 |
AI Engine-ML Intrinsics User Guide (v2024.2)
|
These intrinsics allow shifting full vectors. More...
Overview
These intrinsics allow shifting full vectors.
Special shift used to perform unaligned loads | |
| v128int4 | shiftx (v128int4 a, v128int4 b, int step, unsigned int shift) |
Shifts a by 2^step for step >= 1. Then it ORs the result with b right-shifted shift bytes. Each value except of shift are considered to be 64 elements of bytes. When step is zero, then it behaves like a regular shift. The pseudo-code can be described as follows: | |
| v64int8 | shiftx (v64int8 a, v64int8 b, int step, unsigned int shift) |
Shifts a by 2^step for step >= 1. Then it ORs the result with b right-shifted shift bytes. Each value except of shift are considered to be 64 elements of bytes. When step is zero, then it behaves like a regular shift. The pseudo-code can be described as follows: | |
| v32int16 | shiftx (v32int16 a, v32int16 b, int step, unsigned int shift) |
Shifts a by 2^step for step >= 1. Then it ORs the result with b right-shifted shift bytes. Each value except of shift are considered to be 64 elements of bytes. When step is zero, then it behaves like a regular shift. The pseudo-code can be described as follows: | |
| v16int32 | shiftx (v16int32 a, v16int32 b, int step, unsigned int shift) |
Shifts a by 2^step for step >= 1. Then it ORs the result with b right-shifted shift bytes. Each value except of shift are considered to be 64 elements of bytes. When step is zero, then it behaves like a regular shift. The pseudo-code can be described as follows: | |
| v128uint4 | shiftx (v128uint4 a, v128uint4 b, int step, unsigned int shift) |
Shifts a by 2^step for step >= 1. Then it ORs the result with b right-shifted shift bytes. Each value except of shift are considered to be 64 elements of bytes. When step is zero, then it behaves like a regular shift. The pseudo-code can be described as follows: | |
| v64uint8 | shiftx (v64uint8 a, v64uint8 b, int step, unsigned int shift) |
Shifts a by 2^step for step >= 1. Then it ORs the result with b right-shifted shift bytes. Each value except of shift are considered to be 64 elements of bytes. When step is zero, then it behaves like a regular shift. The pseudo-code can be described as follows: | |
| v32uint16 | shiftx (v32uint16 a, v32uint16 b, int step, unsigned int shift) |
Shifts a by 2^step for step >= 1. Then it ORs the result with b right-shifted shift bytes. Each value except of shift are considered to be 64 elements of bytes. When step is zero, then it behaves like a regular shift. The pseudo-code can be described as follows: | |
| v16uint32 | shiftx (v16uint32 a, v16uint32 b, int step, unsigned int shift) |
Shifts a by 2^step for step >= 1. Then it ORs the result with b right-shifted shift bytes. Each value except of shift are considered to be 64 elements of bytes. When step is zero, then it behaves like a regular shift. The pseudo-code can be described as follows: | |
| v16cint16 | shiftx (v16cint16 a, v16cint16 b, int step, unsigned int shift) |
Shifts a by 2^step for step >= 1. Then it ORs the result with b right-shifted shift bytes. Each value except of shift are considered to be 64 elements of bytes. When step is zero, then it behaves like a regular shift. The pseudo-code can be described as follows: | |
| v8cint32 | shiftx (v8cint32 a, v8cint32 b, int step, unsigned int shift) |
Shifts a by 2^step for step >= 1. Then it ORs the result with b right-shifted shift bytes. Each value except of shift are considered to be 64 elements of bytes. When step is zero, then it behaves like a regular shift. The pseudo-code can be described as follows: | |
| v32bfloat16 | shiftx (v32bfloat16 a, v32bfloat16 b, int step, unsigned int shift) |
Shifts a by 2^step for step >= 1. Then it ORs the result with b right-shifted shift bytes. Each value except of shift are considered to be 64 elements of bytes. When step is zero, then it behaves like a regular shift. The pseudo-code can be described as follows: | |
| v16cbfloat16 | shiftx (v16cbfloat16 a, v16cbfloat16 b, int step, int shift) |
Shifts a by 2^step for step >= 1. Then it ORs the result with b right-shifted shift bytes. Each value except of shift are considered to be 64 elements of bytes. When step is zero, then it behaves like a regular shift. The pseudo-code can be described as follows: | |
