aclnnRingAttentionUpdate

Applicable Products

Function

• Description: The RingAttentionUpdate operator updates the outputs of two FlashAttention operations based on their different softmax max and sum values.

• Formula:

Function Prototype

Each operator is divided into a two-phase API. You must first call the "aclnnRingAttentionUpdateGetWorkspaceSize" API to obtain the workspace size required for computation and the executor that contains the operator computation process, and then call the "aclnnRingAttentionUpdate" API to perform the computation.

aclnnStatus aclnnRingAttentionUpdateGetWorkspaceSize(
  const aclTensor *prevAttnOut,
  const aclTensor *prevSoftmaxMax,
  const aclTensor *prevSoftmaxSum,
  const aclTensor *curAttnOut,
  const aclTensor *curSoftmaxMax,
  const aclTensor *curSoftmaxSum,
  const aclTensor *actualSeqQlenOptional,
  char            *inputLayoutOptional,
  const aclTensor *attnOutOut,
  const aclTensor *softmaxMaxOut,
  const aclTensor *softmaxSumOut,
  uint64_t        *workspaceSize,
  aclOpExecutor  **executor)

aclnnStatus aclnnRingAttentionUpdate(
  void          *workspace,
  uint64_t       workspaceSize,
  aclOpExecutor *executor,
  aclrtStream    stream)

aclnnRingAttentionUpdateGetWorkspaceSize

• Parameters:

  Parameter	Input/Output	Description	Instruction	Data Type	Data Format	Dimension (Shape)	Non-contiguous Tensor
  prevAttnOut (aclTensor*)	Input	prev_attn_out in the formula, the output of the first FlashAttention.	The input shape and the inputLayoutOptional attribute remain consistent.	FLOAT16, FLOAT, BFLOAT16	ND	[T,N,D], [S,B,H]	√
  prevSoftmaxMax (aclTensor*)	Input	prev_softmax_max in the formula, the max result of the softmax from the first time FlashAttention.	The eight numbers in the last dimension are identical and must be positive numbers.	FLOAT	ND	[B,N,S,8]、[T,N,8]	√
  prevSoftmaxSum (aclTensor*)	Input	prev_softmax_sum in the formula, the softmax sum result from the first time FlashAttention.	The input shape must remain consistent with prevSoftmaxMax, with the last dimension having the same 8 numbers and must be a positive number.	FLOAT	ND	[B,N,S,8], [T,N,8]	√
  curAttnOut (aclTensor*)	Input	cur_attn_out in the formula, the output of the second FlashAttention.	The data type and input shape remain consistent with prevAttnOut.	FLOAT16, FLOAT, BFLOAT16	ND	[T,N,D], [S,B,H]	√
  curSoftmaxMax (aclTensor*)	Input	cur_softmax_max in the formula, the softmax max result of the second time FlashAttention.	The input shape must remain consistent with prevSoftmaxMax, with the last dimension having the same 8 numbers and must be a positive number.	FLOAT	ND	[B,N,S,8], [T,N,8]	√
  curSoftmaxSum (aclTensor*)	Input	cur_softmax_sum in the formula, the sum result of the softmax from the second FlashAttention.	The input shape remains consistent with prevSoftmaxMax, with the last dimension having 8 identical numbers, and must be a positive number.	FLOAT	ND	[B,N,S,8], [T,N,8]	√
  actualSeqQlenOptional (aclTensor*)	Input	Cumulative sum of sequence lengths starting from 0.	When the data layout inputLayoutOptional is TND, this parameter must be passed. It is an integer aclTensor that increments from 0 to T.	INT64	-	-	-
  inputLayoutOptional (char*)	Input	Data layout of the attn_out-related inputs.	A char* constant on the host side. Currently supports "TND" and "SBH".	-	-	-	-
  attnOutOut (aclTensor*)	Output	attn_out in the formula, the output after two result updates.	The data type and output shape remain consistent with prevAttnOut.	FLOAT16, FLOAT, BFLOAT16	ND	[T,N,D], [S,B,H]	√
  softmaxMaxOut (aclTensor*)	Output	Represents softmax_max in the formula, which is the max of softmax after the result update through two times.	The output shape remains consistent with prevSoftmaxMax.	FLOAT	ND	[B,N,S,8]、[T,N,8]	√
  softmaxSumOut (aclTensor*)	Output	softmax_sum in the formula, which is the sum of softmax after two result updates.	The output shape remains consistent with prevSoftmaxMax.	FLOAT	ND	[B,N,S,8], [T,N,8]	√
  workspaceSize (uint64_t*)	Output	Size of the workspace to be applied on the Device side.	-	-	-	-	-
  executor (aclOpExecutor**)	Output	Returns the op executor, which contains the operator computation flow.	-	-	-	-	-

  • Atlas A2 training products/Atlas A2 inference products, Atlas A3 training products/Atlas A3 inference products: When the input data layout inputLayoutOptional is TND, D must be a multiple of 64.

