Insert vector

Vector insert intrinsic functions allow substitution of the lanes within a vector value. More...

Topics
	Insert MX vector components
	Insert a vector into a vector
	Insert sparse vector components

Insert 256-bit in 512-bit register
v128int4	insert (v128int4 a, int idx, v64int4 b)
v128uint4	insert (v128uint4 a, int idx, v64uint4 b)
v64int8	insert (v64int8 a, int idx, v32int8 b)
v64uint8	insert (v64uint8 a, int idx, v32uint8 b)
v16cint16	insert (v16cint16 a, int idx, v8cint16 b)
v32int16	insert (v32int16 a, int idx, v16int16 b)
v32uint16	insert (v32uint16 a, int idx, v16uint16 b)
v8cint32	insert (v8cint32 a, int idx, v4cint32 b)
v16int32	insert (v16int32 a, int idx, v8int32 b)
v16uint32	insert (v16uint32 a, int idx, v8uint32 b)
v32bfloat16	insert (v32bfloat16 a, int idx, v16bfloat16 b)
v32float16	insert (v32float16 a, int idx, v16float16 b)
v64bfloat8	insert (v64bfloat8 a, int idx, v32bfloat8 b)
v64float8	insert (v64float8 a, int idx, v32float8 b)
v16float	insert (v16float a, int idx, v8float b)
v16cbfloat16	insert (v16cbfloat16 a, int idx, v8cbfloat16 b)
v8cfloat	insert (v8cfloat a, int idx, v4cfloat b)
v16accfloat	insert (v16accfloat a, int idx, v8accfloat b)
v8caccfloat	insert (v8caccfloat a, int idx, v4caccfloat b)
v16acc32	insert (v16acc32 a, int idx, v8acc32 b)
v8acc64	insert (v8acc64 a, int idx, v4acc64 b)
v4cacc64	insert (v4cacc64 a, int idx, v2cacc64 b)

Insert 256-bit in 1024-bit register
v256int4	insert (v256int4 a, int idx, v64int4 b)
v256uint4	insert (v256uint4 a, int idx, v64uint4 b)
v128int8	insert (v128int8 a, int idx, v32int8 b)
v128uint8	insert (v128uint8 a, int idx, v32uint8 b)
v32cint16	insert (v32cint16 a, int idx, v8cint16 b)
v64int16	insert (v64int16 a, int idx, v16int16 b)
v64uint16	insert (v64uint16 a, int idx, v16uint16 b)
v16cint32	insert (v16cint32 a, int idx, v4cint32 b)
v32int32	insert (v32int32 a, int idx, v8int32 b)
v32uint32	insert (v32uint32 a, int idx, v8uint32 b)
v64bfloat16	insert (v64bfloat16 a, int idx, v16bfloat16 b)
v64float16	insert (v64float16 a, int idx, v16float16 b)
v128bfloat8	insert (v128bfloat8 a, int idx, v32bfloat8 b)
v128float8	insert (v128float8 a, int idx, v32float8 b)
v32float	insert (v32float a, int idx, v8float b)
v32cbfloat16	insert (v32cbfloat16 a, int idx, v8cbfloat16 b)
v16cfloat	insert (v16cfloat a, int idx, v4cfloat b)
v32accfloat	insert (v32accfloat a, int idx, v8accfloat b)
v16caccfloat	insert (v16caccfloat a, int idx, v4caccfloat b)
v32acc32	insert (v32acc32 a, int idx, v8acc32 b)
v16acc64	insert (v16acc64 a, int idx, v4acc64 b)
v8cacc64	insert (v8cacc64 a, int idx, v2cacc64 b)

