aclnnMaskedSoftmaxWithRelPosBias
支持的产品型号
- Atlas 推理系列产品。
- Atlas A2训练系列产品/Atlas 800I A2推理产品。
接口原型
每个算子分为两段式接口,必须先调用“aclnnMaskedSoftmaxWithRelPosBiasGetWorkspaceSize”接口获取计算所需workspace大小以及包含了算子计算流程的执行器,再调用“aclnnMaskedSoftmaxWithRelPosBias”接口执行计算。
aclnnstatus aclnnMaskedSoftmaxWithRelPosBiasGetWorkspaceSize(const aclTensor *x, const aclTensor *attenMaskOptional, const aclTensor *relativePosBias, double scaleValueOptional, int64_t innerPrecisionModeOptional, aclTensor *out, uint64_t *workspaceSize, aclOpExecutor **executor)
aclnnstatus aclnnMaskedSoftmaxWithRelPosBias(void *workspace, uint64_t workspaceSize, aclOpExecutor *executor, aclrtStream stream)
功能描述
算子功能:替换在swinTransformer中使用window attention计算softmax的部分。
计算公式:
aclnnMaskedSoftmaxWithRelPosBiasGetWorkspaceSize
参数说明:
- x(aclTensor*, 计算输入):Device侧的aclTensor,数据类型支持FLOAT、FLOAT16、BFLOAT16,暂不支持非连续的Tensor,数据格式支持ND。其中BFLOAT16仅Atlas A2训练系列产品/Atlas 800I A2推理产品支持。支持shape为4维(B*W, N, S1, S2)和5维(B, W, N, S1, S2)。
- attenMaskOptional(aclTensor*, 计算输入):Device侧的aclTensor,数据类型支持FLOAT、FLOAT16、BFLOAT16,暂不支持非连续的Tensor,数据格式支持ND。其中BFLOAT16仅Atlas A2训练系列产品/Atlas 800I A2推理产品支持。支持shape为3维(W, S1, S2)、4维(W, 1, S1, S2)和5维(1, W, 1, S1, S2),可选输入,当不需要时为空指针, 数据类型需要保持和x一致。
- relativePosBias(aclTensor*, 计算输入):Device侧的aclTensor,数据类型支持FLOAT、FLOAT16、BFLOAT16,暂不支持非连续的Tensor,数据格式支持ND。其中BFLOAT16仅Atlas A2训练系列产品/Atlas 800I A2推理产品支持。支持shape为3维(N, S1, S2)、4维(1, N, S1, S2)和5维(1, 1, N, S1, S2), 数据类型需要保持和x一致。
- scaleValueOptional(double, 计算输入):Host侧的整型,数据类型支持DOUBLE。
- innerPrecisionModeOptional(int64_t, 预留参数):Host侧的整型,数据类型支持INT64。
- out(aclTensor*, 计算输出):Device侧的aclTensor,数据类型支持FLOAT、FLOAT16、BFLOAT16,暂不支持非连续的Tensor,数据格式支持ND。其中BFLOAT16仅Atlas A2训练系列产品/Atlas 800I A2推理产品支持。shape与数据类型需要保持和入参x一致。
- workspaceSize(uint64_t*, 出参):返回需要在Device侧申请的workspace大小。
- executor(aclOpExecutor**, 出参):返回op执行器,包含了算子计算流程。
返回值:
aclnnStatus:返回状态码,具体参见aclnn返回码。
第一段接口完成入参校验,出现以下场景时报错: 返回161001 (ACLNN_ERR_PARAM_NULLPTR):1. 传入的x、attenMaskOptional、relativePosBias或out是空指针。 返回561002 (ACLNN_ERR_INNER_TILING_ERROR): 1. 入参或者出参的数据类型、数据格式或shape不在支持的范围之内。
aclnnMaskedSoftmaxWithRelPosBias
参数说明:
- workspace(void *, 入参):在Device侧申请的workspace内存地址。
- workspaceSize(uint64_t, 入参):在Device侧申请的workspace大小,由第一段接口aclnnMaskedSoftmaxWithRelPosBiasGetWorkspaceSize获取。
- executor(aclOpExecutor *, 入参):op执行器,包含了算子计算流程。
- stream(aclrtStream, 入参):指定执行任务的AscendCL Stream流。
返回值:
aclnnStatus:返回状态码,具体参见aclnn返回码。
约束与限制
Atlas 推理系列产品不支持入参x的最后一个维度S2非32Byte对齐的场景。
需要保证传递给算子的shape所需要的ub空间小于AI处理器版本总ub的大小,该算子所需要的ub空间的总大小minComputeSize如下,其中s2AlignedSize 表示S2对齐32Byte后的结果:
对于attenMaskOptional存在的情况:
对于FLOAT类型,公式如下: dtypeSize = 4; xSize = s2AlignedSize * dtypeSize; softMaskMinTmpSize = 288; minComputeSize = xSize * 8 + softMaskMinTmpSize; 对于FLOAT16或者BFLOAT16类型,公式如下: dtypeSize = 2; xSize = s2AlignedSize * dtypeSize; softMaskMinTmpSize = 288; minComputeSize = xSize * 16 + softMaskMinTmpSize;
对于attenMaskOptional不存在的情况:
对于FLOAT类型,公式如下: dtypeSize = 4; xSize = s2AlignedSize * dtypeSize; softMaskMinTmpSize = 288; minComputeSize = xSize * 6 + softMaskMinTmpSize; 对于FLOAT16或者BFLOAT16类型,公式如下: dtypeSize = 2; xSize = s2AlignedSize * dtypeSize; softMaskMinTmpSize = 288; minComputeSize = xSize* 12 + softMaskMinTmpSize;
