HVX INTRINSIC EXAMPLES(V68)

contact: jilong1993@gmail.com

解释中加入了一些个人理解，不代表编译器行为
打印HVX_VectorPred时，是通过转成uint8_t来打印的，所以标上的是uint8_t或者int8_t的字样, 对应的实际是bit.

Table

ALU
- add
- sub
- abs
- abs diff
- sad （Sum Absolute Diffence）
- max
- min
- avg
- Condacc
- mpy
- Mpa
- shift
Vector Permutation
- valign
- rotr
- shuffle
- deal
- pack
- unpack
- Other
Logic Operation
- compare
- Bit Operation
Vector Transfer
- data type upgrade
- sat
Scatter & Gather
- gather
- scatter
Lut
Other
- store
- splat
- setq
- prefixsum

ALU

add

Vector -> Vector

HVX_Vector Q6_Vub_vadd_VubVub_sat(HVX_Vector Vu, HVX_Vector Vv)

Vd.ub[i] = sat_u8(Vu.ub[i] + Vv.ub[i])  带有饱和的无符号加法

Vu(uint8_t)	0	1	1	255	255	255	127	128	...
Vv(uint8_t)	0	0	1	0	255	1	128	128	...
res(uint8_t)	0	1	2	255	255	255	255	255	...

HVX_Vector Q6_Vb_vadd_VbVb_sat(HVX_Vector Vu, HVX_Vector Vv)

Vd.b[i] = sat_s8(Vu.b[i] + Vv.b[i])  带有饱和的有符号加法

Vu(int8_t)	0	1	1	-1	-1	-1	127	-128	...
Vv(int8_t)	0	0	1	0	-1	1	-128	-128	...
res(int8_t)	0	1	2	-1	-2	0	-1	-128	...

Other Instructions:
HVX_Vector Q6_Vuh_vadd_VuhVuh_sat(HVX_Vector, HVX_Vector)
HVX_Vector Q6_Vh_vadd_VhVh_sat( HVX_Vector, HVX_Vector)
HVX_Vector Q6_Vuw_vadd_VuwVuw_sat(HVX_Vector, HVX_Vector)
HVX_Vector Q6_Vw_vadd_VwVw_sat( HVX_Vector, HVX_Vector)

HVX_Vector Q6_Vb_vadd_VbVb(HVX_Vector Vu, HVX_Vector Vv)

Vd.ub[i] = Vu.ub[i] + Vv.ub[i]  无饱和的加法，ub和b的行为是一样的

Vu(uint8_t)	0	1	1	255	255	255	127	128	...
Vv(uint8_t)	0	0	1	0	255	1	128	128	...
res(uint8_t)	0	1	2	255	254	0	255	0	...

Other Instructions:
HVX_Vector Q6_Vh_vadd_VhVh(HVX_Vector, HVX_Vector)
HVX_Vector Q6_Vw_vadd_VwVw(HVX_Vector, HVX_Vector)

注: 没有非sat的无符号指令, 可以用有符号的非饱和指令代替

TOADD:

HVX_Vector Q6_Vqf16_vadd_VhfVhf(HVX_Vector Vu, HVX_Vector Vv)
HVX_Vector Q6_Vqf16_vadd_Vqf16Vhf(HVX_Vector Vu, HVX_Vector Vv)
HVX_Vector Q6_Vqf16_vadd_Vqf16Vqf16(HVX_Vector Vu, HVX_Vector Vv)
HVX_Vector Q6_Vqf32_vadd_Vqf32Vqf32(HVX_Vector Vu, HVX_Vector Vv)
HVX_Vector Q6_Vqf32_vadd_Vqf32Vsf(HVX_Vector Vu, HVX_Vector Vv)
HVX_Vector Q6_Vqf32_vadd_VsfVsf(HVX_Vector Vu, HVX_Vector Vv)

HVX_Vector Q6_Vw_vadd_VwVwQ_carry_sat(HVX_Vector Vu, HVX_Vector Vv, HVX_VectorPred Qs)
HVX_Vector Q6_Vw_vadd_VwVwQ_carry(HVX_Vector Vu, HVX_Vector Vv, HVX_VectorPred Qp)
HVX_Vector Q6_Vw_vsub_VwVwQ_carry(HVX_Vector Vu, HVX_Vector Vv, HVX_VectorPred Qp)

Vector -> VectorPair

HVX_VectorPair Q6_Wh_vadd_VubVub(HVX_Vector Vu, HVX_Vector Vv)

相加，结果存入更大的数中。（需要注意，存入的顺序是奇偶分离的）
Vdd.v[0].w[i] = Vu.w[i].h[0] + Vv.w[i].h[0];  
Vdd.v[1].w[i] = Vu.w[i].h[1] + Vv.w[i].h[1];

Vu(uint8_t)	0	1	2	3	4	5	6	7	...
Vv(uint8_t)	0	1	2	3	4	5	6	7	...
res.lo(int16_t)	0	4	8	12	16	20	24	28	...
res.hi(int16_t)	2	6	10	14	18	22	26	30	...