Insert 512-bit in 1024-bit register
v256int4	insert (v256int4 a, int idx, v128int4 b)
v256uint4	insert (v256uint4 a, int idx, v128uint4 b)
v128int8	insert (v128int8 a, int idx, v64int8 b)
v128uint8	insert (v128uint8 a, int idx, v64uint8 b)
v32cint16	insert (v32cint16 a, int idx, v16cint16 b)
v64int16	insert (v64int16 a, int idx, v32int16 b)
v64uint16	insert (v64uint16 a, int idx, v32uint16 b)
v16cint32	insert (v16cint32 a, int idx, v8cint32 b)
v32int32	insert (v32int32 a, int idx, v16int32 b)
v32uint32	insert (v32uint32 a, int idx, v16uint32 b)
v32accfloat	insert (v32accfloat a, int idx, v16accfloat b)
v16caccfloat	insert (v16caccfloat a, int idx, v8caccfloat b)
v32acc32	insert (v32acc32 a, int idx, v16acc32 b)
v16acc64	insert (v16acc64 a, int idx, v8acc64 b)
v8cacc64	insert (v8cacc64 a, int idx, v4cacc64 b)
v64bfloat16	insert (v64bfloat16 a, int idx, v32bfloat16 b)
v64float16	insert (v64float16 a, int idx, v32float16 b)
v128bfloat8	insert (v128bfloat8 a, int idx, v64bfloat8 b)
v128float8	insert (v128float8 a, int idx, v64float8 b)
v32float	insert (v32float a, int idx, v16float b)
v32cbfloat16	insert (v32cbfloat16 a, int idx, v16cbfloat16 b)
v16cfloat	insert (v16cfloat a, int idx, v8cfloat b)

Insert 512-bit in 2048-bit register
v64accfloat	insert (v64accfloat a, int idx, v16accfloat b)
v32caccfloat	insert (v32caccfloat a, int idx, v8caccfloat b)
v64acc32	insert (v64acc32 a, int idx, v16acc32 b)
v32acc64	insert (v32acc64 a, int idx, v8acc64 b)
v16cacc64	insert (v16cacc64 a, int idx, v4cacc64 b)

Insert 1024-bit in 2048-bit register
v64accfloat	insert (v64accfloat a, int idx, v32accfloat b)
v32caccfloat	insert (v32caccfloat a, int idx, v16caccfloat b)
v64acc32	insert (v64acc32 a, int idx, v32acc32 b)
v32acc64	insert (v32acc64 a, int idx, v16acc64 b)
v16cacc64	insert (v16cacc64 a, int idx, v8cacc64 b)

Insert v128mx9 into v256mx9
v256mx9	insert (v256mx9 m, int idx, v128mx9 a)

Insert v64mx9 into v256mx9
v256mx9	insert (v256mx9 m, int idx, v64mx9 a)

Detailed Description

Vector insert intrinsic functions allow substitution of the lanes within a vector value.

Below the buffer sizes are as follows: W - 256 bit X - 512 bit Y - 1024 bit

For more information see Integer Vector Types.

Note

All intrinsics require a compile time constant for the idx parameter except those in either of the following two forms:

• upd_w(Y buf,int idx,W val)

upd_hi and upd_lo intrinsic functions

Update the top half or bottom half of the lanes within a data type.

upd_w({X,Y} buf,int idx,W val) 256-bit intrinsic functions

upd_w(buf,0...3,val) update the successive 256-bit lanes into a 512/1024 bit vector.

The following example shows the update of a large 32-way complex vector 16 elements at-a-time using a 256-bit update. These updates are also pipelined.

const v16int16 * input  = d_in;
v64int16 sbuff = undef_v64int16();
...
sbuff = upd_w(sbuff,0, *input_++);   // 00++|08++|____|____    ____|____|____|____
sbuff = upd_w(sbuff,1, *input_++);   // 00..|08..|16++|24++    ____|____|____|____

concat, upd

In order to concatenate two vectors into a new vector twice the size you can either use the concat intrinsic or the correspondent upd intrinsics.

Both options are valid and should generally give the same performance.

Function Documentation

insert() [1/78]

v128bfloat8 insert	(	v128bfloat8	a,
		int	idx,
		v32bfloat8	b )

insert() [2/78]

v128bfloat8 insert	(	v128bfloat8	a,
		int	idx,
		v64bfloat8	b )

insert() [3/78]

v128float8 insert	(	v128float8	a,
		int	idx,
		v32float8	b )

insert() [4/78]

v128float8 insert	(	v128float8	a,
		int	idx,
		v64float8	b )

insert() [5/78]

v128int4 insert	(	v128int4	a,
		int	idx,
		v64int4	b )

insert() [6/78]

v128int8 insert	(	v128int8	a,
		int	idx,
		v32int8	b )

insert() [7/78]

