensorDescriptorPosition
mluOpGetTensorDescriptorPositionAndScale
mluOpGetTensorDescriptorPositionScaleAndOffset
mluOpGetTensorDescriptor_v2
mluOpGetTensorElementNum
mluOpGetTensorSetDescriptor
mluOpGetTensorSetDescriptorSize
mluOpGetThreeNNForwardWorkspaceSize
mluOpGetVoxelizationWorkspaceSize
mluOpIndiceConvolutionBackwardData
mluOpIndiceConvolutionBackwardFilter
mluOpIndiceConvolutionForward
mluOpInitTensorSetMemberDescriptor
mluOpInitTensorSetMemberDescriptorPositionAndScale
mluOpLgamma
mluOpLog
mluOpLogspace
mluOpMakeFFTPlanMany
mluOpMaskedCol2imForward
mluOpMaskedIm2colForward
mluOpMoeDispatchBackwardData
mluOpMoeDispatchBackwardGate
mluOpMoeDispatchForward
mluOpMsDeformAttnBackward
mluOpMsDeformAttnForward
mluOpMutualInformationBackward
mluOpMutualInformationForward
mluOpNms
mluOpNmsRotated
mluOpPointsInBoxes
mluOpPolyNms
mluOpPriorBox
mluOpPsRoiPoolBackward
mluOpPsRoiPoolForward
mluOpPsamaskBackward
mluOpPsamaskForward
mluOpResetTensorDescriptor
mluOpRoiAlignBackward
mluOpRoiAlignBackward_v2
mluOpRoiAlignForward_v2
mluOpRoiAlignRotatedBackward
mluOpRoiAlignRotatedForward
mluOpRoiAwarePool3dBackward
mluOpRoiAwarePool3dForward
mluOpRoiCropBackward
mluOpRoiCropForward
mluOpRoiPointPool3d
mluOpRoiPoolingBackward
mluOpRoiPoolingForward
mluOpRoiawarePool3dBackward
mluOpRoiawarePool3dForward
mluOpRotatedFeatureAlignBackward
mluOpRotatedFeatureAlignForward
mluOpSetAdamWDescAttr
mluOpSetAtomicsMode
mluOpSetCarafeDescriptor
mluOpSetDCNDescriptor
mluOpSetFFTReserveArea
mluOpSetGenCaseDirectory
mluOpSetGenCaseMode
mluOpSetGroupTensorDescriptors
mluOpSetNmsDescriptor
mluOpSetQuantizeRoundMode
mluOpSetQueue
mluOpSetRoiAlignForwardDescriptor_v2
mluOpSetSeqDataDescriptorPositionAndScale
mluOpSetSeqDataDescriptor_v2
mluOpSetSparseConvolutionDescriptor
mluOpSetTensorDescriptor
mluOpSetTensorDescriptorDim
mluOpSetTensorDescriptorDim_v2
mluOpSetTensorDescriptorEx
mluOpSetTensorDescriptorEx_v2
mluOpSetTensorDescriptorOnchipDataType
mluOpSetTensorDescriptorPointerMode
mluOpSetTensorDescriptorPosition
mluOpSetTensorDescriptorPositionAndScale
mluOpSetTensorDescriptorPositionScaleAndOffset
mluOpSetTensorDescriptor_v2
mluOpSqrt
mluOpSqrtBackward
mluOpSyncBatchNormBackwardElemt
mluOpSyncBatchNormBackwardElemtV2
mluOpSyncBatchNormBackwardReduce
mluOpSyncBatchNormBackwardReduce_v2
mluOpSyncBatchNormElemt
mluOpSyncBatchNormGatherStatsWithCounts
mluOpSyncBatchNormStats
mluOpSyncBatchNormStats_v2
mluOpSyncBatchnormBackwardReduce
mluOpSyncBatchnormBackwardReduce_v2
mluOpThreeInterpolateBackward
mluOpThreeInterpolateForward
mluOpThreeNNForward
mluOpTinShiftBackward
mluOpTinShiftForward
mluOpUpdateContextInformation
mluOpVoxelPoolingForward
mluOpVoxelization
mluOpYoloBox
setClusterNum