Concatenate vectors

Vector concat intrinsic functions allow concatenation of vector values to create a larger one. More...

Topics
	Concatenate four vectors
	Concatenate two vectors

Functions
v4cacc64	concat (v2cacc64 a, v2cacc64 b)
v8cacc64	concat (v2cacc64 a, v2cacc64 b, v2cacc64 c, v2cacc64 d)
v8acc64	concat (v4acc64 a, v4acc64 b)
v16acc64	concat (v4acc64 a, v4acc64 b, v4acc64 c, v4acc64 d)
v8caccfloat	concat (v4caccfloat a, v4caccfloat b)
v16caccfloat	concat (v4caccfloat a, v4caccfloat b, v4caccfloat c, v4caccfloat d)
v16acc32	concat (v8acc32 a, v8acc32 b)
v32acc32	concat (v8acc32 a, v8acc32 b, v8acc32 c, v8acc32 d)
v16accfloat	concat (v8accfloat a, v8accfloat b)
v32accfloat	concat (v8accfloat a, v8accfloat b, v8accfloat c, v8accfloat d)

Concat v64int4 vectors
v128int4	concat (v64int4 a0, v64int4 a1)
v256int4	concat (v64int4 a0, v64int4 a1, v64int4 a2, v64int4 a3)

Concat v64uint4 vectors
v128uint4	concat (v64uint4 a0, v64uint4 a1)
v256uint4	concat (v64uint4 a0, v64uint4 a1, v64uint4 a2, v64uint4 a3)

Concat v32int8 vectors
v64int8	concat (v32int8 a0, v32int8 a1)
v128int8	concat (v32int8 a0, v32int8 a1, v32int8 a2, v32int8 a3)

Concat v32uint8 vectors
v64uint8	concat (v32uint8 a0, v32uint8 a1)
v128uint8	concat (v32uint8 a0, v32uint8 a1, v32uint8 a2, v32uint8 a3)

Concat v8cint16 vectors
v16cint16	concat (v8cint16 a0, v8cint16 a1)
v32cint16	concat (v8cint16 a0, v8cint16 a1, v8cint16 a2, v8cint16 a3)

Concat v16int16 vectors
v32int16	concat (v16int16 a0, v16int16 a1)
v64int16	concat (v16int16 a0, v16int16 a1, v16int16 a2, v16int16 a3)

Concat v16uint16 vectors
v32uint16	concat (v16uint16 a0, v16uint16 a1)
v64uint16	concat (v16uint16 a0, v16uint16 a1, v16uint16 a2, v16uint16 a3)

Concat v4cint32 vectors
v8cint32	concat (v4cint32 a0, v4cint32 a1)
v16cint32	concat (v4cint32 a0, v4cint32 a1, v4cint32 a2, v4cint32 a3)

Concat v8int32 vectors
v16int32	concat (v8int32 a0, v8int32 a1)
v32int32	concat (v8int32 a0, v8int32 a1, v8int32 a2, v8int32 a3)

Concat v8uint32 vectors
v16uint32	concat (v8uint32 a0, v8uint32 a1)
v32uint32	concat (v8uint32 a0, v8uint32 a1, v8uint32 a2, v8uint32 a3)

Concat v16bfloat16 vectors
v32bfloat16	concat (v16bfloat16 a0, v16bfloat16 a1)
v64bfloat16	concat (v16bfloat16 a0, v16bfloat16 a1, v16bfloat16 a2, v16bfloat16 a3)

Concat v16float16 vectors
v32float16	concat (v16float16 a0, v16float16 a1)
v64float16	concat (v16float16 a0, v16float16 a1, v16float16 a2, v16float16 a3)

Concat v32bfloat8 vectors
v64bfloat8	concat (v32bfloat8 a0, v32bfloat8 a1)
v128bfloat8	concat (v32bfloat8 a0, v32bfloat8 a1, v32bfloat8 a2, v32bfloat8 a3)

Concat v32float8 vectors
v64float8	concat (v32float8 a0, v32float8 a1)
v128float8	concat (v32float8 a0, v32float8 a1, v32float8 a2, v32float8 a3)

Concat v8float vectors
v16float	concat (v8float a0, v8float a1)
v32float	concat (v8float a0, v8float a1, v8float a2, v8float a3)

Concat v8cbfloat16 vectors
v16cbfloat16	concat (v8cbfloat16 a0, v8cbfloat16 a1)
v32cbfloat16	concat (v8cbfloat16 a0, v8cbfloat16 a1, v8cbfloat16 a2, v8cbfloat16 a3)

