Shuffle

Intrinsics allowing you perform vector shuffles. More...

Overview

Intrinsics allowing you perform vector shuffles.

This table gives the different shuffle modes. For an illustration see Illustration of Shuffle Modes.

Element Size	Matrix Transpose Dimensions	Mode Value
		Forward operation (Deinterleaving)		Backward Operation (Interleaving)
		extract low	extract high	extract low	extract high
8-bit	64x2 → 2x64	shuffle_T8_64x2_lo	shuffle_T8_64x2_hi	shuffle_T8_2x64_lo	shuffle_T8_2x64_hi
16-bit	32x2 → 2x32	shuffle_T16_32x2_lo	shuffle_T16_32x2_hi	shuffle_T16_2x32_lo	shuffle_T16_2x32_hi
32-bit	16x2 → 2x16	shuffle_T32_16x2_lo	shuffle_T32_16x2_hi	shuffle_T32_2x16_lo	shuffle_T32_2x16_hi
64-bit	8x2 → 2x8	shuffle_T64_8x2_lo	shuffle_T64_8x2_hi	shuffle_T64_2x8_lo	shuffle_T64_2x8_hi
128-bit	4x2 → 2x4	shuffle_T128_4x2_lo	shuffle_T128_4x2_hi	shuffle_T128_2x4_lo	shuffle_T128_2x4_hi
256-bit	2x2 → 2x2	shuffle_T256_2x2_lo	shuffle_T256_2x2_hi	shuffle_T256_2x2_lo	shuffle_T256_2x2_hi
512-bit	1x2 → 2x1	shuffle_T512_1x2_lo	shuffle_T512_1x2_hi	shuffle_T512_1x2_lo	shuffle_T512_1x2_hi
16-bit	16x4 → 4x16	shuffle_T16_16x4_lo	shuffle_T16_16x4_hi	shuffle_T16_4x16_lo	shuffle_T16_4x16_hi
16-bit	8x4 → 4x8	shuffle_T16_8x4	-	shuffle_T16_4x8	-
32-bit	8x4 → 4x8	shuffle_T32_8x4_lo	shuffle_T32_8x4_hi	shuffle_T32_4x8_lo	shuffle_T32_4x8_hi
32-bit	4x4 → 4x4	shuffle_T32_4x4	-	shuffle_T32_4x4	-
8-bit	8x8 → 8x8	shuffle_T8_8x8	-	shuffle_T32_8x8	-
8-bit	16x4 → 4x16	shuffle_T8_16x4	-	shuffle_T8_4x16	-
16-bit	16x[a,b] → 16x[b,a]	shuffle_T16_1x2_flip	-	shuffle_T16_1x2_flip	-
16-bit	4x4 → 4x4	shuffle_T16_4x4	-	shuffle_T16_4x4	-
16-bit	4x2 → 2x4	shuffle_T16_4x2	-	shuffle_T16_2x4	-
16-bit	8x2 → 2x8	shuffle_T16_8x2	-	shuffle_T16_2x8	-
16-bit	16x2 → 2x16	shuffle_T16_16x2	-	shuffle_T16_2x16	-
8-bit	8x4 → 4x8	shuffle_T8_8x4	-	shuffle_T8_4x8	-

Modules
	Illustration of Shuffle Modes

Enumerations
enum	eShuffleMode {   shuffle_T8_64x2_lo = 0 , shuffle_T8_64x2_hi = 1 , shuffle_T16_32x2_lo = 2 , shuffle_T16_32x2_hi = 3 ,   shuffle_T32_16x2_lo = 4 , shuffle_T32_16x2_hi = 5 , shuffle_T64_8x2_lo = 6 , shuffle_T64_8x2_hi = 7 ,   shuffle_T128_4x2_lo = 8 , shuffle_T128_4x2_hi = 9 , shuffle_T256_2x2_lo = 10 , shuffle_T256_2x2_hi = 11 ,   shuffle_T128_2x4_lo = 12 , shuffle_T128_2x4_hi = 13 , shuffle_T64_2x8_lo = 14 , shuffle_T64_2x8_hi = 15 ,   shuffle_T32_2x16_lo = 16 , shuffle_T32_2x16_hi = 17 , shuffle_T16_2x32_lo = 18 , shuffle_T16_2x32_hi = 19 ,   shuffle_T8_2x64_lo = 20 , shuffle_T8_2x64_hi = 21 , shuffle_T512_1x2_lo = 22 , shuffle_T512_1x2_hi = 23 ,   shuffle_T16_16x4_lo = 24 , shuffle_T16_16x4_hi = 25 , shuffle_T16_4x16_lo = 26 , shuffle_T16_4x16_hi = 27 ,   shuffle_T16_8x4 = 28 , shuffle_T16_4x8 = 29 , shuffle_T32_8x4_lo = 30 , shuffle_T32_8x4_hi = 31 ,   shuffle_T32_4x8_lo = 32 , shuffle_T32_4x8_hi = 33 , shuffle_T32_4x4 = 34 , shuffle_T8_8x8 = 35 ,   shuffle_T8_16x4 = 36 , shuffle_T8_4x16 = 37 , shuffle_T16_1x2_flip = 38 , shuffle_T16_4x4 = 39 ,   shuffle_T16_4x2 = 40 , shuffle_T16_2x4 = 41 , shuffle_T16_8x2 = 42 , shuffle_T16_2x8 = 43 ,   shuffle_T16_16x2 = 44 , shuffle_T16_2x16 = 45 , shuffle_T8_8x4 = 46 , shuffle_T8_4x8 = 47 }
	Shuffle modes. More...