调用示例
示例代码如下,仅供参考,具体编译和执行过程请参考编译与运行样例。
#include <iostream>
#include <vector>
#include "acl/acl.h"
#include "aclnnop/aclnn_masked_softmax_with_rel_pos_bias.h"
#define CHECK_RET(cond, return_expr) \
do { \
if (!(cond)) { \
return_expr; \
} \
} while (0)
#define LOG_PRINT(message, ...) \
do { \
printf(message, ##__VA_ARGS__); \
} while (0)
int64_t GetShapeSize(const std::vector<int64_t>& shape) {
int64_t shapeSize = 1;
for (auto i : shape) {
shapeSize *= i;
}
return shapeSize;
}
int Init(int32_t deviceId, aclrtStream* stream) {
// 固定写法,AscendCL初始化
auto ret = aclInit(nullptr);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclInit failed. ERROR: %d\n", ret); return ret);
ret = aclrtSetDevice(deviceId);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtSetDevice failed. ERROR: %d\n", ret); return ret);
ret = aclrtCreateStream(stream);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtCreateStream failed. ERROR: %d\n", ret); return ret);
return 0;
}
template <typename T>
int CreateAclTensor(const std::vector<T>& hostData, const std::vector<int64_t>& shape, void** deviceAddr,
aclDataType dataType, aclTensor** tensor) {
auto size = GetShapeSize(shape) * sizeof(T);
// 调用aclrtMalloc申请device侧内存
auto ret = aclrtMalloc(deviceAddr, size, ACL_MEM_MALLOC_HUGE_FIRST);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtMalloc failed. ERROR: %d\n", ret); return ret);
// 调用aclrtMemcpy将host侧数据拷贝到device侧内存上
ret = aclrtMemcpy(*deviceAddr, size, hostData.data(), size, ACL_MEMCPY_HOST_TO_DEVICE);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtMemcpy failed. ERROR: %d\n", ret); return ret);
// 计算连续tensor的strides
std::vector<int64_t> strides(shape.size(), 1);
for (int64_t i = shape.size() - 2; i >= 0; i--) {
strides[i] = shape[i + 1] * strides[i + 1];
}
// 调用aclCreateTensor接口创建aclTensor
*tensor = aclCreateTensor(shape.data(), shape.size(), dataType, strides.data(), 0, aclFormat::ACL_FORMAT_ND,
shape.data(), shape.size(), *deviceAddr);
return 0;
}
int main() {
// 1. (固定写法)device/stream初始化,参考AscendCL对外接口列表
// 根据自己的实际device填写deviceId
int32_t deviceId = 0;
aclrtStream stream;
auto ret = Init(deviceId, &stream);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("Init acl failed. ERROR: %d\n", ret); return ret);
// 2. 构造输入与输出,需要根据API的接口自定义构造
std::vector<int64_t> xShape = {1, 1, 1, 2, 16};
std::vector<int64_t> attenMaskOptionalShape = {1, 2, 16};
std::vector<int64_t> relativePosBiasShape = {1, 2, 16};
std::vector<int64_t> outShape = {1, 1, 1, 2, 16};
void* xDeviceAddr = nullptr;
void* attenMaskOptionalDeviceAddr = nullptr;
void* relativePosBiasDeviceAddr = nullptr;
void* outDeviceAddr = nullptr;
aclTensor* x = nullptr;
aclTensor* attenMaskOptional = nullptr;
aclTensor* relativePosBias = nullptr;
aclTensor* out = nullptr;
std::vector<float> xHostData = {1.08, -1.56, -1.3, -2.01, 2.18, -2.23, -3.58, 3.22, 1.25, -0.56, -0.3, -1.01, 1.08, -1.13, -3.08, -2.22, -0.08, -2.56, 1.35, 1.01, 0.35, -1.03, -1.28, 1.22, 0.08, -2.56, -1.01, -1.01, -0.18, -6.23, 4.55, -1.82};