| v16accfloat | shiftx (v16accfloat a, v16accfloat b, int step, unsigned int shift) |
Shifts a by 2^step for step >= 1. Then it ORs the result with b right-shifted shift bytes. Each value except of shift are considered to be 64 elements of bytes. When step is zero, then it behaves like a regular shift. The pseudo-code can be described as follows: | |
| v16float | shiftx (v16float a, v16float b, int step, unsigned int shift) |
Shifts a by 2^step for step >= 1. Then it ORs the result with b right-shifted shift bytes. Each value except of shift are considered to be 64 elements of bytes. When step is zero, then it behaves like a regular shift. The pseudo-code can be described as follows: | |
| v8cfloat | shiftx (v8cfloat a, v8cfloat b, int step, unsigned int shift) |
Shifts a by 2^step for step >= 1. Then it ORs the result with b right-shifted shift bytes. Each value except of shift are considered to be 64 elements of bytes. When step is zero, then it behaves like a regular shift. The pseudo-code can be described as follows: | |
Lane-by-lane vector shift (in bytes) | |
| v128int4 | shift_bytes (v128int4 a, v128int4 b, unsigned int shift) |
| Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift:shift+64]. | |
| v64int8 | shift_bytes (v64int8 a, v64int8 b, unsigned int shift) |
| Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift:shift+64]. | |
| v32int16 | shift_bytes (v32int16 a, v32int16 b, unsigned int shift) |
| Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift:shift+64]. | |
| v16int32 | shift_bytes (v16int32 a, v16int32 b, unsigned int shift) |
| Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift:shift+64]. | |
| v128uint4 | shift_bytes (v128uint4 a, v128uint4 b, unsigned int shift) |
| Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift:shift+64]. | |
| v64uint8 | shift_bytes (v64uint8 a, v64uint8 b, unsigned int shift) |
| Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift:shift+64]. | |
| v32uint16 | shift_bytes (v32uint16 a, v32uint16 b, unsigned int shift) |
| Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift:shift+64]. | |
| v16uint32 | shift_bytes (v16uint32 a, v16uint32 b, unsigned int shift) |
| Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift:shift+64]. | |
| v16cint16 | shift_bytes (v16cint16 a, v16cint16 b, unsigned int shift) |
| Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift:shift+64]. | |
| v8cint32 | shift_bytes (v8cint32 a, v8cint32 b, unsigned int shift) |
| Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift:shift+64]. | |
| v32bfloat16 | shift_bytes (v32bfloat16 a, v32bfloat16 b, unsigned int shift) |
| Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift:shift+64]. | |
| v16cbfloat16 | shift_bytes (v16cbfloat16 a, v16cbfloat16 b, int shift) |
| Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift:shift+64]. | |
| v16accfloat | shift_bytes (v16accfloat a, v16accfloat b, unsigned int shift) |
| Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift:shift+64]. | |
| v16float | shift_bytes (v16float a, v16float b, unsigned int shift) |
| Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift:shift+64]. | |
| v8cfloat | shift_bytes (v8cfloat a, v8cfloat b, unsigned int shift) |
| Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift:shift+64]. | |
Lane-by-lane vector shift (in elems) | |
| v64int8 | shift (v64int8 a, v64int8 b, unsigned int shift) |
| Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift*elem_size:shift*elem_size+64]. | |
| v32int16 | shift (v32int16 a, v32int16 b, unsigned int shift) |
| Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift*elem_size:shift*elem_size+64]. | |
| v16int32 | shift (v16int32 a, v16int32 b, unsigned int shift) |
| Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift*elem_size:shift*elem_size+64]. | |
| v64uint8 | shift (v64uint8 a, v64uint8 b, unsigned int shift) |
| Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift*elem_size:shift*elem_size+64]. | |
| v32uint16 | shift (v32uint16 a, v32uint16 b, unsigned int shift) |
| Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift*elem_size:shift*elem_size+64]. | |
| v16uint32 | shift (v16uint32 a, v16uint32 b, unsigned int shift) |
| Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift*elem_size:shift*elem_size+64]. | |
| v16cint16 | shift (v16cint16 a, v16cint16 b, unsigned int shift) |
| Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift*elem_size:shift*elem_size+64]. | |
| v8cint32 | shift (v8cint32 a, v8cint32 b, unsigned int shift) |
| Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift*elem_size:shift*elem_size+64]. | |
| v32bfloat16 | shift (v32bfloat16 a, v32bfloat16 b, unsigned int shift) |
| Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift*elem_size:shift*elem_size+64]. | |
| v16cbfloat16 | shift (v16cbfloat16 a, v16cbfloat16 b, int shift) |
| Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift*elem_size:shift*elem_size+64]. | |
| v16accfloat | shift (v16accfloat a, v16accfloat b, unsigned int shift) |
| Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift*elem_size:shift*elem_size+64]. | |
| v16float | shift (v16float a, v16float b, unsigned int shift) |
| Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift*elem_size:shift*elem_size+64]. | |
| v8cfloat | shift (v8cfloat a, v8cfloat b, unsigned int shift) |
| Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift*elem_size:shift*elem_size+64]. | |
Function Documentation
◆ shift() [1/13]
| v16accfloat shift | ( | v16accfloat | a, |
| v16accfloat | b, | ||
| unsigned int | shift | ||
| ) |
Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift*elem_size:shift*elem_size+64].
- Parameters
-
a value to be concatenated b value to be concatenated shift number of elements to be shifted
- Returns
- shifted vector
◆ shift() [2/13]
| v16cbfloat16 shift | ( | v16cbfloat16 | a, |
| v16cbfloat16 | b, | ||
| int | shift | ||
| ) |
Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift*elem_size:shift*elem_size+64].
- Parameters
-
a value to be concatenated b value to be concatenated shift number of elements to be shifted
- Returns
- shifted vector
◆ shift() [3/13]
Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift*elem_size:shift*elem_size+64].
- Parameters
-
a value to be concatenated b value to be concatenated shift number of elements to be shifted
- Returns
- shifted vector
◆ shift() [4/13]
Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift*elem_size:shift*elem_size+64].
- Parameters
-
a value to be concatenated b value to be concatenated shift number of elements to be shifted
- Returns
- shifted vector
◆ shift() [5/13]
Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift*elem_size:shift*elem_size+64].
- Parameters
-
a value to be concatenated b value to be concatenated shift number of elements to be shifted
- Returns
- shifted vector
◆ shift() [6/13]
Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift*elem_size:shift*elem_size+64].
- Parameters
-
a value to be concatenated b value to be concatenated shift number of elements to be shifted
- Returns
- shifted vector
◆ shift() [7/13]
| v32bfloat16 shift | ( | v32bfloat16 | a, |
| v32bfloat16 | b, | ||
| unsigned int | shift | ||
| ) |
Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift*elem_size:shift*elem_size+64].
- Parameters
-
a value to be concatenated b value to be concatenated shift number of elements to be shifted
- Returns
- shifted vector
◆ shift() [8/13]
Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift*elem_size:shift*elem_size+64].
- Parameters
-
a value to be concatenated b value to be concatenated shift number of elements to be shifted
- Returns
- shifted vector
◆ shift() [9/13]
Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift*elem_size:shift*elem_size+64].
- Parameters
-
a value to be concatenated b value to be concatenated shift number of elements to be shifted
- Returns
- shifted vector
◆ shift() [10/13]
Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift*elem_size:shift*elem_size+64].
- Parameters
-
a value to be concatenated b value to be concatenated shift number of elements to be shifted
- Returns
- shifted vector
◆ shift() [11/13]
Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift*elem_size:shift*elem_size+64].
- Parameters
-
a value to be concatenated b value to be concatenated shift number of elements to be shifted
- Returns
- shifted vector
◆ shift() [12/13]
Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift*elem_size:shift*elem_size+64].