v128int8 insert	(	v128int8	a,
		int	idx,
		v64int8	b )

insert() [8/78]

v128uint4 insert	(	v128uint4	a,
		int	idx,
		v64uint4	b )

insert() [9/78]

v128uint8 insert	(	v128uint8	a,
		int	idx,
		v32uint8	b )

insert() [10/78]

v128uint8 insert	(	v128uint8	a,
		int	idx,
		v64uint8	b )

insert() [11/78]

v16acc32 insert	(	v16acc32	a,
		int	idx,
		v8acc32	b )

insert() [12/78]

v16acc64 insert	(	v16acc64	a,
		int	idx,
		v4acc64	b )

insert() [13/78]

v16acc64 insert	(	v16acc64	a,
		int	idx,
		v8acc64	b )

insert() [14/78]

v16accfloat insert	(	v16accfloat	a,
		int	idx,
		v8accfloat	b )

insert() [15/78]

v16cacc64 insert	(	v16cacc64	a,
		int	idx,
		v4cacc64	b )

insert() [16/78]

v16cacc64 insert	(	v16cacc64	a,
		int	idx,
		v8cacc64	b )

insert() [17/78]

v16caccfloat insert	(	v16caccfloat	a,
		int	idx,
		v4caccfloat	b )

insert() [18/78]

v16caccfloat insert	(	v16caccfloat	a,
		int	idx,
		v8caccfloat	b )

insert() [19/78]

v16cbfloat16 insert	(	v16cbfloat16	a,
		int	idx,
		v8cbfloat16	b )

insert() [20/78]

v16cfloat insert	(	v16cfloat	a,
		int	idx,
		v4cfloat	b )

insert() [21/78]

v16cfloat insert	(	v16cfloat	a,
		int	idx,
		v8cfloat	b )

insert() [22/78]

v16cint16 insert	(	v16cint16	a,
		int	idx,
		v8cint16	b )

insert() [23/78]

v16cint32 insert	(	v16cint32	a,
		int	idx,
		v4cint32	b )

insert() [24/78]

v16cint32 insert	(	v16cint32	a,
		int	idx,
		v8cint32	b )

insert() [25/78]

v16float insert	(	v16float	a,
		int	idx,
		v8float	b )

insert() [26/78]

v16int32 insert	(	v16int32	a,
		int	idx,
		v8int32	b )

insert() [27/78]

v16uint32 insert	(	v16uint32	a,
		int	idx,
		v8uint32	b )

insert() [28/78]

v256int4 insert	(	v256int4	a,
		int	idx,
		v128int4	b )

insert() [29/78]

v256int4 insert	(	v256int4	a,
		int	idx,
		v64int4	b )

insert() [30/78]

v256mx9 insert	(	v256mx9	m,
		int	idx,
		v128mx9	a )

insert() [31/78]

v256mx9 insert	(	v256mx9	m,
		int	idx,
		v64mx9	a )

insert() [32/78]

v256uint4 insert	(	v256uint4	a,
		int	idx,
		v128uint4	b )

insert() [33/78]

v256uint4 insert	(	v256uint4	a,
		int	idx,
		v64uint4	b )

insert() [34/78]

v32acc32 insert	(	v32acc32	a,
		int	idx,
		v16acc32	b )

insert() [35/78]

v32acc32 insert	(	v32acc32	a,
		int	idx,
		v8acc32	b )

insert() [36/78]

v32acc64 insert	(	v32acc64	a,
		int	idx,
		v16acc64	b )

insert() [37/78]

v32acc64 insert	(	v32acc64	a,
		int	idx,
		v8acc64	b )

insert() [38/78]

v32accfloat insert	(	v32accfloat	a,
		int	idx,
		v16accfloat	b )

insert() [39/78]

v32accfloat insert	(	v32accfloat	a,
		int	idx,
		v8accfloat	b )

insert() [40/78]

v32bfloat16 insert	(	v32bfloat16	a,
		int	idx,
		v16bfloat16	b )

insert() [41/78]

v32caccfloat insert	(	v32caccfloat	a,
		int	idx,
		v16caccfloat	b )

insert() [42/78]

v32caccfloat insert	(	v32caccfloat	a,
		int	idx,
		v8caccfloat	b )

