aclnnRepeatInterleave
Applicable Products
| Product | Supported |
|---|---|
| √ | |
| √ | |
| √ | |
| × | |
| × | |
| × |
Function
Description: After flattening the tensor self, repeats each element the corresponding number of times specified in the repeats tensor.
Example:
Assume the input tensor is ([[a, b], [c, d], [e, f]]) and repeats is ([1, 2, 2, 1, 1, 1]). The resulting tensor is tensor([a, b, b, c, c, d, e, f]). The Number of Elements in the output tensor is 8, which equals the sum of all elements in repeats. After flattening the tensor, input becomes ([a, b, c, d, e, f]). This tensor is replicated in one-to-one correspondence with repeats: a is repeated once, b is repeated twice, c is repeated twice, and so on.
Assume the input tensor is ([[a, b], [c, d], [e, f]]) and repeats is ([2]). The final generated tensor is tensor([a, a, b, b, c, c, d, d, e, e, f, f]). After flattening the tensor, the input becomes ([a, b, c, d, e, f]). Each element in this tensor is repeated by the number of times specified in repeats, meaning each element is repeated twice. Note: This scenario is equivalent to repeats being (2).
Function Prototype
Each operator is divided into a two-phase API. The "aclnnRepeatInterleaveGetWorkspaceSize" API must be called first to obtain the workspace size required for computation and the executor that encapsulates the operator computation flow. Then, the "aclnnRepeatInterleave" API is called to perform the computation.
aclnnStatus aclnnRepeatInterleaveGetWorkspaceSize(
const aclTensor* self,
const aclTensor* repeats,
int64_t outputSize,
aclTensor* out,
uint64_t* workspaceSize,
aclOpExecutor** executor)aclnnStatus aclnnRepeatInterleave(
void* workspace,
uint64_t workspaceSize,
aclOpExecutor* executor,
aclrtStream stream)aclnnRepeatInterleaveGetWorkspaceSize
Parameters:
Parameter Input/Output Description Instruction Data Type Data Format Dimension (Shape) Non-Contiguous Tensor self (aclTensor*) Input Input tensor to be replicated as described in Function. Supports empty tensors. UINT8, INT8, INT16, INT32, INT64, BOOL, FLOAT16, BFLOAT16, FLOAT ND 0-8 √ repeats (aclTensor*) Input Number of repetitions. Supports empty tensors.
Only 0D/1D tensors are allowed.
If it is a 1D tensor, the size of repeats must be 1 or the Number of Elements of self.INT32, INT64 ND 0-1 √ outputSize (int64_t) Input Final size of the tensor after the repeat operation. When repeats contains only one element, outputSize = Number of Elements of self * value of repeats.
When repeats contains multiple values, outputSize = sum of values of repeats.- - - - out (aclTensor*) Output Output tensor after data replication as described in Function. The data type must be the same as self.
The shape is a 1D tensor, and the shape size is the same as outputSize.UINT8, INT8, INT16, INT32, INT64, BOOL, FLOAT16, BFLOAT16, FLOAT ND 1 √ workspaceSize (uint64_t*) Output Returns the workspace size to be allocated on the Device side. - - - - - Executor (aclOpExecutor**) Output Returns the op executor, which contains the operator computation flow. - - - - - 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 Return Code Description ACLNN_ERR_PARAM_NULLPTR 161001 self, repeats, or out contains a null pointer. ACLNN_ERR_PARAM_INVALID 161002 The data types of self and repeats are not within the supported range. The data types of self and out are different. repeats is not a 0D/1D tensor. When repeats is a 1D tensor, the size of repeats is neither 1 nor the number of elements in self. The number of dimensions of self exceeds 8. When self is an empty tensor, but repeats is not an empty tensor, not a 0-dimensional tensor with one element, and not a 1-dimensional tensor with one element. When self is not an empty tensor, but repeats is an empty tensor.
aclnnRepeatInterleave
Parameters:
Parameter Input/Output Description workspace Input Address of the workspace memory applied for on the Device side. workspaceSize Input Size of the workspace applied for on the Device side, obtained from the first-phase API aclnnRepeatInterleaveGetWorkspaceSize. executor Input Operator executor, which contains the operator computation flow. stream Input Specifies the stream for task execution. Return value
aclnnStatus: return code. For details, see aclnn Return Code.