Shuffle two vectors
Shuffle two vectors Parameters a First vector to shuffle b Second vector to shuffle mode Shuffling mode (of type eShuffleMode) Returns shuffled vector
v16int32	shuffle (v16int32, v16int32, unsigned int)
v32int16	shuffle (v32int16, v32int16, unsigned int)
v64int8	shuffle (v64int8, v64int8, unsigned int)
v128int4	shuffle (v128int4, v128int4, unsigned int)
v16uint32	shuffle (v16uint32, v16uint32, unsigned int)
v32uint16	shuffle (v32uint16, v32uint16, unsigned int)
v64uint8	shuffle (v64uint8, v64uint8, unsigned int)
v128uint4	shuffle (v128uint4, v128uint4, unsigned int)
v8cint32	shuffle (v8cint32, v8cint32, unsigned int)
v16cint16	shuffle (v16cint16, v16cint16, unsigned int)
v32bfloat16	shuffle (v32bfloat16, v32bfloat16, unsigned int)
v16cbfloat16	shuffle (v16cbfloat16, v16cbfloat16, unsigned int)
v16float	shuffle (v16float, v16float, unsigned int)
v8cfloat	shuffle (v8cfloat, v8cfloat, unsigned int)

Shuffle one vector
Shuffle one vector Parameters a Vector to shuffle mode Shuffling mode (of type eShuffleMode) Returns shuffled vector
v16int32	shuffle (v16int32 a, unsigned int mode)
v32int16	shuffle (v32int16 a, unsigned int mode)
v64int8	shuffle (v64int8 a, unsigned int mode)
v128int4	shuffle (v128int4 a, unsigned int mode)
v16uint32	shuffle (v16uint32 a, unsigned int mode)
v32uint16	shuffle (v32uint16 a, unsigned int mode)
v64uint8	shuffle (v64uint8 a, unsigned int mode)
v128uint4	shuffle (v128uint4 a, unsigned int mode)
v8cint32	shuffle (v8cint32 a, unsigned int mode)
v16cint16	shuffle (v16cint16 a, unsigned int mode)
v32bfloat16	shuffle (v32bfloat16 a, unsigned int mode)
v16cbfloat16	shuffle (v16cbfloat16 a, unsigned int mode)
v16float	shuffle (v16float a, unsigned int mode)
v8cfloat	shuffle (v8cfloat a, unsigned int mode)

Updating all elements with same value and shuffle
Broadcasts input value to all vector lanes Parameters b value to be broadcasted
v64int8	shuffle_s8 (int b, unsigned int m)
v32int16	shuffle_s16 (int b, unsigned int m)
v16int32	shuffle_s32 (int b, unsigned int m)
v16int32	shuffle_v2s32 (v2int32 b, unsigned int m)
v64uint8	shuffle_u8 (unsigned int b, unsigned int m)
v32uint16	shuffle_u16 (unsigned int b, unsigned int m)
v16uint32	shuffle_u32 (unsigned int b, unsigned int m)
v16uint32	shuffle_v2u32 (v2uint32 b, unsigned int m)
v16cint16	shuffle_c16 (cint16 b, unsigned int m)
v32bfloat16	shuffle_bfloat16 (bfloat16 b, unsigned int m)
v16float	shuffle_float (float b, unsigned int m)
v16int32	shuffle_s64 (long long b, unsigned int m)
v16uint32	shuffle_u64 (unsigned long long b, unsigned int m)
v8cint32	shuffle_c32 (cint32 b, unsigned int m)