insert() [43/78]

v32cbfloat16 insert	(	v32cbfloat16	a,
		int	idx,
		v16cbfloat16	b )

insert() [44/78]

v32cbfloat16 insert	(	v32cbfloat16	a,
		int	idx,
		v8cbfloat16	b )

insert() [45/78]

v32cint16 insert	(	v32cint16	a,
		int	idx,
		v16cint16	b )

insert() [46/78]

v32cint16 insert	(	v32cint16	a,
		int	idx,
		v8cint16	b )

insert() [47/78]

v32float insert	(	v32float	a,
		int	idx,
		v16float	b )

insert() [48/78]

v32float insert	(	v32float	a,
		int	idx,
		v8float	b )

insert() [49/78]

v32float16 insert	(	v32float16	a,
		int	idx,
		v16float16	b )

insert() [50/78]

v32int16 insert	(	v32int16	a,
		int	idx,
		v16int16	b )

insert() [51/78]

v32int32 insert	(	v32int32	a,
		int	idx,
		v16int32	b )

insert() [52/78]

v32int32 insert	(	v32int32	a,
		int	idx,
		v8int32	b )

insert() [53/78]

v32uint16 insert	(	v32uint16	a,
		int	idx,
		v16uint16	b )

insert() [54/78]

v32uint32 insert	(	v32uint32	a,
		int	idx,
		v16uint32	b )

insert() [55/78]

v32uint32 insert	(	v32uint32	a,
		int	idx,
		v8uint32	b )

insert() [56/78]

v4cacc64 insert	(	v4cacc64	a,
		int	idx,
		v2cacc64	b )

insert() [57/78]

v64acc32 insert	(	v64acc32	a,
		int	idx,
		v16acc32	b )

insert() [58/78]

v64acc32 insert	(	v64acc32	a,
		int	idx,
		v32acc32	b )

insert() [59/78]

v64accfloat insert	(	v64accfloat	a,
		int	idx,
		v16accfloat	b )

insert() [60/78]

v64accfloat insert	(	v64accfloat	a,
		int	idx,
		v32accfloat	b )

insert() [61/78]

v64bfloat16 insert	(	v64bfloat16	a,
		int	idx,
		v16bfloat16	b )

insert() [62/78]

v64bfloat16 insert	(	v64bfloat16	a,
		int	idx,
		v32bfloat16	b )

insert() [63/78]

v64bfloat8 insert	(	v64bfloat8	a,
		int	idx,
		v32bfloat8	b )

insert() [64/78]

v64float16 insert	(	v64float16	a,
		int	idx,
		v16float16	b )

insert() [65/78]

v64float16 insert	(	v64float16	a,
		int	idx,
		v32float16	b )

insert() [66/78]

v64float8 insert	(	v64float8	a,
		int	idx,
		v32float8	b )

insert() [67/78]

v64int16 insert	(	v64int16	a,
		int	idx,
		v16int16	b )

insert() [68/78]

v64int16 insert	(	v64int16	a,
		int	idx,
		v32int16	b )

insert() [69/78]

v64int8 insert	(	v64int8	a,
		int	idx,
		v32int8	b )

insert() [70/78]

v64uint16 insert	(	v64uint16	a,
		int	idx,
		v16uint16	b )

insert() [71/78]

v64uint16 insert	(	v64uint16	a,
		int	idx,
		v32uint16	b )

insert() [72/78]

v64uint8 insert	(	v64uint8	a,
		int	idx,
		v32uint8	b )

insert() [73/78]

v8acc64 insert	(	v8acc64	a,
		int	idx,
		v4acc64	b )

insert() [74/78]

v8cacc64 insert	(	v8cacc64	a,
		int	idx,
		v2cacc64	b )

insert() [75/78]

v8cacc64 insert	(	v8cacc64	a,
		int	idx,
		v4cacc64	b )

insert() [76/78]

v8caccfloat insert	(	v8caccfloat	a,
		int	idx,
		v4caccfloat	b )

insert() [77/78]

v8cfloat insert	(	v8cfloat	a,
		int	idx,
		v4cfloat	b )

insert() [78/78]

v8cint32 insert	(	v8cint32	a,
		int	idx,
		v4cint32	b )