• Return Value

  aclnnStatus: return code. For details, see aclnn Return Code.

  The first-phase API performs input parameter validation and returns an error in the following scenarios:

  Return Value	Error Code	Description
  ACLNN_ERR_PARAM_NULLPTR	161001	Indicates that prevAttnOut, prevSoftmaxMax, prevSoftmaxSum, curAttnOut, curSoftmaxMax, curSoftmaxSum, attnOutOut, softmaxMaxOut, or softmaxSumOut is a null pointer.
  ACLNN_ERR_PARAM_INVALID	161002	The data type of prevAttnOut, prevSoftmaxMax, prevSoftmaxSum, curAttnOut, curSoftmaxMax, curSoftmaxSum, attnOutOut, softmaxMaxOut, or softmaxSumOut is not within the supported range.
  ACLNN_ERR_PARAM_INVALID	561002	When actualSeqQlenOptional is provided, the input data format is not within the supported range.
  		The shape of prevAttnOut, prevSoftmaxMax, prevSoftmaxSum, curAttnOut, curSoftmaxMax, curSoftmaxSum, attnOutOut, softmaxMaxOut, or softmaxSumOut is empty.
  		The shape of prevAttnOut, prevSoftmaxMax, prevSoftmaxSum, curAttnOut, curSoftmaxMax, curSoftmaxSum, attnOutOut, softmaxMaxOut, or softmaxSumOut is not supported.

aclnnRingAttentionUpdate

• Parameters:

  Parameter	Input/Output	Description
  workspace	Input	Address of the workspace memory applied on the Device side.
  workspaceSize	Input	Size of the workspace applied on the Device side, obtained through the first-phase API aclnnRingAttentionUpdateGetWorkspaceSize.
  executor	Input	Operator executor that contains the operator computation flow.
  stream	Input	Specifies the stream that executes the task.

• Return value:

  aclnnStatus: return code. For details, see aclnn Return Code.

Constraints

• Deterministic computation:

  • aclnnRingAttentionUpdate defaults to a deterministic implementation.

  • When inputLayoutOptional is set to "TND", actualSeqQlenOptional is mandatory.

  • When inputLayoutOptional is set to "TND":

    • N:

      • Atlas A2 training products/Atlas A2 inference products, Atlas A3 training products/Atlas A3 inference products: N <= 256.

    • D:

      • Atlas A2 training products/Atlas A2 inference products, Atlas A3 training products/Atlas A3 inference products: D <= 768 and D is a multiple of 64.

Example

The following provides example code for reference only. For details about compilation and running, see Compile and Run Samples.

#include <iostream>
#include <vector>
#include "acl/acl.h"
#include "aclnnop/aclnn_ring_attention_update.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 shape_size = 1;
  for (auto i : shape) {
    shape_size *= i;
  }
  return shape_size;
}

int Init(int32_t deviceId, aclrtStream* stream) {
  // Boilerplate. Initialize resources.
  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);
  // Call aclrtMalloc to apply for device-side memory.
  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);

  // Call aclrtMemcpy to copy host-side data to device-side memory.
  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);

  // Compute the strides of a contiguous tensor.
  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];
  }

  // Call aclCreateTensor to create an 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. (Boilerplate) Initialize the device/stream. See the acl API manual.
  // Fill in the deviceId based on your actual device.
  int32_t deviceId = 0;
  aclrtStream stream;
  auto ret = Init(deviceId, &stream);
  // Handle the check based on your needs.
  CHECK_RET(ret == 0, LOG_PRINT("Init acl failed. ERROR: %d\n", ret); return ret);
  // 2. Construct the input and output. Customize the construction based on the API interface.
  int64_t batchNum = 1;
  int64_t headNum = 1;
  int64_t seqSize = 2;
  int64_t headDim = 4;
  int64_t headSize = headNum * headDim;

  std::vector<int64_t> prevAttnOutShape = {seqSize, batchNum, headSize};
  std::vector<int64_t> prevSoftmaxMaxShape = {batchNum, headNum, seqSize, 8};
  std::vector<int64_t> prevSoftmaxSumShape = {batchNum, headNum, seqSize, 8};
  std::vector<int64_t> curAttnOutShape = {seqSize, batchNum, headSize};
  std::vector<int64_t> curSoftmaxMaxShape = {batchNum, headNum, seqSize, 8};
  std::vector<int64_t> curSoftmaxSumShape = {batchNum, headNum, seqSize, 8};
  std::vector<int64_t> actualSeqQlenOptionalShape = {batchNum, headNum};