std::vector<float> attenMaskOptionalHostData = {2, 2, 2, 2, 3, 3, 3, 3, 2, 2, 2, 2, 3, 3, 3, 3, 2, 2, 2, 2, 3, 3, 3, 3, 2, 2, 2, 2, 3, 3, 3, 3,};
std::vector<float> relativePosBiasHostData = {1, 1, 1, 1, 4, 4, 4, 4, 1, 1, 1, 1, 4, 4, 4, 4, 1, 1, 1, 1, 4, 4, 4, 4, 1, 1, 1, 1, 4, 4, 4, 4};
std::vector<float> outHostData = {0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0};
ret = CreateAclTensor(xHostData, xShape, &xDeviceAddr, aclDataType::ACL_FLOAT, &x);
CHECK_RET(ret == ACL_SUCCESS, return ret);
ret = CreateAclTensor(attenMaskOptionalHostData, attenMaskOptionalShape, &attenMaskOptionalDeviceAddr, aclDataType::ACL_FLOAT, &attenMaskOptional);
CHECK_RET(ret == ACL_SUCCESS, return ret);
ret = CreateAclTensor(relativePosBiasHostData, relativePosBiasShape, &relativePosBiasDeviceAddr, aclDataType::ACL_FLOAT, &relativePosBias);
CHECK_RET(ret == ACL_SUCCESS, return ret);
ret = CreateAclTensor(outHostData, outShape, &outDeviceAddr, aclDataType::ACL_FLOAT, &out);
CHECK_RET(ret == ACL_SUCCESS, return ret);
// 3. 调用CANN算子库API,需要修改为具体的Api名称
uint64_t workspaceSize = 0;
aclOpExecutor* executor;
// 调用aclnnMaskedSoftmaxWithRelPosBias第一段接口
ret = aclnnMaskedSoftmaxWithRelPosBiasGetWorkspaceSize(x, attenMaskOptional, relativePosBias, 1.0, 0, out, &workspaceSize, &executor);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnMaskedSoftmaxWithRelPosBiasGetWorkspaceSize failed. ERROR: %d\n", ret); return ret);
// 根据第一段接口计算出的workspaceSize申请device内存
void* workspaceAddr = nullptr;
if (workspaceSize > 0) {
ret = aclrtMalloc(&workspaceAddr, workspaceSize, ACL_MEM_MALLOC_HUGE_FIRST);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("allocate workspace failed. ERROR: %d\n", ret); return ret);
}
// 调用aclnnMaskedSoftmaxWithRelPosBias第二段接口
ret = aclnnMaskedSoftmaxWithRelPosBias(workspaceAddr, workspaceSize, executor, stream);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnMaskedSoftmaxWithRelPosBias failed. ERROR: %d\n", ret); return ret);
// 4. (固定写法)同步等待任务执行结束
ret = aclrtSynchronizeStream(stream);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtSynchronizeStream failed. ERROR: %d\n", ret); return ret);
// 5. 获取输出的值,将device侧内存上的结果拷贝至host侧,需要根据具体API的接口定义修改
auto size = GetShapeSize(outShape);
std::vector<float> resultData(size, 0);
ret = aclrtMemcpy(resultData.data(), resultData.size() * sizeof(resultData[0]), outDeviceAddr,
size * sizeof(resultData[0]), ACL_MEMCPY_DEVICE_TO_HOST);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("copy result from device to host failed. ERROR: %d\n", ret); return ret);
for (int64_t i = 0; i < size; i++) {
LOG_PRINT("result[%ld] is: %f\n", i, resultData[i]);
}
// 6. 释放aclTensor和aclScalar,需要根据具体API的接口定义修改
aclDestroyTensor(x);
aclDestroyTensor(attenMaskOptional);
aclDestroyTensor(relativePosBias);
aclDestroyTensor(out);
// 7. 释放device资源
aclrtFree(xDeviceAddr);
aclrtFree(attenMaskOptionalDeviceAddr);
aclrtFree(relativePosBiasDeviceAddr);
aclrtFree(outDeviceAddr);
if (workspaceSize > 0) {
aclrtFree(workspaceAddr);
}
aclrtDestroyStream(stream);
aclrtResetDevice(deviceId);
aclFinalize();
return 0;
}