Other Instructions:
HVX_VectorPair Q6_Wh_vadd_VubVub(HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPair Q6_Ww_vadd_VhVh( HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPair Q6_Ww_vadd_VuhVuh(HVX_Vector Vu, HVX_Vector Vv)
不存在b + b 到 h的指令 :(

VectorPair -> VectorPair

VectorPair的加法，与Vector的加法类似。（区分sat和非sat模式）

HVX_VectorPair Q6_Wb_vadd_WbWb(HVX_VectorPair Vuu, HVX_VectorPair Vvv)

Vuu(uint8_t)	0	1	2	3	4	5	6	7	...
Vvv(uint8_t)	0	1	2	3	4	5	6	7	...
res_pair(uint8_t)	0	2	4	6	8	10	12	14	...

Other Instructions:
HVX_VectorPair Q6_Wb_vadd_WbWb( HVX_VectorPair Vuu, HVX_VectorPair Vvv)
HVX_VectorPair Q6_Wb_vadd_WbWb_sat( HVX_VectorPair Vuu, HVX_VectorPair Vvv)
HVX_VectorPair Q6_Wh_vadd_WhWh( HVX_VectorPair Vuu, HVX_VectorPair Vvv)
HVX_VectorPair Q6_Wh_vadd_WhWh_sat( HVX_VectorPair Vuu, HVX_VectorPair Vvv)
HVX_VectorPair Q6_Ww_vadd_WwWw( HVX_VectorPair Vuu, HVX_VectorPair Vvv)
HVX_VectorPair Q6_Ww_vadd_WwWw_sat( HVX_VectorPair Vuu, HVX_VectorPair Vvv)
HVX_VectorPair Q6_Wub_vadd_WubWub_sat( HVX_VectorPair Vuu, HVX_VectorPair Vvv)
HVX_VectorPair Q6_Wuh_vadd_WuhWuh_sat( HVX_VectorPair Vuu, HVX_VectorPair Vvv)
HVX_VectorPair Q6_Wuw_vadd_WuwWuw_sat( HVX_VectorPair Vuu, HVX_VectorPair Vvv)

HVX_VectorPair Q6_Wh_vaddacc_WhVubVub(HVX_VectorPair Vxx, HVX_Vector Vu, HVX_Vector Vv)

Vu(uint8_t)	0	1	1	255	255	255	127	128	...
Vv(uint8_t)	0	0	1	0	255	1	128	128	...
Vxx.lo(int16_t)	1	0	1	0	1	0	1	0	...
Vxx.hi(int16_t)	1	0	1	0	1	0	1	0	...
res_pair.lo(int16_t)	1	2	511	255	1	0	1	0	...
res_pair.hi(int16_t)	2	255	257	256	1	0	1	0	...

HVX_VectorPair Q6_Ww_vaddacc_WwVhVh( HVX_VectorPair Vxx, HVX_Vector Vu, HVX_Vector Vv)

Vu(int16_t)	0	1	32767	32767	32767	-32768	-32768	-32768	...
Vv(int16_t)	0	0	1	32767	-32768	1	32767	-32768	...
Vxx.lo(int32_t)	1	1	1	1	1	1	1	1	...
Vxx.hi(int32_t)	1	1	1	1	1	1	1	1	...
res_pair.lo(int32_t)	1	32769	0	0	3	3	3	3	...
res_pair.hi(int32_t)	2	65535	-32766	-65535	1	1	1	1	...