Concat v4cfloat vectors
v8cfloat	concat (v4cfloat a0, v4cfloat a1)
v16cfloat	concat (v4cfloat a0, v4cfloat a1, v4cfloat a2, v4cfloat a3)

Concat v128int4 vectors
v256int4	concat (v128int4 a0, v128int4 a1)

Concat v128uint4 vectors
v256uint4	concat (v128uint4 a0, v128uint4 a1)

Concat v64int8 vectors
v128int8	concat (v64int8 a0, v64int8 a1)

Concat v64uint8 vectors
v128uint8	concat (v64uint8 a0, v64uint8 a1)

Concat v16cint16 vectors
v32cint16	concat (v16cint16 a0, v16cint16 a1)

Concat v32int16 vectors
v64int16	concat (v32int16 a0, v32int16 a1)

Concat v32uint16 vectors
v64uint16	concat (v32uint16 a0, v32uint16 a1)

Concat v8cint32 vectors
v16cint32	concat (v8cint32 a0, v8cint32 a1)

Concat v16int32 vectors
v32int32	concat (v16int32 a0, v16int32 a1)

Concat v16uint32 vectors
v32uint32	concat (v16uint32 a0, v16uint32 a1)

Concat v16accfloat vectors
v32accfloat	concat (v16accfloat a0, v16accfloat a1)
v64accfloat	concat (v16accfloat a0, v16accfloat a1, v16accfloat a2, v16accfloat a3)

Concat v8caccfloat vectors
v16caccfloat	concat (v8caccfloat a0, v8caccfloat a1)
v32caccfloat	concat (v8caccfloat a0, v8caccfloat a1, v8caccfloat a2, v8caccfloat a3)

Concat v16acc32 vectors
v32acc32	concat (v16acc32 a0, v16acc32 a1)
v64acc32	concat (v16acc32 a0, v16acc32 a1, v16acc32 a2, v16acc32 a3)

Concat v8acc64 vectors
v16acc64	concat (v8acc64 a0, v8acc64 a1)
v32acc64	concat (v8acc64 a0, v8acc64 a1, v8acc64 a2, v8acc64 a3)

Concat v4cacc64 vectors
v8cacc64	concat (v4cacc64 a0, v4cacc64 a1)
v16cacc64	concat (v4cacc64 a0, v4cacc64 a1, v4cacc64 a2, v4cacc64 a3)

Concat v32bfloat16 vectors
v64bfloat16	concat (v32bfloat16 a0, v32bfloat16 a1)

Concat v32float16 vectors
v64float16	concat (v32float16 a0, v32float16 a1)

Concat v64bfloat8 vectors
v128bfloat8	concat (v64bfloat8 a0, v64bfloat8 a1)

Concat v64float8 vectors
v128float8	concat (v64float8 a0, v64float8 a1)

Concat v16float vectors
v32float	concat (v16float a0, v16float a1)

Concat v16cbfloat16 vectors
v32cbfloat16	concat (v16cbfloat16 a0, v16cbfloat16 a1)

Concat v8cfloat vectors
v16cfloat	concat (v8cfloat a0, v8cfloat a1)

Concat v32accfloat vectors
v64accfloat	concat (v32accfloat a0, v32accfloat a1)

Concat v16caccfloat vectors
v32caccfloat	concat (v16caccfloat a0, v16caccfloat a1)

Concat v32acc32 vectors
v64acc32	concat (v32acc32 a0, v32acc32 a1)

Concat v16acc64 vectors
v32acc64	concat (v16acc64 a0, v16acc64 a1)

Concat v8cacc64 vectors
v16cacc64	concat (v8cacc64 a0, v8cacc64 a1)

Detailed Description

Vector concat intrinsic functions allow concatenation of vector values to create a larger one.

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++    ____|____|____|____

wset, xset, yset

Same as update but set value of new buf and not one that exists already. For example:

wset_w(idx,val) is the same as upd_w(undef_type(),idx,val)