Constraints
Deterministic computation:
- aclnnRepeatInterleave defaults to a deterministic implementation.
Input shape constraints: repeats can only be a 0D or 1D tensor. If it is a 1D tensor, the size of repeats must be 1 or the number of elements in self.
Input value range constraints: The values in the repeats tensor must be natural numbers.
Other restrictions: The value of outputSize must meet the following calculation results: When repeats contains only one element, outputSize = Number of Elements of self * value of repeats. When repeats contains multiple values, outputSize = sum of the values of repeats.
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_repeat_interleave.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) {
// Boilerplate: resource initialization.
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 the 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);
// Calculate 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 the aclCreateTensor API 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);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("Init acl failed. ERROR: %d\n", ret); return ret);
// 2. Construct the input and output. Custom construction is required based on the API.
std::vector<int64_t> selfShape = {2, 3};
std::vector<int64_t> repeatsShape = {6};
std::vector<int64_t> outShape = {21};
void* selfDeviceAddr = nullptr;
void* repeatsDeviceAddr = nullptr;
void* outDeviceAddr = nullptr;
aclTensor* self = nullptr;
aclTensor* repeats = nullptr;
aclTensor* out = nullptr;
int64_t output_size = 21;
std::vector<float> selfHostData = {3, 4, 5, -3, -4, -5};
std::vector<int64_t> repeatsHostData = {1, 2, 3, 4, 5, 6};
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};
// Create the self aclTensor.
ret = CreateAclTensor(selfHostData, selfShape, &selfDeviceAddr, aclDataType::ACL_FLOAT, &self);
CHECK_RET(ret == ACL_SUCCESS, return ret);
// Create the repeats aclTensor.
ret = CreateAclTensor(repeatsHostData, repeatsShape, &repeatsDeviceAddr, aclDataType::ACL_INT64, &repeats);
CHECK_RET(ret == ACL_SUCCESS, return ret);
// Create the out aclTensor.
ret = CreateAclTensor(outHostData, outShape, &outDeviceAddr, aclDataType::ACL_FLOAT, &out);
CHECK_RET(ret == ACL_SUCCESS, return ret);
// 3. Call the CANN operator library API. Replace with the specific API name.
uint64_t workspaceSize = 0;
aclOpExecutor* executor;
// Call the first-phase API of aclnnRepeatInterleave.
ret = aclnnRepeatInterleaveGetWorkspaceSize(self, repeats, output_size, out, &workspaceSize, &executor);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnRepeatInterleaveGetWorkspaceSize 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 aclnnRepeatInterleave.
ret = aclnnRepeatInterleave(workspaceAddr, workspaceSize, executor, stream);
CHECK_RET(ret == ACL_SUCCESS, LOG_PRINT("aclnnRepeatInterleave failed. ERROR: %d\n", ret); return ret);
// 4. (Boilerplate) Wait for the task execution to complete synchronously.
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 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. Release aclTensor and aclScalar. Modify based on the specific API definition.
aclDestroyTensor(self);
aclDestroyTensor(repeats);
aclDestroyTensor(out);
// 7. Release device resources. Modify based on the specific API definition.
aclrtFree(selfDeviceAddr);
aclrtFree(repeatsDeviceAddr);
aclrtFree(outDeviceAddr);
if (workspaceSize > 0) {
aclrtFree(workspaceAddr);
}
aclrtDestroyStream(stream);
aclrtResetDevice(deviceId);
aclFinalize();
return 0;
}