HVX_VectorPair Q6_Ww_vaddacc_WwVuhVuh(HVX_VectorPair Vxx, HVX_Vector Vu, HVX_Vector Vv)

Vu(uint16_t)	0	1	1	65535	65535	65535	32767	32768	...
Vv(uint16_t)	0	0	1	0	65535	1	32768	32768	...
Vxx.lo(int32_t)	1	1	1	1	1	1	1	1	...
Vxx.hi(int32_t)	1	1	1	1	1	1	1	1	...
res_pair.lo(int32_t)	1	3	131071	65536	3	3	3	3	...
res_pair.hi(int32_t)	2	65536	65537	65537	1	1	1	1	...

sub

1 Vector -> Vector

HVX_Vector Q6_Vub_vsub_VubVub_sat(HVX_Vector Vu, HVX_Vector Vv)

Vd.ub[i] = sat_u8(Vu.ub[i] - Vv.ub[i]) 带饱和的无符号减法

Vu(uint8_t)	0	1	0	255	255	255	127	128	...
Vv(uint8_t)	0	0	1	0	255	1	128	128	...
res(uint8_t)	0	1	0	255	0	254	0	0	...

HVX_Vector Q6_Vb_vsub_VbVb_sat(HVX_Vector Vu, HVX_Vector Vv)

Vd.b[i] = sat_s8(Vu.b[i] - Vv.b[i]) 带饱和的有符号减法

Vu(int8_t)	0	1	1	-1	-1	-1	127	-128	...
Vv(int8_t)	0	0	1	0	-1	1	-128	-128	...
res(int8_t)	0	1	0	-1	0	-2	127	0	...

Other Instructions:
HVX_Vector Q6_Vh_vsub_VhVh_sat( HVX_Vector Vu, HVX_Vector Vv)
HVX_Vector Q6_Vuh_vsub_VuhVuh_sat(HVX_Vector Vu, HVX_Vector Vv)
HVX_Vector Q6_Vw_vsub_VwVw_sat( HVX_Vector Vu, HVX_Vector Vv)
HVX_Vector Q6_Vuw_vsub_VuwVuw_sat(HVX_Vector Vu, HVX_Vector Vv)

HVX_Vector Q6_Vb_vsub_VbVb(HVX_Vector Vu, HVX_Vector Vv)

Vd.ub[i] = Vu.ub[i] - Vv.ub[i] 带饱和的有符号减法, b和ub的行为一致。

Vu(uint8_t)	0	1	1	255	255	255	127	128	...
Vv(uint8_t)	0	0	1	0	255	1	128	128	...
res(uint8_t)	0	1	0	255	0	254	255	0	...

Other Instructions:
HVX_Vector Q6_Vh_vsub_VhVh(HVX_Vector Vu, HVX_Vector Vv)
HVX_Vector Q6_Vw_vsub_VwVw(HVX_Vector Vu, HVX_Vector Vv)
同加法, 没有无符号整型的非饱和减法, 可以用有符号整型的非饱和减法代替

HVX_Vector Q6_Vub_vsub_VubVb_sat(HVX_Vector Vu, HVX_Vector Vv)

Vd.ub[i] = sat_u8(Vu.ub[i] = Vv.b[i])  无符号ub减去有符号b，饱和后得到ub。（只有ub的指令，无对应的uh和uw指令 :(

Vu(uint8_t)	0	1	0	1	255	0	255	128	...
Vv(int8_t)	0	0	1	1	0	-1	1	-128	...
res(uint8_t)	0	1	0	0	255	1	254	255	...

2 Vector -> VectorPair

HVX_VectorPair Q6_Wh_vsub_VubVub(HVX_Vector Vu, HVX_Vector Vu)

ub相减得到h，无精度损失。注意：数据存入分奇偶
Vdd.v[0].h[i] = Vu.uh[i].ub[0] - Vv.uh[i].ub[0];
Vdd.v[1].h[i] = Vu.uh[i].ub[1] - Vv.uh[i].ub[1] ;

Vu(uint8_t)	0	1	0	255	255	255	127	128	...
Vv(uint8_t)	0	0	1	0	255	1	128	128	...
res.lo(int16_t)	0	-1	0	-1	0	0	0	0	...
res.hi(int16_t)	1	255	254	0	0	0	0	0	...

Other Instructions:
HVX_VectorPair Q6_Ww_vsub_VhVh(HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPair Q6_Ww_vsub_VuhVuh(HVX_Vector Vu, HVX_Vector Vv)
不存在b - b = h的指令

3 VectorPair -> VectorPair

对应的VectorPair减法指令，与Vector类似，区分sat和非sat模式

HVX_VectorPair Q6_Wub_vsub_WubWub_sat(HVX_VectorPair Vuu, HVX_VectorPair Vvv)

Vuu(uint8_t)	5	5	5	5	5	5	5	5	...
Vvv(uint8_t)	0	1	2	3	4	5	6	7	...
res_pair(uint8_t)	5	4	3	2	1	0	0	0	...