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

concat() [1/76]

v256int4 concat	(	v128int4	a0,
		v128int4	a1 )

concat() [2/76]

v256uint4 concat	(	v128uint4	a0,
		v128uint4	a1 )

concat() [3/76]

v32acc32 concat	(	v16acc32	a0,
		v16acc32	a1 )

concat() [4/76]

v64acc32 concat	(	v16acc32	a0,
		v16acc32	a1,
		v16acc32	a2,
		v16acc32	a3 )

concat() [5/76]

v32acc64 concat	(	v16acc64	a0,
		v16acc64	a1 )

concat() [6/76]

v32accfloat concat	(	v16accfloat	a0,
		v16accfloat	a1 )

concat() [7/76]

v64accfloat concat	(	v16accfloat	a0,
		v16accfloat	a1,
		v16accfloat	a2,
		v16accfloat	a3 )

concat() [8/76]

v32bfloat16 concat	(	v16bfloat16	a0,
		v16bfloat16	a1 )

concat() [9/76]

v64bfloat16 concat	(	v16bfloat16	a0,
		v16bfloat16	a1,
		v16bfloat16	a2,
		v16bfloat16	a3 )

concat() [10/76]

v32caccfloat concat	(	v16caccfloat	a0,
		v16caccfloat	a1 )

concat() [11/76]

v32cbfloat16 concat	(	v16cbfloat16	a0,
		v16cbfloat16	a1 )

concat() [12/76]

v32cint16 concat	(	v16cint16	a0,
		v16cint16	a1 )

concat() [13/76]

v32float concat	(	v16float	a0,
		v16float	a1 )

concat() [14/76]

v32float16 concat	(	v16float16	a0,
		v16float16	a1 )

concat() [15/76]

v64float16 concat	(	v16float16	a0,
		v16float16	a1,
		v16float16	a2,
		v16float16	a3 )

concat() [16/76]

v32int16 concat	(	v16int16	a0,
		v16int16	a1 )

concat() [17/76]

v64int16 concat	(	v16int16	a0,
		v16int16	a1,
		v16int16	a2,
		v16int16	a3 )

concat() [18/76]

v32int32 concat	(	v16int32	a0,
		v16int32	a1 )

concat() [19/76]

v32uint16 concat	(	v16uint16	a0,
		v16uint16	a1 )

concat() [20/76]

v64uint16 concat	(	v16uint16	a0,
		v16uint16	a1,
		v16uint16	a2,
		v16uint16	a3 )

concat() [21/76]

v32uint32 concat	(	v16uint32	a0,
		v16uint32	a1 )

concat() [22/76]

v4cacc64 concat	(	v2cacc64	a,
		v2cacc64	b )

concat() [23/76]

v8cacc64 concat	(	v2cacc64	a,
		v2cacc64	b,
		v2cacc64	c,
		v2cacc64	d )

concat() [24/76]

v64acc32 concat	(	v32acc32	a0,
		v32acc32	a1 )

concat() [25/76]

v64accfloat concat	(	v32accfloat	a0,
		v32accfloat	a1 )

concat() [26/76]

v64bfloat16 concat	(	v32bfloat16	a0,
		v32bfloat16	a1 )

concat() [27/76]

v64bfloat8 concat	(	v32bfloat8	a0,
		v32bfloat8	a1 )

concat() [28/76]

v128bfloat8 concat	(	v32bfloat8	a0,
		v32bfloat8	a1,
		v32bfloat8	a2,
		v32bfloat8	a3 )

concat() [29/76]

v64float16 concat	(	v32float16	a0,
		v32float16	a1 )

concat() [30/76]

v64float8 concat	(	v32float8	a0,
		v32float8	a1 )

concat() [31/76]

v128float8 concat	(	v32float8	a0,
		v32float8	a1,
		v32float8	a2,
		v32float8	a3 )

concat() [32/76]

v64int16 concat	(	v32int16	a0,
		v32int16	a1 )

concat() [33/76]

v64int8 concat	(	v32int8	a0,
		v32int8	a1 )

concat() [34/76]

v128int8 concat	(	v32int8	a0,
		v32int8	a1,
		v32int8	a2,
		v32int8	a3 )

concat() [35/76]

v64uint16 concat	(	v32uint16	a0,
		v32uint16	a1 )

concat() [36/76]

v64uint8 concat	(	v32uint8	a0,
		v32uint8	a1 )

concat() [37/76]

v128uint8 concat	(	v32uint8	a0,
		v32uint8	a1,
		v32uint8	a2,
		v32uint8	a3 )

concat() [38/76]

v8acc64 concat	(	v4acc64	a,
		v4acc64	b )

concat() [39/76]

v16acc64 concat	(	v4acc64	a,
		v4acc64	b,
		v4acc64	c,
		v4acc64	d )

concat() [40/76]

v8cacc64 concat	(	v4cacc64	a0,
		v4cacc64	a1 )

concat() [41/76]

v16cacc64 concat	(	v4cacc64	a0,
		v4cacc64	a1,
		v4cacc64	a2,
		v4cacc64	a3 )