- Parameters
-
a value to be concatenated b value to be concatenated shift number of elements to be shifted
- Returns
- shifted vector
◆ shift() [13/13]
Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift*elem_size:shift*elem_size+64].
- Parameters
-
a value to be concatenated b value to be concatenated shift number of elements to be shifted
- Returns
- shifted vector
◆ shift_bytes() [1/15]
Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift:shift+64].
- Parameters
-
a value to be concatenated b value to be concatenated shift number of bytes to be shifted
- Returns
- shifted vector
◆ shift_bytes() [2/15]
Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift:shift+64].
- Parameters
-
a value to be concatenated b value to be concatenated shift number of bytes to be shifted
- Returns
- shifted vector
◆ shift_bytes() [3/15]
| v16accfloat shift_bytes | ( | v16accfloat | a, |
| v16accfloat | b, | ||
| unsigned int | shift | ||
| ) |
Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift:shift+64].
- Parameters
-
a value to be concatenated b value to be concatenated shift number of bytes to be shifted
- Returns
- shifted vector
◆ shift_bytes() [4/15]
| v16cbfloat16 shift_bytes | ( | v16cbfloat16 | a, |
| v16cbfloat16 | b, | ||
| int | shift | ||
| ) |
Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift:shift+64].
- Parameters
-
a value to be concatenated b value to be concatenated shift number of bytes to be shifted
- Returns
- shifted vector
◆ shift_bytes() [5/15]
Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift:shift+64].
- Parameters
-
a value to be concatenated b value to be concatenated shift number of bytes to be shifted
- Returns
- shifted vector
◆ shift_bytes() [6/15]
Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift:shift+64].
- Parameters
-
a value to be concatenated b value to be concatenated shift number of bytes to be shifted
- Returns
- shifted vector
◆ shift_bytes() [7/15]
Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift:shift+64].
- Parameters
-
a value to be concatenated b value to be concatenated shift number of bytes to be shifted
- Returns
- shifted vector
◆ shift_bytes() [8/15]
Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift:shift+64].
- Parameters
-
a value to be concatenated b value to be concatenated shift number of bytes to be shifted
- Returns
- shifted vector
◆ shift_bytes() [9/15]
| v32bfloat16 shift_bytes | ( | v32bfloat16 | a, |
| v32bfloat16 | b, | ||
| unsigned int | shift | ||
| ) |
Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift:shift+64].
- Parameters
-
a value to be concatenated b value to be concatenated shift number of bytes to be shifted
- Returns
- shifted vector
◆ shift_bytes() [10/15]
Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift:shift+64].
- Parameters
-
a value to be concatenated b value to be concatenated shift number of bytes to be shifted
- Returns
- shifted vector
◆ shift_bytes() [11/15]
Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift:shift+64].
- Parameters
-
a value to be concatenated b value to be concatenated shift number of bytes to be shifted
- Returns
- shifted vector
◆ shift_bytes() [12/15]
Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift:shift+64].
- Parameters
-
a value to be concatenated b value to be concatenated shift number of bytes to be shifted
- Returns
- shifted vector
◆ shift_bytes() [13/15]
Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift:shift+64].
- Parameters
-
a value to be concatenated b value to be concatenated shift number of bytes to be shifted
- Returns
- shifted vector
◆ shift_bytes() [14/15]
Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift:shift+64].
- Parameters
-
a value to be concatenated b value to be concatenated shift number of bytes to be shifted
- Returns
- shifted vector
◆ shift_bytes() [15/15]
Concatenates a and b, interprets them as a vector of 128 bytes and returns a::b[shift:shift+64].