Shuffle a sparse vector
Shuffle a sparse vector Parameters a Vector to shuffle mode Shuffling mode (of type eShuffleMode) Returns shuffled vector
v256int4_sparse	shuffle (v256int4_sparse qx, int itlv)
v128int8_sparse	shuffle (v128int8_sparse qx, int itlv)
v64int16_sparse	shuffle (v64int16_sparse qx, int itlv)
v256uint4_sparse	shuffle (v256uint4_sparse qx, int itlv)
v128uint8_sparse	shuffle (v128uint8_sparse qx, int itlv)
v64uint16_sparse	shuffle (v64uint16_sparse qx, int itlv)
v64bfloat16_sparse	shuffle (v64bfloat16_sparse qx, int itlv)

Enumeration Type Documentation

eShuffleMode

enum eShuffleMode

Shuffle modes.

Definition of valid modes for vector shuffle

Enumerator
shuffle_T8_64x2_lo	Transpose 64x2 matrix of 8 bit values. Extract low 512 bits of result. See illustration.
shuffle_T8_64x2_hi	Transpose 64x2 matrix of 8 bit values. Extract high 512 bits of result. See illustration.
shuffle_T16_32x2_lo	Transpose 32x2 matrix of 16 bit values. Extract low 512 bits of result. See illustration.
shuffle_T16_32x2_hi	Transpose 32x2 matrix of 16 bit values. Extract high 512 bits of result. See illustration.
shuffle_T32_16x2_lo	Transpose 16x2 matrix of 32 bit values. Extract low 512 bits of result. See illustration.
shuffle_T32_16x2_hi	Transpose 16x2 matrix of 32 bit values. Extract high 512 bits of result. See illustration.
shuffle_T64_8x2_lo	Transpose 8x2 matrix of 64 bit values. Extract low 512 bits of result. See illustration.
shuffle_T64_8x2_hi	Transpose 8x2 matrix of 64 bit values. Extract high 512 bits of result. See illustration.
shuffle_T128_4x2_lo	Transpose 4x2 matrix of 128 bit values. Extract low 512 bits of result. See illustration.
shuffle_T128_4x2_hi	Transpose 4x2 matrix of 128 bit values. Extract high 512 bits of result. See illustration.
shuffle_T256_2x2_lo	Transpose 2x2 matrix of 256 bit values. Extract low 512 bits of result. See illustration.
shuffle_T256_2x2_hi	Transpose 2x2 matrix of 256 bit values. Extract high 512 bits of result. See illustration.
shuffle_T128_2x4_lo	Transpose 2x4 matrix of 128 bit values. Extract low 512 bits of result. See illustration.
shuffle_T128_2x4_hi	Transpose 2x4 matrix of 128 bit values. Extract high 512 bits of result. See illustration.
shuffle_T64_2x8_lo	Transpose 2x8 matrix of 64 bit values. Extract low 512 bits of result. See illustration.
shuffle_T64_2x8_hi	Transpose 2x8 matrix of 64 bit values. Extract high 512 bits of result. See illustration.
shuffle_T32_2x16_lo	Transpose 2x16 matrix of 32 bit values. Extract low 512 bits of result. See illustration.
shuffle_T32_2x16_hi	Transpose 2x16 matrix of 32 bit values. Extract high 512 bits of result. See illustration.
shuffle_T16_2x32_lo	Transpose 2x32 matrix of 16 bit values. Extract low 512 bits of result. See illustration.
shuffle_T16_2x32_hi	Transpose 2x32 matrix of 16 bit values. Extract high 512 bits of result. See illustration.
shuffle_T8_2x64_lo	Transpose 2x64 matrix of 8 bit values. Extract low 512 bits of result. See illustration.
shuffle_T8_2x64_hi	Transpose 2x64 matrix of 8 bit values. Extract high 512 bits of result. See illustration.
shuffle_T512_1x2_lo	Transpose 1x2 matrix of 512 bit values. Extract low 512 bits of result. See illustration.
shuffle_T512_1x2_hi	Transpose 1x2 matrix of 512 bit values. Extract high 512 bits of result. See illustration.
shuffle_T16_16x4_lo	Transpose 16x4 matrix of 16 bit values. Extract low 512 bits of result. See illustration.
shuffle_T16_16x4_hi	Transpose 16x4 matrix of 16 bit values. Extract high 512 bits of result. See illustration.
shuffle_T16_4x16_lo	Transpose 4x16 matrix of 16 bit values. Extract low 512 bits of result. See illustration.
shuffle_T16_4x16_hi	Transpose 4x16 matrix of 16 bit values. Extract high 512 bits of result. See illustration.
shuffle_T16_8x4	Transpose 8x4 matrix of 16 bit values. Extract resulting 512 bit vector. See illustration.
shuffle_T16_4x8	Transpose 4x8 matrix of 16 bit values. Extract resulting 512 bit vector. See illustration.
shuffle_T32_8x4_lo	Transpose 8x4 matrix of 32 bit values. Extract low 512 bits of result. See illustration.
shuffle_T32_8x4_hi	Transpose 8x4 matrix of 32 bit values. Extract high 512 bits of result. See illustration.
shuffle_T32_4x8_lo	Transpose 4x8 matrix of 32 bit values. Extract low 512 bits of result. See illustration.
shuffle_T32_4x8_hi	Transpose 4x8 matrix of 32 bit values. Extract high 512 bits of result. See illustration.
shuffle_T32_4x4	Transpose 4x4 matrix of 32 bit values. Extract resulting 512 bit vector. See illustration.
shuffle_T8_8x8	Transpose 4x4 matrix of 8 bit values. Extract resulting 512 bit vector. See illustration.
shuffle_T8_16x4	Transpose 16x4 matrix of 8 bit values. Extract resulting 512 bit vector. See illustration.
shuffle_T8_4x16	Transpose 4x16 matrix of 8 bit values. Extract resulting 512 bit vector. See illustration.
shuffle_T16_1x2_flip	Flip inner dimension of 16x2 matrix of 16 bit values. Extract resulting 512 bit vector. See illustration.
shuffle_T16_4x4	Transpose 4x4 matrix of 16 bit values. Extract resulting 512 bit vector. See illustration.
shuffle_T16_4x2	Transpose 4x2 matrix of 16 bit values. Extract resulting 512 bit vector. See illustration.
shuffle_T16_2x4	Transpose 2x4 matrix of 16 bit values. Extract resulting 512 bit vector. See illustration.
shuffle_T16_8x2	Transpose 8x2 matrix of 16 bit values. Extract resulting 512 bit vector. See illustration.
shuffle_T16_2x8	Transpose 2x8 matrix of 16 bit values. Extract resulting 512 bit vector. See illustration.
shuffle_T16_16x2	Transpose 16x2 matrix of 16 bit values. Extract resulting 512 bit vector. See illustration.
shuffle_T16_2x16	Transpose 2x16 matrix of 16 bit values. Extract resulting 512 bit vector. See illustration.
shuffle_T8_8x4	Transpose 8x4 matrix of 8 bit values. Extract resulting 512 bit vector. See illustration.
shuffle_T8_4x8	Transpose 4x8 matrix of 8 bit values. Extract resulting 512 bit vector. See illustration.

