Shuffle

Intrinsics allowing you perform vector shuffles. More...

Topics
	Illustration of Shuffle Modes

Shuffle two vectors
Shuffle two vectors of the same datatype according to a given mode. Parameters a First vector to Shuffle b Second vector to Shuffle mode shuffle mode (of type eInterleaveMode) Returns Shuffle vector
v16acc32	shuffle (v16acc32 a, v16acc32 b, unsigned int mode)
v16int32	shuffle (v16int32 a, v16int32 b, unsigned int mode)
v32int16	shuffle (v32int16 a, v32int16 b, unsigned int mode)
v64int8	shuffle (v64int8 a, v64int8 b, unsigned int mode)
v128int4	shuffle (v128int4 a, v128int4 b, unsigned int mode)
v16uint32	shuffle (v16uint32 a, v16uint32 b, unsigned int mode)
v32uint16	shuffle (v32uint16 a, v32uint16 b, unsigned int mode)
v64uint8	shuffle (v64uint8 a, v64uint8 b, unsigned int mode)
v128uint4	shuffle (v128uint4 a, v128uint4 b, unsigned int mode)
v16cbfloat16	shuffle (v16cbfloat16 a, v16cbfloat16 b, unsigned int mode)
v8cint32	shuffle (v8cint32 a, v8cint32 b, unsigned int mode)
v8cfloat	shuffle (v8cfloat a, v8cfloat b, unsigned int mode)
v16cint16	shuffle (v16cint16 a, v16cint16 b, unsigned int mode)
v32bfloat16	shuffle (v32bfloat16 a, v32bfloat16 b, unsigned int mode)
v32float16	shuffle (v32float16 a, v32float16 b, unsigned int mode)
v64float8	shuffle (v64float8 a, v64float8 b, unsigned int mode)
v64bfloat8	shuffle (v64bfloat8 a, v64bfloat8 b, unsigned int mode)
v16float	shuffle (v16float a, v16float b, unsigned int mode)

Interleave one vector
Shuffle one vector given a shuffle mode Parameters a Vector to shuffle mode shuffle mode (of type eInterleaveMode) Returns shuffle 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)
v16cbfloat16	shuffle (v16cbfloat16 a, unsigned int mode)
v8cint32	shuffle (v8cint32 a, unsigned int mode)
v8cfloat	shuffle (v8cfloat a, unsigned int mode)
v16cint16	shuffle (v16cint16 a, unsigned int mode)
v32bfloat16	shuffle (v32bfloat16 a, unsigned int mode)
v32float16	shuffle (v32float16 a, unsigned int mode)
v64float8	shuffle (v64float8 a, unsigned int mode)
v64bfloat8	shuffle (v64bfloat8 a, unsigned int mode)
v16float	shuffle (v16float 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 mode (of type eInterleaveMode)
v64int8	shuffle_s8 (int b, unsigned int mode)
v32int16	shuffle_s16 (int b, unsigned int mode)
v16int32	shuffle_s32 (int b, unsigned int mode)
v16int32	shuffle_s64 (mask64 b, unsigned int mode)
v16int32	shuffle_v2s32 (v2int32 b, unsigned int mode)
v64uint8	shuffle_u8 (unsigned int b, unsigned int mode)
v32uint16	shuffle_u16 (unsigned int b, unsigned int mode)
v16uint32	shuffle_u32 (unsigned int b, unsigned int mode)
v16uint32	shuffle_u64 (mask64 b, unsigned int mode)
v16uint32	shuffle_v2u32 (v2uint32 b, unsigned int mode)
v16cint16	shuffle_c16 (cint16 b, unsigned int mode)
v16cbfloat16	shuffle_cbf16 (v1cbfloat16 b, unsigned int mode)
v16cbfloat16	shuffle_cbf16 (cbfloat16 b, unsigned int mode)
v8cint32	shuffle_c32 (cint32_w64 b, unsigned int mode)
v8cint32	shuffle_c32 (cint32 b, unsigned int mode)
v8cfloat	shuffle_c32 (cfloat_w64 b, unsigned int mode)
v8cfloat	shuffle_c32 (cfloat b, unsigned int mode)
v32bfloat16	shuffle_bfloat16 (bfloat16 b, unsigned int mode)
v32float16	shuffle_float16 (float16 b, unsigned int mode)
v64float8	shuffle_float8 (float8 b, unsigned int mode)
v64bfloat8	shuffle_bfloat8 (bfloat8 b, unsigned int mode)
v16float	shuffle_float (float b, unsigned int mode)