  std::vector<int64_t> attnOutShape = {seqSize, batchNum, headSize};
  std::vector<int64_t> softmaxMaxShape = {batchNum, headNum, seqSize, 8};
  std::vector<int64_t> softmaxSumShape = {batchNum, headNum, seqSize, 8};

  void* prevAttnOutDeviceAddr = nullptr;
  void* prevSoftmaxMaxDeviceAddr = nullptr;
  void* prevSoftmaxSumDeviceAddr = nullptr;
  void* curAttnOutDeviceAddr = nullptr;
  void* curSoftmaxMaxDeviceAddr = nullptr;
  void* curSoftmaxSumDeviceAddr = nullptr;
  void* actualSeqQlenOptionalDeviceAddr = nullptr;

  void* attnOutDeviceAddr = nullptr;
  void* softmaxMaxDeviceAddr = nullptr;
  void* softmaxSumDeviceAddr = nullptr;

  aclTensor* prevAttnOut = nullptr;
  aclTensor* prevSoftmaxMax = nullptr;
  aclTensor* prevSoftmaxSum = nullptr;
  aclTensor* curAttnOut = nullptr;
  aclTensor* curSoftmaxMax = nullptr;
  aclTensor* curSoftmaxSum = nullptr;
  aclTensor* actualSeqQlenOptional = nullptr;

  aclTensor* attnOut = nullptr;
  aclTensor* softmaxMax = nullptr;
  aclTensor* softmaxSum = nullptr;

  std::vector<float> prevAttnOutHostData(seqSize * batchNum * headSize, 1);
  std::vector<float> prevSoftmaxMaxHostData(batchNum * headNum * seqSize * 8, 1);
  std::vector<float> prevSoftmaxSumHostData(batchNum * headNum * seqSize * 8, 1);
  std::vector<float> curAttnOutHostData(seqSize * batchNum * headSize, 1);
  std::vector<float> curSoftmaxMaxHostData(batchNum * headNum * seqSize * 8, 1);
  std::vector<float> curSoftmaxSumHostData(batchNum * headNum * seqSize * 8, 1);
  std::vector<float> actualSeqQlenOptionalHostData(batchNum * headNum, 1);

  std::vector<float> attnOutHostData(seqSize * batchNum * headSize, 1);
  std::vector<float> softmaxMaxHostData(batchNum * headNum * seqSize * 8, 1);
  std::vector<float> softmaxSumHostData(batchNum * headNum * seqSize * 8, 1);

  char* inputLayoutOptional = "SBH";
  // Create the prevAttnOut aclTensor.
  ret = CreateAclTensor(prevAttnOutHostData, prevAttnOutShape, &prevAttnOutDeviceAddr, aclDataType::ACL_FLOAT, &prevAttnOut);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  // Create the prevSoftmaxMax aclTensor.
  ret = CreateAclTensor(prevSoftmaxMaxHostData, prevSoftmaxMaxShape, &prevSoftmaxMaxDeviceAddr, aclDataType::ACL_FLOAT, &prevSoftmaxMax);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  // Create prevSoftmaxSum aclTensor.
  ret = CreateAclTensor(prevSoftmaxSumHostData, prevSoftmaxSumShape, &prevSoftmaxSumDeviceAddr, aclDataType::ACL_FLOAT, &prevSoftmaxSum);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  // Create curAttnOut aclTensor.
  ret = CreateAclTensor(curAttnOutHostData, curAttnOutShape, &curAttnOutDeviceAddr, aclDataType::ACL_FLOAT, &curAttnOut);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  // Create curSoftmaxMax aclTensor.
  ret = CreateAclTensor(curSoftmaxMaxHostData, curSoftmaxMaxShape, &curSoftmaxMaxDeviceAddr, aclDataType::ACL_FLOAT, &curSoftmaxMax);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  // Create curSoftmaxSum aclTensor.
  ret = CreateAclTensor(curSoftmaxSumHostData, curSoftmaxSumShape, &curSoftmaxSumDeviceAddr, aclDataType::ACL_FLOAT, &curSoftmaxSum);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  // Create actualSeqQlenOptional aclTensor.
  ret = CreateAclTensor(actualSeqQlenOptionalHostData, actualSeqQlenOptionalShape, &actualSeqQlenOptionalDeviceAddr, aclDataType::ACL_INT64, &actualSeqQlenOptional);
  CHECK_RET(ret == ACL_SUCCESS, return ret);

  // Create the attnOut aclTensor.
  ret = CreateAclTensor(attnOutHostData, attnOutShape, &attnOutDeviceAddr, aclDataType::ACL_FLOAT, &attnOut);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  // Create the softmaxMax aclTensor.
  ret = CreateAclTensor(softmaxMaxHostData, softmaxMaxShape, &softmaxMaxDeviceAddr, aclDataType::ACL_FLOAT, &softmaxMax);
  CHECK_RET(ret == ACL_SUCCESS, return ret);
  // Create the softmaxSum aclTensor.
  ret = CreateAclTensor(softmaxSumHostData, softmaxSumShape, &softmaxSumDeviceAddr, aclDataType::ACL_FLOAT, &softmaxSum);
  CHECK_RET(ret == ACL_SUCCESS, return ret);