Function Documentation

shuffle() [1/35]

v128int4 shuffle	(	v128int4	a,
		unsigned int	mode
	)

shuffle() [2/35]

v128int4 shuffle	(	v128int4	,
		v128int4	,
		unsigned int
	)

shuffle() [3/35]

v128int8_sparse shuffle	(	v128int8_sparse	qx,
		int	itlv
	)

shuffle() [4/35]

v128uint4 shuffle	(	v128uint4	a,
		unsigned int	mode
	)

shuffle() [5/35]

v128uint4 shuffle	(	v128uint4	,
		v128uint4	,
		unsigned int
	)

shuffle() [6/35]

v128uint8_sparse shuffle	(	v128uint8_sparse	qx,
		int	itlv
	)

shuffle() [7/35]

v16cbfloat16 shuffle	(	v16cbfloat16	a,
		unsigned int	mode
	)

shuffle() [8/35]

v16cbfloat16 shuffle	(	v16cbfloat16	,
		v16cbfloat16	,
		unsigned int
	)

shuffle() [9/35]

v16cint16 shuffle	(	v16cint16	a,
		unsigned int	mode
	)

shuffle() [10/35]

v16cint16 shuffle	(	v16cint16	,
		v16cint16	,
		unsigned int
	)

shuffle() [11/35]

v16float shuffle	(	v16float	a,
		unsigned int	mode
	)

shuffle() [12/35]