Interleave a sparse vector
Interleave a sparse vector Parameters a Vector to interleave mode Interleaving mode (of type eInterleaveMode) Returns interleaved vector
v512int4_sparse	shuffle (v512int4_sparse qy, int itlv)
v256int8_sparse	shuffle (v256int8_sparse qy, int itlv)
v128int16_sparse	shuffle (v128int16_sparse qy, int itlv)
v256int4_sparse	shuffle (v256int4_sparse qx, int itlv)
v128int8_sparse	shuffle (v128int8_sparse qx, int itlv)
v64int16_sparse	shuffle (v64int16_sparse qx, int itlv)
v512uint4_sparse	shuffle (v512uint4_sparse qy, int itlv)
v256uint8_sparse	shuffle (v256uint8_sparse qy, int itlv)
v128uint16_sparse	shuffle (v128uint16_sparse qy, 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)

Interleave mx (block floating-point) vectors
Interleave one vector Parameters a Vector to interleave mode Interleaving mode (first byte controls mantissa interleaving, second byte controls exponent interleaving) Returns interleaved vector
v64mx9	shuffle (v64mx9 a, v64mx9 b, unsigned mode)
v128mx6	shuffle (v128mx6 a, v128mx6 b, unsigned mode)
v128mx4	shuffle (v128mx4 a, v128mx4 b, unsigned mode)
v64mx9	shuffle (v64mx9 a, unsigned mode)
v128mx6	shuffle (v128mx6 a, unsigned mode)
v128mx4	shuffle (v128mx4 a, unsigned mode)

Detailed Description

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	-
64-bit	4x4 → 4x4	shuffle_T64_4x4_lo	shuffle_T64_4x4_hi	shuffle_T64_4x4_lo	shuffle_T64_4x4_hi
8-bit	8x8 → 8x8	shuffle_T8_8x8	-	shuffle_T8_8x8	-
8-bit	16x4 → 4x16	shuffle_T8_16x4	-	shuffle_T8_4x16	-
8-bit	16x[0,1,2,3] → 16x[3,2,1,0]	shuffle_T8_1x4_flip	-	shuffle_T8_1x4_flip	-
16-bit	16x[0,1] → 16x[1,0]	shuffle_T16_1x2_flip	-	shuffle_T16_1x2_flip	-
32-bit	1x[0..15] → 1x[15..0]	shuffle_T32_1x16_flip	-	shuffle_T32_1x16_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	-
32-bit	8x2 → 2x8	shuffle_T32_8x2	-	shuffle_T32_2x8	-
64-bit	4x2 → 2x4	shuffle_T64_4x2	-	shuffle_T8_2x4	-
16-bit	8x8 → 8x8	shuffle_T16_8x8	-	shuffle_T8_8x8	-

Function Documentation

shuffle() [1/53]

v128int16_sparse shuffle	(	v128int16_sparse	qy,
		int	itlv )

shuffle() [2/53]

v128int4 shuffle	(	v128int4	a,
		unsigned int	mode )

shuffle() [3/53]

v128int4 shuffle	(	v128int4	a,
		v128int4	b,
		unsigned int	mode )

shuffle() [4/53]

v128int8_sparse shuffle	(	v128int8_sparse	qx,
		int	itlv )

shuffle() [5/53]

v128mx4 shuffle	(	v128mx4	a,
		unsigned	mode )

shuffle() [6/53]

v128mx4 shuffle	(	v128mx4	a,
		v128mx4	b,
		unsigned	mode )

shuffle() [7/53]

v128mx6 shuffle	(	v128mx6	a,
		unsigned	mode )

shuffle() [8/53]

