Sub

Function Usage

Performs subtraction based on elements using the following formula, where PAR indicates the number of elements that can be processed by the Vector Unit in one iteration.

$\text{[math]}$

Prototype

Computation of the entire tensor
1

dstLocal = src0Local - src1Local;

Computation of the first n data elements of a tensor

        
             template <typename T>
__aicore__ inline void Sub(const LocalTensor<T>& dstLocal, const LocalTensor<T>& src0Local, const LocalTensor<T>& src1Local, const int32_t& calCount)

High-dimensional tensor sharding computation

Bitwise mask mode

          
               template <typename T, bool isSetMask = true>
__aicore__ inline void Sub(const LocalTensor<T>& dstLocal, const LocalTensor<T>& src0Local, const LocalTensor<T>& src1Local, uint64_t mask[], const uint8_t repeatTimes, const BinaryRepeatParams& repeatParams)

Contiguous mask mode

          
               template <typename T, bool isSetMask = true>
__aicore__ inline void Sub(const LocalTensor<T>& dstLocal, const LocalTensor<T>& src0Local, const LocalTensor<T>& src1Local, uint64_t mask, const uint8_t repeatTimes, const BinaryRepeatParams& repeatParams)

Parameters

**Table 1** Parameters in the template
Parameter	Description
T	Operand data type.
isSetMask	Indicates whether to set mask inside the API. true: sets mask inside the API. false: sets mask outside the API. Developers need to use the SetVectorMask API to set the mask value. In this mode, the mask value in the input parameter of this API must be set to MASK_PLACEHOLDER.

**Table 2** Parameters
Parameter	Input/Output	Description
dstLocal	Output	Destination operand. The type is LocalTensor, and the supported TPosition is VECIN, VECCALC, or VECOUT. The start address of the LocalTensor must be 32-byte aligned. For the Atlas Training Series Product , the supported data types are half, float, and int32_t.
src0Local and src1Local	Input	Source operand. The type is LocalTensor, and the supported TPosition is VECIN, VECCALC, or VECOUT. The start address of the LocalTensor must be 32-byte aligned. The two source operands must have the same data type as the destination operand. For the Atlas Training Series Product , the supported data types are half, float, and int32_t.
calCount	Input	Number of elements of the input data.
mask	Input	mask is used to control the elements that participate in computation in each iteration. Contiguous mode: indicates the number of contiguous elements that participate in computation. The value range is related to the operand data type. The maximum number of elements that can be processed in each iteration varies according to the data type. When the operand is 16-bit, mask ∈ [1, 128]. When the operand is 32-bit, mask ∈ [1, 64]. When the operand is 64-bit, mask ∈ [1, 32]. Bitwise mode: controls the elements that participate in computation by bit. If a bit is set to 1, the corresponding element participates in the computation. If a bit is set to 0, the corresponding element is masked in the computation. The parameter type is a uint64_t array whose length is 2. For example, if mask = [0, 8] and 8 = 0b1000, only the fourth element participates in computation. The parameter value range is related to the operand data type. The maximum number of elements that can be processed in each iteration varies according to the data type. When the operand is 16-bit, mask[0] and mask[1] ∈ [0, 2⁶⁴ -1] and cannot be 0 at the same time. When the operand is 32-bit, mask[1] is 0 and mask[0] ∈ (0, 2⁶⁴ – 1]. When the operand is 64-bit, mask[1] is 0 and mask[0] ∈ (0, 2³² – 1].
repeatTimes	Input	Number of iteration repeats. The Vector Unit reads 256 bytes of contiguous data for computation each time. To read the complete data for processing, the unit needs to read the input data in multiple repeats. repeatTimes indicates the number of iterations. For details about this parameter, see Common Parameters.
repeatParams	Input	Parameters that control the operand address strides. They are of the BinaryRepeatParams type, and contain such parameters as those that specify the address stride of the operand for the same data block between adjacent iterations and address stride of the operand between different data blocks in a single iteration. For details about the address stride of the operand between adjacent iterations, see repeatStride. For details about the address stride of the operand between different data blocks in a single iteration, see dataBlockStride.

Returns

None

Availability

Atlas Training Series Product

Precautions

To save memory space when using high-dimensional tensor sharding computation APIs, you can define a tensor shared by the source and destination operands (by address overlapping). The general instruction restrictions are as follows.
- For a single repeat (repeatTimes = 1), the source operand must completely overlap the destination operand.
- For multiple repeats (repeatTimes > 1), if there is a dependency between the source operand and the destination operand, that is, the destination operand of the Nth iteration is the source operand of the (N + 1)th iteration, address overlapping is not allowed. Address overlapping is supported in the following cases: (1) The data type is half, int32_t, or float, and the destination operand overlaps the second source operand; (2) src1RepStride or dstRepStride is 0; (3) src0 and src1 do not overlap.

When the entire tensor computation API is used for symbol overloading, the computation workload is the total length of the destination LocalTensor.
For details about the alignment requirements of the operand address offset, see General Restrictions.

Example

This example shows only part of the code used in the computation process. If you need to run the sample code, copy the code snippet and replace the Compute function in the two-operand instruction template provided in More Samples.

Example of high-dimensional tensor sharding computation (contiguous mask mode)

        
             uint64_t mask = 128;
// repeatTimes = 4. 128 elements are computed in one iteration, and 512 elements are computed in total.
// dstBlkStride, src0BlkStride, src1BlkStride = 1. Data is continuously read and written in a single iteration.
// dstRepStride, src0RepStride, src1RepStride = 8. Data is continuously read and written between adjacent iterations.
AscendC::Sub(dstLocal, src0Local, src1Local, mask, 4, { 1, 1, 1, 8, 8, 8 });

Example of high-dimensional tensor sharding computation (bitwise mask mode)

        
             uint64_t mask[2] = { UINT64_MAX, UINT64_MAX };
// repeatTimes = 4. 128 elements are computed in one iteration, and 512 elements are computed in total.
// dstBlkStride, src0BlkStride, src1BlkStride = 1. Data is continuously read and written in a single iteration.
// dstRepStride, src0RepStride, src1RepStride = 8. Data is continuously read and written between adjacent iterations.
AscendC::Sub(dstLocal, src0Local, src1Local, mask, 4, { 1, 1, 1, 8, 8, 8 });

Example of computing the first n data elements of a tensor

        
             AscendC::Sub(dstLocal, src0Local, src1Local, 512);

Example computation of the entire tensor
1

dstLocal = src0Local - src1Local;

Result example:

Input (src0Local): [1 2 3 ... 512]
Input (src1Local): [513 514 515 ... 1024]
Output (dstLocal): [-512 -512 -512 ... -512]

Parent topic: Two-Operand Instructions