  // 3. Call the CANN operator library API. This needs to be modified to the specific API.
  uint64_t workspaceSize = 0;
  aclOpExecutor* executor;
  // Call the first-phase API of aclnnRingAttentionUpdate.
  ret = aclnnRingAttentionUpdateGetWorkspaceSize(prevAttnOut, prevSoftmaxMax, prevSoftmaxSum,
                                                 curAttnOut, curSoftmaxMax, curSoftmaxSum,
                                                 actualSeqQlenOptional, inputLayoutOptional,
                                                 attnOut, softmaxMax, softmaxSum, &workspaceSize, &executor);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnRingAttentionUpdateGetWorkspaceSize failed. ERROR: %d\n", ret); return ret);
  // Apply for device memory based on the workspaceSize calculated by the first-phase API.
  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;);
  }
  // Call the second-phase API of aclnnRingAttentionUpdate.
  ret = aclnnRingAttentionUpdate(workspaceAddr, workspaceSize, executor, stream);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnRingAttentionUpdate failed. ERROR: %d\n", ret); return ret);
  // 4. (Boilerplate) Synchronously wait for the task execution to complete.
  ret = aclrtSynchronizeStream(stream);
  CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclrtSynchronizeStream failed. ERROR: %d\n", ret); return ret);
  // 5. Obtain the output value and copy the result from the device memory to the host memory. Modify based on the specific API definition.
  auto attnOutSize = GetShapeSize(attnOutShape);
  std::vector<float> attnOutResultData(attnOutSize, 0);
  ret = aclrtMemcpy(attnOutResultData.data(), attnOutResultData.size() * sizeof(attnOutResultData[0]), attnOutDeviceAddr, attnOutSize * sizeof(float),
                    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 < attnOutSize; i++) {
    LOG_PRINT("attnOutResultData[%ld] is: %f\n", i, attnOutResultData[i]);
  }

  auto softmaxMaxSize = GetShapeSize(softmaxMaxShape);
  std::vector<float> softmaxMaxResultData(softmaxMaxSize, 0);
  ret = aclrtMemcpy(softmaxMaxResultData.data(), softmaxMaxResultData.size() * sizeof(softmaxMaxResultData[0]), softmaxMaxDeviceAddr, softmaxMaxSize * sizeof(float),
                    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 < softmaxMaxSize; i++) {
    LOG_PRINT("softmaxMaxResultData[%ld] is: %f\n", i, softmaxMaxResultData[i]);
  }

  auto softmaxSumSize = GetShapeSize(softmaxSumShape);
  std::vector<float> softmaxSumResultData(softmaxSumSize, 0);
  ret = aclrtMemcpy(softmaxSumResultData.data(), softmaxSumResultData.size() * sizeof(softmaxSumResultData[0]), softmaxSumDeviceAddr, softmaxSumSize * sizeof(float),
                    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 < softmaxSumSize; i++) {
    LOG_PRINT("softmaxSumResultData[%ld] is: %f\n", i, softmaxSumResultData[i]);
  }

  // 6. Release aclTensor and aclScalar. Modify based on the specific API definition.
  aclDestroyTensor(prevAttnOut);
  aclDestroyTensor(prevSoftmaxMax);
  aclDestroyTensor(prevSoftmaxSum);
  aclDestroyTensor(curAttnOut);
  aclDestroyTensor(curSoftmaxMax);
  aclDestroyTensor(curSoftmaxSum);
  aclDestroyTensor(actualSeqQlenOptional);
  aclDestroyTensor(attnOut);
  aclDestroyTensor(softmaxMax);
  aclDestroyTensor(softmaxSum);

  // 7. Release device resources. Modify based on the specific API definition.
  aclrtFree(prevAttnOutDeviceAddr);
  aclrtFree(prevSoftmaxMaxDeviceAddr);
  aclrtFree(prevSoftmaxSumDeviceAddr);
  aclrtFree(curAttnOutDeviceAddr);
  aclrtFree(curSoftmaxMaxDeviceAddr);
  aclrtFree(curSoftmaxSumDeviceAddr);
  aclrtFree(actualSeqQlenOptionalDeviceAddr);
  aclrtFree(attnOutDeviceAddr);
  aclrtFree(softmaxMaxDeviceAddr);
  aclrtFree(softmaxSumDeviceAddr);

  if (workspaceSize > 0) {
    aclrtFree(workspaceAddr);
  }
  aclrtDestroyStream(stream);
  aclrtResetDevice(deviceId);
  aclFinalize();
  return 0;
}