v128mx6 shuffle	(	v128mx6	a,
		v128mx6	b,
		unsigned	mode )

shuffle() [9/53]

v128uint16_sparse shuffle	(	v128uint16_sparse	qy,
		int	itlv )

shuffle() [10/53]

v128uint4 shuffle	(	v128uint4	a,
		unsigned int	mode )

shuffle() [11/53]

v128uint4 shuffle	(	v128uint4	a,
		v128uint4	b,
		unsigned int	mode )

shuffle() [12/53]

v128uint8_sparse shuffle	(	v128uint8_sparse	qx,
		int	itlv )

shuffle() [13/53]

v16acc32 shuffle	(	v16acc32	a,
		v16acc32	b,
		unsigned int	mode )

shuffle() [14/53]

v16cbfloat16 shuffle	(	v16cbfloat16	a,
		unsigned int	mode )

shuffle() [15/53]

v16cbfloat16 shuffle	(	v16cbfloat16	a,
		v16cbfloat16	b,
		unsigned int	mode )

shuffle() [16/53]

v16cint16 shuffle	(	v16cint16	a,
		unsigned int	mode )

shuffle() [17/53]

v16cint16 shuffle	(	v16cint16	a,
		v16cint16	b,
		unsigned int	mode )

shuffle() [18/53]

v16float shuffle	(	v16float	a,
		unsigned int	mode )

shuffle() [19/53]

v16float shuffle	(	v16float	a,
		v16float	b,
		unsigned int	mode )

shuffle() [20/53]

v16int32 shuffle	(	v16int32	a,
		unsigned int	mode )

shuffle() [21/53]

v16int32 shuffle	(	v16int32	a,
		v16int32	b,
		unsigned int	mode )

shuffle() [22/53]

v16uint32 shuffle	(	v16uint32	a,
		unsigned int	mode )

shuffle() [23/53]

v16uint32 shuffle	(	v16uint32	a,
		v16uint32	b,
		unsigned int	mode )

shuffle() [24/53]

v256int4_sparse shuffle	(	v256int4_sparse	qx,
		int	itlv )

shuffle() [25/53]

v256int8_sparse shuffle	(	v256int8_sparse	qy,
		int	itlv )

shuffle() [26/53]

v256uint4_sparse shuffle	(	v256uint4_sparse	qx,
		int	itlv )

shuffle() [27/53]

v256uint8_sparse shuffle	(	v256uint8_sparse	qy,
		int	itlv )

shuffle() [28/53]

v32bfloat16 shuffle	(	v32bfloat16	a,
		unsigned int	mode )

shuffle() [29/53]

v32bfloat16 shuffle	(	v32bfloat16	a,
		v32bfloat16	b,
		unsigned int	mode )

shuffle() [30/53]

v32float16 shuffle	(	v32float16	a,
		unsigned int	mode )

shuffle() [31/53]

v32float16 shuffle	(	v32float16	a,
		v32float16	b,
		unsigned int	mode )

shuffle() [32/53]

v32int16 shuffle	(	v32int16	a,
		unsigned int	mode )

shuffle() [33/53]

v32int16 shuffle	(	v32int16	a,
		v32int16	b,
		unsigned int	mode )

shuffle() [34/53]

v32uint16 shuffle	(	v32uint16	a,
		unsigned int	mode )

shuffle() [35/53]

v32uint16 shuffle	(	v32uint16	a,
		v32uint16	b,
		unsigned int	mode )

shuffle() [36/53]

v512int4_sparse shuffle	(	v512int4_sparse	qy,
		int	itlv )

shuffle() [37/53]

v512uint4_sparse shuffle	(	v512uint4_sparse	qy,
		int	itlv )

shuffle() [38/53]

v64bfloat8 shuffle	(	v64bfloat8	a,
		unsigned int	mode )

shuffle() [39/53]

v64bfloat8 shuffle	(	v64bfloat8	a,
		v64bfloat8	b,
		unsigned int	mode )

shuffle() [40/53]

v64float8 shuffle	(	v64float8	a,
		unsigned int	mode )

shuffle() [41/53]

