ShiftRight

Applicability

Product	Supported
Atlas A3 training products / Atlas A3 inference products	√
Atlas A2 training products / Atlas A2 inference products	√
Atlas 200I/500 A2 inference products	√
Atlas inference product 's AI Core	x
Atlas inference product 's Vector Core	x
Atlas training products	x

Function

Performs right shift on the source operand element-wise. The shift distance is determined by the scalar in the formula. The right shift operation is classified into the following types based on the data type of the source operand:

If the data type is unsigned, logical right shift is performed. In this case, the binary number is shifted right by a specified number of bits. The least significant bit is discarded, and the most significant bit is padded with 0. For example, after the binary number 1010101010101010 (uint16_t) is logically shifted right by one bit, the result is 0101010101010101.
If the data type is signed, arithmetic right shift is performed. In this case, the binary number is shifted right by a specified number of bits. The least significant bit is discarded, and the sign bit is copied to the most significant bit. For example, after the binary number 1010101010101010 (int16_t) is arithmetically shifted right by one bit, the result is 1101010101010101; after the binary number is arithmetically shifted left by three bits, the result is 1111010101010101.

$\text{[math]}$

Prototype

Computation of the first n data elements of a tensor

        
             template <typename T, bool isSetMask = true>
__aicore__ inline void ShiftRight(const LocalTensor<T>& dst, const LocalTensor<T>& src, const T& scalarValue, const int32_t& count)

High-dimensional tensor sharding computation

Bitwise mask mode

          
               template <typename T, bool isSetMask = true>
__aicore__ inline void ShiftRight(const LocalTensor<T>& dst, const LocalTensor<T>& src, const T& scalarValue, uint64_t mask[], const uint8_t repeatTime, const UnaryRepeatParams& repeatParams, bool roundEn = false)

Contiguous mask mode

          
               template <typename T, bool isSetMask = true>
__aicore__ inline void ShiftRight(const LocalTensor<T>& dst, const LocalTensor<T>& src, const T& scalarValue, uint64_t mask, const uint8_t repeatTime, const UnaryRepeatParams& repeatParams, bool roundEn = false)

When dst and src use the TensorTrait type, the data type of scalarValue (corresponding to the LiteType in TensorTrait) may differ from that of TensorTrait. To address this, the template type U is introduced to represent the data type of scalarValue. std::enable_if is used to check whether LiteType extracted from T is the same as U. If they match, the API compiles successfully; otherwise, the compilation fails. The API prototype is defined as follows:

Computation of the first n data elements of a tensor

        
             template <typename T, typename U, bool isSetMask = true, typename Std::enable_if<Std::is_same<PrimT<T>, U>::value, bool>::type = true>
__aicore__ inline void ShiftRight(const LocalTensor<T>& dst, const LocalTensor<T>& src, const U& scalarValue, const int32_t& count)

High-dimensional tensor sharding computation

Bitwise mask mode

          
               template <typename T, typename U, bool isSetMask = true, typename Std::enable_if<Std::is_same<PrimT<T>, U>::value, bool>::type = true>
__aicore__ inline void ShiftRight(const LocalTensor<T>& dst, const LocalTensor<T>& src, const U& scalarValue, uint64_t mask[], const uint8_t repeatTime, const UnaryRepeatParams& repeatParams, bool roundEn)

Contiguous mask mode

          
               template <typename T, typename U, bool isSetMask = true, typename Std::enable_if<Std::is_same<PrimT<T>, U>::value, bool>::type = true>
__aicore__ inline void ShiftRight(const LocalTensor<T>& dst, const LocalTensor<T>& src, const U& scalarValue, uint64_t mask, const uint8_t repeatTime, const UnaryRepeatParams& repeatParams, bool roundEn)

Parameters

**Table 1** Template parameters
Parameter	Description
T	Data type of the operand. For Atlas A2 training products / Atlas A2 inference products , the supported data types are uint16_t, int16_t, uint32_t, and int32_t. For Atlas A3 training products / Atlas A3 inference products , the supported data types are uint16_t, int16_t, uint32_t, and int32_t. For Atlas 200I/500 A2 inference products , the supported data types are uint16_t, int16_t, uint32_t, and int32_t.
U	Data type of scalarValue. For Atlas A2 training products / Atlas A2 inference products , the supported data types are uint16_t, int16_t, uint32_t, and int32_t. For Atlas A3 training products / Atlas A3 inference products , the supported data types are uint16_t, int16_t, uint32_t, and int32_t. For Atlas 200I/500 A2 inference products , the supported data types are uint16_t, int16_t, uint32_t, and int32_t.
isSetMask	Whether to set the mask mode and mask value inside the API. true: Set inside the API. The APIs for high-dimensional tensor sharding and the APIs for computing the first n data elements of a tensor internally use the normal mode or counter mode of the mask. In general, keep isSetMask as the default value, which allows the API to automatically set the mask mode and mask value based on the mask or count parameter passed in by developers. false: Set outside the API. For high-dimensional tensor sharding computation APIs, in scenarios with high performance requirements, you need to use SetMaskNorm/SetMaskCount to set the mask mode and use SetVectorMask to set the mask value. The mask input parameter of this API must be set to MASK_PLACEHOLDER. For APIs that compute the first n data elements of a tensor, in scenarios with high performance requirements, you need to set the mask mode to counter mode using SetMaskCount and set the mask value using SetVectorMask. The count parameter in this API does not take effect. You are advised to set it to 1. For the following models, the isSetMask parameter in the API for computing the first n data elements of a tensor does not take effect. Keep it as the default value. Atlas 200I/500 A2 inference products