Other Instructions:
HVX_VectorPair Q6_Wb_vsub_WbWb( HVX_VectorPair Vuu, HVX_VectorPair Vvv)
HVX_VectorPair Q6_Wb_vsub_WbWb_sat( HVX_VectorPair Vuu, HVX_VectorPair Vvv)
HVX_VectorPair Q6_Wh_vsub_WhWh( HVX_VectorPair Vuu, HVX_VectorPair Vvv)
HVX_VectorPair Q6_Wh_vsub_WhWh_sat( HVX_VectorPair Vuu, HVX_VectorPair Vvv)
HVX_VectorPair Q6_Ww_vsub_WwWw( HVX_VectorPair Vuu, HVX_VectorPair Vvv)
HVX_VectorPair Q6_Ww_vsub_WwWw_sat( HVX_VectorPair Vuu, HVX_VectorPair Vvv)
HVX_VectorPair Q6_Wub_vsub_WubWub_sat( HVX_VectorPair Vuu, HVX_VectorPair Vvv)
HVX_VectorPair Q6_Wuh_vsub_WuhWuh_sat( HVX_VectorPair Vuu, HVX_VectorPair Vvv)
HVX_VectorPair Q6_Wuw_vsub_WuwWuw_sat( HVX_VectorPair Vuu, HVX_VectorPair Vvv)

abs

HVX_Vector Q6_Vb_vabs_Vb(HVX_Vector Vu)

注意:  abs得到的结果并不是sat的，所以如果Vu是-128，那么结果就是-128，而不是127
Vd.b[i] = abs(Vu.b[i])

Vu(int8_t)	0	1	2	127	-1	-2	-64	-128	...
res(int8_t)	0	1	2	127	1	2	64	-128	...

HVX_Vector Q6_Vb_vabs_Vb_sat(HVX_Vector Vu)

Vd.b[i] = sat_s8(abs(Vu.b[i]))  饱和的abs指令

Vu(int8_t)	0	1	2	127	-1	-2	-64	-128	...
res(int8_t)	0	1	2	127	1	2	64	127	...

Other Instructions：
HVX_Vector Q6_Vh_vabs_Vh(HVX_Vector Vu)
HVX_Vector Q6_Vw_vabs_Vw(HVX_Vector Vu)
HVX_Vector Q6_Vh_vabs_Vh_sat(HVX_Vector Vu)
HVX_Vector Q6_Vw_vabs_Vw_sat(HVX_Vector Vu)

abs diff

HVX_Vector Q6_Vub_vabsdiff_VubVub(HVX_Vector Vu,HVX_Vector Vv)

Vd.ub[i]  = abs(Vu.ub[i] - Vv.ub[i])   无符号数之间的绝对误差值

Vu(uint8_t)	0	1	2	127	255	254	64	128	...
Vv(uint8_t)	1	2	1	0	0	255	12	127	...
res(uint8_t)	1	1	1	127	255	1	52	1	...

HVX_Vector Q6_Vuh_vabsdiff_VhVh(HVX_Vector Vu,HVX_Vector Vv)

Vd.uh[i] = abs (Vu.h[i] - Vv.h[i])  有符号数之间的绝对误差值

Vu(int16_t)	32767	-32768	32767	-32768	1	16	1	65408	...
Vv(int16_t)	-32768	32767	0	0	0	1	-1	65531	...
res(uint16_t)	65535	65535	32767	32768	1	15	2	123	...

Other Instructions:

HVX_Vector Q6_Vuh_vabsdiff_VuhVuh(HVX_Vector Vu,HVX_Vector Vv)
HVX_Vector Q6_Vuw_vabsdiff_VwVw(HVX_Vector Vu,HVX_Vector Vv)

sad （Sum Absolute Diffence）

HVX_VectorPair Q6_Wuw_vdsad_WuhRuh(HVX_VectorPair Vuu, Word32 Rt)
Rt = 0xffff0001 (Rt.uh[0] = 0x0001, Rt.uh[1] = 0xFFFF)

将Vuu.uw[i]中的高低位uh 减去 Rt中的高低位uh， 得到的差值的绝对值加起来
Vdd.v[0].uw[i] = ABS(Vuu.v[0].uw[i].uh[0] - Rt.uh[0]);
Vdd.v[0].uw[i] += ABS(Vuu.v[0].uw[i].uh[1] - Rt.uh[1]);
Vdd.v[1].uw[i] = ABS(Vuu.v[0].uw[i].uh[1] - Rt.uh[0]);
Vdd.v[1].uw[i] += ABS(Vuu.v[1].uw[i].uh[0] - Rt.uh[1]) ;

e.g. Vdd.v[0].uw[0] = abs(Vuu.v[0].uw[0].uh[0] - Rt.uh[0]) + abs(Vuu.v[0].uw[0].uh[1] - Rt.uh[1]) = abs(0 - 1) + abs(1 - 0xFFFF) = 0xFFFF = 65535

Vuu.lo(uint16_t)	0	1	2	3	4	5	6	7	...
Vuu.hi(uint16_t)	64	65	66	67	68	69	70	71	...
res_pair.lo(uint32_t)	65535	65533	65533	65533	65533	65533	65533	65533	...
res_pair.hi(uint32_t)	65471	65471	65471	65471	65471	65471	65471	65471	...