- Parameters
-
a value to be concatenated b value to be concatenated shift number of bytes to be shifted
- Returns
- shifted vector
◆ shiftx() [1/15]
Shifts a by 2^step for step >= 1. Then it ORs the result with b right-shifted shift bytes. Each value except of shift are considered to be 64 elements of bytes. When step is zero, then it behaves like a regular shift. The pseudo-code can be described as follows:
- Parameters
-
a value to be concatenated b value to be concatenated step amount of preshift on a (0: 0 bits, 1: 32 bits, 2: 64 bits, 3: 128 bits, 4: 256 bits) shift number of bytes to shift b
- Returns
- shifted vector
◆ shiftx() [2/15]
Shifts a by 2^step for step >= 1. Then it ORs the result with b right-shifted shift bytes. Each value except of shift are considered to be 64 elements of bytes. When step is zero, then it behaves like a regular shift. The pseudo-code can be described as follows:
- Parameters
-
a value to be concatenated b value to be concatenated step amount of preshift on a (0: 0 bits, 1: 32 bits, 2: 64 bits, 3: 128 bits, 4: 256 bits) shift number of bytes to shift b
- Returns
- shifted vector
◆ shiftx() [3/15]
| v16accfloat shiftx | ( | v16accfloat | a, |
| v16accfloat | b, | ||
| int | step, | ||
| unsigned int | shift | ||
| ) |
Shifts a by 2^step for step >= 1. Then it ORs the result with b right-shifted shift bytes. Each value except of shift are considered to be 64 elements of bytes. When step is zero, then it behaves like a regular shift. The pseudo-code can be described as follows:
- Parameters
-
a value to be concatenated b value to be concatenated step amount of preshift on a (0: 0 bits, 1: 32 bits, 2: 64 bits, 3: 128 bits, 4: 256 bits) shift number of bytes to shift b
- Returns
- shifted vector
◆ shiftx() [4/15]
| v16cbfloat16 shiftx | ( | v16cbfloat16 | a, |
| v16cbfloat16 | b, | ||
| int | step, | ||
| int | shift | ||
| ) |
Shifts a by 2^step for step >= 1. Then it ORs the result with b right-shifted shift bytes. Each value except of shift are considered to be 64 elements of bytes. When step is zero, then it behaves like a regular shift. The pseudo-code can be described as follows:
- Parameters
-
a value to be concatenated b value to be concatenated step amount of preshift on a (0: 0 bits, 1: 32 bits, 2: 64 bits, 3: 128 bits, 4: 256 bits) shift number of bytes to shift b
- Returns
- shifted vector
◆ shiftx() [5/15]
Shifts a by 2^step for step >= 1. Then it ORs the result with b right-shifted shift bytes. Each value except of shift are considered to be 64 elements of bytes. When step is zero, then it behaves like a regular shift. The pseudo-code can be described as follows:
- Parameters
-
a value to be concatenated b value to be concatenated step amount of preshift on a (0: 0 bits, 1: 32 bits, 2: 64 bits, 3: 128 bits, 4: 256 bits) shift number of bytes to shift b
- Returns
- shifted vector
◆ shiftx() [6/15]
Shifts a by 2^step for step >= 1. Then it ORs the result with b right-shifted shift bytes. Each value except of shift are considered to be 64 elements of bytes. When step is zero, then it behaves like a regular shift. The pseudo-code can be described as follows:
- Parameters
-
a value to be concatenated b value to be concatenated step amount of preshift on a (0: 0 bits, 1: 32 bits, 2: 64 bits, 3: 128 bits, 4: 256 bits) shift number of bytes to shift b
- Returns
- shifted vector
◆ shiftx() [7/15]
Shifts a by 2^step for step >= 1. Then it ORs the result with b right-shifted shift bytes. Each value except of shift are considered to be 64 elements of bytes. When step is zero, then it behaves like a regular shift. The pseudo-code can be described as follows:
- Parameters
-
a value to be concatenated b value to be concatenated step amount of preshift on a (0: 0 bits, 1: 32 bits, 2: 64 bits, 3: 128 bits, 4: 256 bits) shift number of bytes to shift b
- Returns
- shifted vector
◆ shiftx() [8/15]