concat() [42/76]

v8caccfloat concat	(	v4caccfloat	a,
		v4caccfloat	b )

concat() [43/76]

v16caccfloat concat	(	v4caccfloat	a,
		v4caccfloat	b,
		v4caccfloat	c,
		v4caccfloat	d )

concat() [44/76]

v8cfloat concat	(	v4cfloat	a0,
		v4cfloat	a1 )

concat() [45/76]

v16cfloat concat	(	v4cfloat	a0,
		v4cfloat	a1,
		v4cfloat	a2,
		v4cfloat	a3 )

concat() [46/76]

v8cint32 concat	(	v4cint32	a0,
		v4cint32	a1 )

concat() [47/76]

v16cint32 concat	(	v4cint32	a0,
		v4cint32	a1,
		v4cint32	a2,
		v4cint32	a3 )

concat() [48/76]

v128bfloat8 concat	(	v64bfloat8	a0,
		v64bfloat8	a1 )

concat() [49/76]

v128float8 concat	(	v64float8	a0,
		v64float8	a1 )

concat() [50/76]

v128int4 concat	(	v64int4	a0,
		v64int4	a1 )

concat() [51/76]

v256int4 concat	(	v64int4	a0,
		v64int4	a1,
		v64int4	a2,
		v64int4	a3 )

concat() [52/76]

v128int8 concat	(	v64int8	a0,
		v64int8	a1 )

concat() [53/76]

v128uint4 concat	(	v64uint4	a0,
		v64uint4	a1 )

concat() [54/76]

v256uint4 concat	(	v64uint4	a0,
		v64uint4	a1,
		v64uint4	a2,
		v64uint4	a3 )

concat() [55/76]

v128uint8 concat	(	v64uint8	a0,
		v64uint8	a1 )

concat() [56/76]

v16acc32 concat	(	v8acc32	a,
		v8acc32	b )

concat() [57/76]

v32acc32 concat	(	v8acc32	a,
		v8acc32	b,
		v8acc32	c,
		v8acc32	d )

concat() [58/76]

v16acc64 concat	(	v8acc64	a0,
		v8acc64	a1 )

concat() [59/76]

v32acc64 concat	(	v8acc64	a0,
		v8acc64	a1,
		v8acc64	a2,
		v8acc64	a3 )

concat() [60/76]

v16accfloat concat	(	v8accfloat	a,
		v8accfloat	b )

concat() [61/76]

v32accfloat concat	(	v8accfloat	a,
		v8accfloat	b,
		v8accfloat	c,
		v8accfloat	d )

concat() [62/76]

v16cacc64 concat	(	v8cacc64	a0,
		v8cacc64	a1 )

concat() [63/76]

v16caccfloat concat	(	v8caccfloat	a0,
		v8caccfloat	a1 )

concat() [64/76]

v32caccfloat concat	(	v8caccfloat	a0,
		v8caccfloat	a1,
		v8caccfloat	a2,
		v8caccfloat	a3 )

concat() [65/76]

v16cbfloat16 concat	(	v8cbfloat16	a0,
		v8cbfloat16	a1 )

concat() [66/76]

v32cbfloat16 concat	(	v8cbfloat16	a0,
		v8cbfloat16	a1,
		v8cbfloat16	a2,
		v8cbfloat16	a3 )

concat() [67/76]

v16cfloat concat	(	v8cfloat	a0,
		v8cfloat	a1 )

concat() [68/76]

v16cint16 concat	(	v8cint16	a0,
		v8cint16	a1 )

concat() [69/76]

v32cint16 concat	(	v8cint16	a0,
		v8cint16	a1,
		v8cint16	a2,
		v8cint16	a3 )

concat() [70/76]

v16cint32 concat	(	v8cint32	a0,
		v8cint32	a1 )

concat() [71/76]

v16float concat	(	v8float	a0,
		v8float	a1 )

concat() [72/76]

v32float concat	(	v8float	a0,
		v8float	a1,
		v8float	a2,
		v8float	a3 )

concat() [73/76]

v16int32 concat	(	v8int32	a0,
		v8int32	a1 )

concat() [74/76]

v32int32 concat	(	v8int32	a0,
		v8int32	a1,
		v8int32	a2,
		v8int32	a3 )

concat() [75/76]

v16uint32 concat	(	v8uint32	a0,
		v8uint32	a1 )

concat() [76/76]

v32uint32 concat	(	v8uint32	a0,
		v8uint32	a1,
		v8uint32	a2,
		v8uint32	a3 )