ATVOSS默认核配置详解

DefaultKernelConfig

【免费下载链接】atvossATVOSS（Ascend C Templates for Vector Operator Subroutines）是一套基于Ascend C开发的Vector算子库，致力于为昇腾硬件上的Vector类融合算子提供极简、高效、高性能、高拓展的编程方式。项目地址: https://gitcode.com/cann/atvoss

功能说明

默认的kernel层schedule配置参数的数据结构。

所属头文件链接

/include/elewise/kernel/builder.h

数据结构

struct DefaultKernelConfig { uint32_t blockNum = 1; uint64_t unitNumPerCore = 0; uint64_t moreUnitCoreNum = 0; uint64_t tailNum = 0; uint64_t unitNum = 1; };

数据结构成员说明

约束说明

NA

使用示例

template <typename InputDtype, typename OutputDtype> struct AddSubConfig { struct AddSubCompute { template <template <typename> class Tensor> __host_aicore__ constexpr auto Compute() const { auto in1 = Atvoss::PlaceHolder<1, Tensor<InputDtype>, Atvoss::ParamUsage::IN>(); auto in2 = Atvoss::PlaceHolder<2, Tensor<InputDtype>, Atvoss::ParamUsage::IN>(); auto in3 = Atvoss::PlaceHolder<3, InputDtype, Atvoss::ParamUsage::IN>(); auto out = Atvoss::PlaceHolder<4, Tensor<OutputDtype>, Atvoss::ParamUsage::OUT>(); return (out = in1 + in2 - in3); }; }; static constexpr Atvoss::Ele::DefaultKernelPolicy kernelPolicy{Atvoss::Ele::DefaultSegmentPolicy::UniformSegment}; using ArchTag = Atvoss::Arch::DAV_3510; using BlockOp = Atvoss::Ele::BlockBuilder<AddSubCompute, ArchTag>; using KernelOp = Atvoss::Ele::KernelBuilder< BlockOp, kernelPolicy, // 🔥🔥🔥 使用示例 🔥🔥🔥 Atvoss::Ele::DefaultKernelConfig // 🔥🔥🔥 使用示例 🔥🔥🔥 >; using DeviceOp = Atvoss::DeviceAdapter<KernelOp>; }; template <typename InputDtype, typename OutputDtype> static void Run() { /* ACL init and stream create */ ... Atvoss::Tensor<InputDtype> in1(deviceIn1, {{3, 4, 0, 0, 0, 0, 0, 0}}, 2); Atvoss::Tensor<InputDtype> in2(deviceIn2, {{3, 4, 0, 0, 0, 0, 0, 0}}, 2); InputDtype in3 = 5.0; Atvoss::Tensor<OutputDtype> out(deviceOut, {{3, 4, 0, 0, 0, 0, 0, 0}}, 2); auto arguments = Atvoss::ArgumentsBuilder{}.inputOutput(in1, in2, in3, out).attr("dim", 5).build(); using DeviceOp = typename AddSubConfig<InputDtype, OutputDtype>::DeviceOp; DeviceOp deviceOp; deviceOp.Run(arguments, stream); } int main(int argc, char const* argv[]) { Run<float, float>(); return 0; }

【免费下载链接】atvossATVOSS（Ascend C Templates for Vector Operator Subroutines）是一套基于Ascend C开发的Vector算子库，致力于为昇腾硬件上的Vector类融合算子提供极简、高效、高性能、高拓展的编程方式。项目地址: https://gitcode.com/cann/atvoss