Shifts a by 2^step for step >= 1. Then it ORs the result with b right-shifted shift bytes. Each value except of shift are considered to be 64 elements of bytes. When step is zero, then it behaves like a regular shift. The pseudo-code can be described as follows:
- Parameters
-
a value to be concatenated b value to be concatenated step amount of preshift on a (0: 0 bits, 1: 32 bits, 2: 64 bits, 3: 128 bits, 4: 256 bits) shift number of bytes to shift b
- Returns
- shifted vector
◆ shiftx() [9/15]
| v32bfloat16 shiftx | ( | v32bfloat16 | a, |
| v32bfloat16 | b, | ||
| int | step, | ||
| unsigned int | shift | ||
| ) |
Shifts a by 2^step for step >= 1. Then it ORs the result with b right-shifted shift bytes. Each value except of shift are considered to be 64 elements of bytes. When step is zero, then it behaves like a regular shift. The pseudo-code can be described as follows:
- Parameters
-
a value to be concatenated b value to be concatenated step amount of preshift on a (0: 0 bits, 1: 32 bits, 2: 64 bits, 3: 128 bits, 4: 256 bits) shift number of bytes to shift b
- Returns
- shifted vector
◆ shiftx() [10/15]
Shifts a by 2^step for step >= 1. Then it ORs the result with b right-shifted shift bytes. Each value except of shift are considered to be 64 elements of bytes. When step is zero, then it behaves like a regular shift. The pseudo-code can be described as follows:
- Parameters
-
a value to be concatenated b value to be concatenated step amount of preshift on a (0: 0 bits, 1: 32 bits, 2: 64 bits, 3: 128 bits, 4: 256 bits) shift number of bytes to shift b
- Returns
- shifted vector
◆ shiftx() [11/15]
Shifts a by 2^step for step >= 1. Then it ORs the result with b right-shifted shift bytes. Each value except of shift are considered to be 64 elements of bytes. When step is zero, then it behaves like a regular shift. The pseudo-code can be described as follows:
- Parameters
-
a value to be concatenated b value to be concatenated step amount of preshift on a (0: 0 bits, 1: 32 bits, 2: 64 bits, 3: 128 bits, 4: 256 bits) shift number of bytes to shift b
- Returns
- shifted vector
◆ shiftx() [12/15]
Shifts a by 2^step for step >= 1. Then it ORs the result with b right-shifted shift bytes. Each value except of shift are considered to be 64 elements of bytes. When step is zero, then it behaves like a regular shift. The pseudo-code can be described as follows:
- Parameters
-
a value to be concatenated b value to be concatenated step amount of preshift on a (0: 0 bits, 1: 32 bits, 2: 64 bits, 3: 128 bits, 4: 256 bits) shift number of bytes to shift b
- Returns
- shifted vector
◆ shiftx() [13/15]
Shifts a by 2^step for step >= 1. Then it ORs the result with b right-shifted shift bytes. Each value except of shift are considered to be 64 elements of bytes. When step is zero, then it behaves like a regular shift. The pseudo-code can be described as follows:
- Parameters
-
a value to be concatenated b value to be concatenated step amount of preshift on a (0: 0 bits, 1: 32 bits, 2: 64 bits, 3: 128 bits, 4: 256 bits) shift number of bytes to shift b
- Returns
- shifted vector
◆ shiftx() [14/15]
Shifts a by 2^step for step >= 1. Then it ORs the result with b right-shifted shift bytes. Each value except of shift are considered to be 64 elements of bytes. When step is zero, then it behaves like a regular shift. The pseudo-code can be described as follows:
- Parameters
-
a value to be concatenated b value to be concatenated step amount of preshift on a (0: 0 bits, 1: 32 bits, 2: 64 bits, 3: 128 bits, 4: 256 bits) shift number of bytes to shift b
- Returns
- shifted vector
◆ shiftx() [15/15]
Shifts a by 2^step for step >= 1. Then it ORs the result with b right-shifted shift bytes. Each value except of shift are considered to be 64 elements of bytes. When step is zero, then it behaves like a regular shift. The pseudo-code can be described as follows:
- Parameters
-
a value to be concatenated b value to be concatenated step amount of preshift on a (0: 0 bits, 1: 32 bits, 2: 64 bits, 3: 128 bits, 4: 256 bits) shift number of bytes to shift b
- Returns
- shifted vector