v64float8 shuffle	(	v64float8	a,
		v64float8	b,
		unsigned int	mode )

shuffle() [42/53]

v64int16_sparse shuffle	(	v64int16_sparse	qx,
		int	itlv )

shuffle() [43/53]

v64int8 shuffle	(	v64int8	a,
		unsigned int	mode )

shuffle() [44/53]

v64int8 shuffle	(	v64int8	a,
		v64int8	b,
		unsigned int	mode )

shuffle() [45/53]

v64mx9 shuffle	(	v64mx9	a,
		unsigned	mode )

shuffle() [46/53]

v64mx9 shuffle	(	v64mx9	a,
		v64mx9	b,
		unsigned	mode )

shuffle() [47/53]

v64uint16_sparse shuffle	(	v64uint16_sparse	qx,
		int	itlv )

shuffle() [48/53]

v64uint8 shuffle	(	v64uint8	a,
		unsigned int	mode )

shuffle() [49/53]

v64uint8 shuffle	(	v64uint8	a,
		v64uint8	b,
		unsigned int	mode )

shuffle() [50/53]

v8cfloat shuffle	(	v8cfloat	a,
		unsigned int	mode )

shuffle() [51/53]

v8cfloat shuffle	(	v8cfloat	a,
		v8cfloat	b,
		unsigned int	mode )

shuffle() [52/53]

v8cint32 shuffle	(	v8cint32	a,
		unsigned int	mode )

shuffle() [53/53]

v8cint32 shuffle	(	v8cint32	a,
		v8cint32	b,
		unsigned int	mode )

shuffle_bfloat16()

v32bfloat16 shuffle_bfloat16	(	bfloat16	b,
		unsigned int	mode )

shuffle_bfloat8()

v64bfloat8 shuffle_bfloat8	(	bfloat8	b,
		unsigned int	mode )

shuffle_c16()

v16cint16 shuffle_c16	(	cint16	b,
		unsigned int	mode )

shuffle_c32() [1/4]

v8cfloat shuffle_c32	(	cfloat	b,
		unsigned int	mode )

shuffle_c32() [2/4]

v8cfloat shuffle_c32	(	cfloat_w64	b,
		unsigned int	mode )

shuffle_c32() [3/4]

v8cint32 shuffle_c32	(	cint32	b,
		unsigned int	mode )

shuffle_c32() [4/4]

v8cint32 shuffle_c32	(	cint32_w64	b,
		unsigned int	mode )

shuffle_cbf16() [1/2]

v16cbfloat16 shuffle_cbf16	(	cbfloat16	b,
		unsigned int	mode )

shuffle_cbf16() [2/2]

v16cbfloat16 shuffle_cbf16	(	v1cbfloat16	b,
		unsigned int	mode )

shuffle_float()

v16float shuffle_float	(	float	b,
		unsigned int	mode )

shuffle_float16()

v32float16 shuffle_float16	(	float16	b,
		unsigned int	mode )

shuffle_float8()

v64float8 shuffle_float8	(	float8	b,
		unsigned int	mode )

shuffle_s16()

v32int16 shuffle_s16	(	int	b,
		unsigned int	mode )

shuffle_s32()

v16int32 shuffle_s32	(	int	b,
		unsigned int	mode )

shuffle_s64()

v16int32 shuffle_s64	(	mask64	b,
		unsigned int	mode )

shuffle_s8()

v64int8 shuffle_s8	(	int	b,
		unsigned int	mode )

shuffle_u16()

v32uint16 shuffle_u16	(	unsigned int	b,
		unsigned int	mode )

shuffle_u32()

v16uint32 shuffle_u32	(	unsigned int	b,
		unsigned int	mode )

shuffle_u64()

v16uint32 shuffle_u64	(	mask64	b,
		unsigned int	mode )

shuffle_u8()

v64uint8 shuffle_u8	(	unsigned int	b,
		unsigned int	mode )

shuffle_v2s32()

v16int32 shuffle_v2s32	(	v2int32	b,
		unsigned int	mode )

shuffle_v2u32()

v16uint32 shuffle_v2u32	(	v2uint32	b,
		unsigned int	mode )