**Table 2** Parameters
Parameter	Input/Output	Description
dst	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.
src	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. Its data type must be the same as that of the destination operand.
scalarValue	Input	Right shift distance. Its data type must be the same as that of the tensor element in the destination operand. For Atlas A2 training products / Atlas A2 inference products , when src is of type uint16_t or int16_t, the value range of scalarValue is [0, 16]; when src is of type uint32_t or int32_t, the value range of scalarValue is [0, 32]. For Atlas A3 training products / Atlas A3 inference products , when src is of type uint16_t or int16_t, the value range of scalarValue is [0, 16]; when src is of type uint32_t or int32_t, the value range of scalarValue is [0, 32]. For Atlas 200I/500 A2 inference products , when src is of type uint16_t or int16_t, the value range of scalarValue is [0, 16]; when src is of type uint32_t or int32_t, the value range of scalarValue is [0, 32].
count	Input	Number of elements involved in the computation.
mask/mask[]	Input	mask is used to control the elements that participate in computation in each iteration. Bitwise mode: controls which elements are involved in computation bit by bit. A bit value of 1 means the corresponding element participates in computation, while 0 means it does not. The mask value is an array. The array length and the value range of the array elements are related to the operand data type. When the operand is 16-bit, the array length is 2, mask[0] and mask[1] ∈ [0, 2⁶⁴ -1] and cannot be 0 at the same time. When the operand is 32-bit, the array length is 1 and mask[0] ∈ (0, 2⁶⁴ – 1]. When the operand is 64-bit, the array length is 1 and mask[0] ∈ (0, 2³² – 1]. For example, if mask = [0, 8] and 8 = 0b1000, only the fourth element participates in computation. 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 repeat 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].
repeatTime	Input	Number of iteration repeats. The vector compute unit reads 256 bytes of contiguous data for computation each time. To process the input data, the data needs to be read and computed over multiple repeats. repeatTime indicates the number of repeats. For details about this parameter, see High-dimensional Sharding APIs.
repeatParams	Input	Structure for controlling element-wise operations. For details, see UnaryRepeatParams.
roundEn	Input	A bool. Set to True to enable the rounding function. Set to False to disable the rounding function. This parameter is valid only when src is of type int16_t or int32_t. For example, with the rounding function enabled and src of type int16_t, the 5-bit arithmetic right shift of src is computed as follows: src_ele = 17 = 0b0000000000010001 (the fifth bit is 1) dst_ele = arithmetic_right_shift(src_ele, 5) + 1 = 0b0000000000000000 + 1 = 0b0000000000000001 For Atlas 200I/500 A2 inference products , the rounding function cannot be enabled and the value can only be false.

Returns

None

Restrictions

For details about the operand address alignment requirements, see General Address Alignment Restrictions.
For details about the constraints on operand address overlapping, see General Address Overlapping Restrictions.

Example

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

        
             uint64_t mask = 128;
int16_t scalar = 2;
// repeatTime = 4. 128 elements are processed in a single iteration. To compute 512 elements, four iterations are required.
// dstBlkStride, srcBlkStride = 1. The interval between addresses of src0 data involved in computation within each iteration is one data block, indicating that data is read and written continuously within a single iteration.
// dstRepStride, srcRepStride = 8. The interval between addresses of adjacent iterations is eight data blocks, indicating that data is continuously read and written between adjacent iterations.
AscendC::ShiftRight(dstLocal, srcLocal, scalar, mask, 4, { 1, 1, 8, 8 }, false);

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

        
             uint64_t mask[2] = { UINT64_MAX, UINT64_MAX };
int16_t scalar = 2;
// repeatTime = 4. 128 elements are processed in a single iteration. To compute 512 elements, four iterations are required.
// dstBlkStride, srcBlkStride = 1. The interval between addresses of src0 data involved in computation within each iteration is one data block, indicating that data is read and written continuously within a single iteration.
// dstRepStride, srcRepStride = 8. The interval between addresses of adjacent iterations is eight data blocks, indicating that data is continuously read and written between adjacent iterations.
AscendC::ShiftRight(dstLocal, srcLocal, scalar, mask, 4, {1, 1, 8, 8}, false);

Example of computing the first n data elements of a tensor

        
             int16_t scalar = 2;
AscendC::ShiftRight(dstLocal, srcLocal, scalar, 512);

Result example:

Input (srcLocal): [1 2 3 ... 512]
Input (scalar): 2
Output (dstLocal): [0 0 0 1 1 1 1 ... 128]

Parent topic: Logic-based Computation