v16float shuffle	(	v16float	,
		v16float	,
		unsigned int
	)

shuffle() [13/35]

v16int32 shuffle	(	v16int32	a,
		unsigned int	mode
	)

shuffle() [14/35]

v16int32 shuffle	(	v16int32	,
		v16int32	,
		unsigned int
	)

shuffle() [15/35]

v16uint32 shuffle	(	v16uint32	a,
		unsigned int	mode
	)

shuffle() [16/35]

v16uint32 shuffle	(	v16uint32	,
		v16uint32	,
		unsigned int
	)

shuffle() [17/35]

v256int4_sparse shuffle	(	v256int4_sparse	qx,
		int	itlv
	)

shuffle() [18/35]

v256uint4_sparse shuffle	(	v256uint4_sparse	qx,
		int	itlv
	)

shuffle() [19/35]

v32bfloat16 shuffle	(	v32bfloat16	a,
		unsigned int	mode
	)

shuffle() [20/35]

v32bfloat16 shuffle	(	v32bfloat16	,
		v32bfloat16	,
		unsigned int
	)

shuffle() [21/35]

v32int16 shuffle	(	v32int16	a,
		unsigned int	mode
	)

shuffle() [22/35]

v32int16 shuffle	(	v32int16	,
		v32int16	,
		unsigned int
	)

shuffle() [23/35]

v32uint16 shuffle	(	v32uint16	a,
		unsigned int	mode
	)

shuffle() [24/35]

v32uint16 shuffle	(	v32uint16	,
		v32uint16	,
		unsigned int
	)

shuffle() [25/35]

v64bfloat16_sparse shuffle	(	v64bfloat16_sparse	qx,
		int	itlv
	)

shuffle() [26/35]

v64int16_sparse shuffle	(	v64int16_sparse	qx,
		int	itlv
	)

shuffle() [27/35]

v64int8 shuffle	(	v64int8	a,
		unsigned int	mode
	)

shuffle() [28/35]

v64int8 shuffle	(	v64int8	,
		v64int8	,
		unsigned int
	)

shuffle() [29/35]

v64uint16_sparse shuffle	(	v64uint16_sparse	qx,
		int	itlv
	)

shuffle() [30/35]

v64uint8 shuffle	(	v64uint8	a,
		unsigned int	mode
	)

shuffle() [31/35]

v64uint8 shuffle	(	v64uint8	,
		v64uint8	,
		unsigned int
	)

shuffle() [32/35]

v8cfloat shuffle	(	v8cfloat	a,
		unsigned int	mode
	)

shuffle() [33/35]

v8cfloat shuffle	(	v8cfloat	,
		v8cfloat	,
		unsigned int
	)

shuffle() [34/35]

v8cint32 shuffle	(	v8cint32	a,
		unsigned int	mode
	)

shuffle() [35/35]

v8cint32 shuffle	(	v8cint32	,
		v8cint32	,
		unsigned int
	)

shuffle_bfloat16()

v32bfloat16 shuffle_bfloat16	(	bfloat16	b,
		unsigned int	m
	)

shuffle_c16()

v16cint16 shuffle_c16	(	cint16	b,
		unsigned int	m
	)

shuffle_c32()

v8cint32 shuffle_c32	(	cint32	b,
		unsigned int	m
	)

shuffle_float()

v16float shuffle_float	(	float	b,
		unsigned int	m
	)

shuffle_s16()

v32int16 shuffle_s16	(	int	b,
		unsigned int	m
	)

shuffle_s32()

v16int32 shuffle_s32	(	int	b,
		unsigned int	m
	)

shuffle_s64()

v16int32 shuffle_s64	(	long long	b,
		unsigned int	m
	)

shuffle_s8()

v64int8 shuffle_s8	(	int	b,
		unsigned int	m
	)

shuffle_u16()

v32uint16 shuffle_u16	(	unsigned int	b,
		unsigned int	m
	)

shuffle_u32()

v16uint32 shuffle_u32	(	unsigned int	b,
		unsigned int	m
	)

shuffle_u64()

v16uint32 shuffle_u64	(	unsigned long long	b,
		unsigned int	m
	)

shuffle_u8()

v64uint8 shuffle_u8	(	unsigned int	b,
		unsigned int	m
	)

shuffle_v2s32()

v16int32 shuffle_v2s32	(	v2int32	b,
		unsigned int	m
	)

shuffle_v2u32()

v16uint32 shuffle_v2u32	(	v2uint32	b,
		unsigned int	m
	)