HVX_VectorPair Q6_Wuw_vdsadacc_WuwWuhRuh(HVX_VectorPair Vxx, HVX_VectorPair Vuu, Word32 Rt) Rt = 0xffff0001

在上面Q6_Wuw_vdsad_WuhRuh运算的基础上，再叠加一个Vxx

Vxx.lo(uint32_t)	0	1	2	3	4	5	6	7	...
Vxx.hi(uint32_t)	32	33	34	35	36	37	38	39	...
Vuu.lo(uint32_t)	0	1	2	3	4	5	6	7	...
Vuu.hi(uint32_t)	64	65	66	67	68	69	70	71	...
res_pair.lo(uint32_t)	65535	65534	65535	65536	65537	65538	65539	65540	...
res_pair.hi(uint32_t)	65503	65504	65505	65506	65507	65508	65509	65510	...

HVX_VectorPair Q6_Wuw_vrsad_WubRubI(HVX_VectorPair Vuu, Word32 Rt, Word32 Iu1 = 0) Rt = 0xffff0001

请自行理解:(
Vdd.v[0].uw[i]  = ABS(Vuu.v[#u?1:0].uw[i].ub[0] - Rt.ub[(0- #u)&3]);
Vdd.v[0].uw[i] += ABS(Vuu.v[0 ].uw[i].ub[1] - Rt.ub[(1-#u)&3]);
Vdd.v[0].uw[i] += ABS(Vuu.v[0 ].uw[i].ub[2] - Rt.ub[(2-#u)&3]);
Vdd.v[0].uw[i] += ABS(Vuu.v[0 ].uw[i].ub[3] - Rt.ub[(3-#u)&3]);
Vdd.v[1].uw[i]   = ABS(Vuu.v[1 ].uw[i].ub[0] - Rt.ub[(2-#u)&3]);
Vdd.v[1].uw[i] += ABS(Vuu.v[1 ].uw[i].ub[1] - Rt.ub[(3-#u)&3]);
Vdd.v[1].uw[i] += ABS(Vuu.v[#u?1:0].uw[i].ub[2] - Rt.ub[(0 - #u)&3]);
Vdd.v[1].uw[i] += ABS(Vuu.v[0 ].uw[i].ub[3] - Rt.ub[(1-#u)&3]) ;

Vuu.lo(uint8_t)	0	1	2	3	4	5	6	7	...
Vuu.hi(uint8_t)	128	129	130	131	132	133	134	135	...
res_pair.lo(uint32_t)	507	505	505	505	505	505	505	505	...
res_pair.hi(uint32_t)	257	257	257	257	257	257	257	257	...

HVX_VectorPair Q6_Wuw_vrsad_WubRubI(HVX_VectorPair Vuu, Word32 Rt, Word32 Iu1 = 1) Rt = 0xffff0001

Vuu.lo(uint8_t)	0	1	2	3	4	5	6	7	...
Vuu.hi(uint8_t)	128	129	130	131	132	133	134	135	...
res_pair.lo(uint32_t)	381	381	381	381	381	381	381	381	...
res_pair.hi(uint32_t)	381	381	381	381	381	381	381	381	...

HVX_VectorPair Q6_Wuw_vrsadacc_WuwWubRubI(HVX_VectorPair Vxx, HVX_VectorPair Vuu, Word32 Rt, Word32 Iu1 = 0) Rt = 0xffff0001

在Q6_Wuw_vrsad_WubRubI的基础上再叠加一个Vxx

Vxx.lo(uint32_t)	0	1	2	3	4	5	6	7	...
Vxx.hi(uint32_t)	32	33	34	35	36	37	38	39	...
Vuu.lo(uint32_t)	0	1	2	3	4	5	6	7	...
Vuu.hi(uint32_t)	128	129	130	131	132	133	134	135	...
res_pair.lo(uint32_t)	507	506	507	508	509	510	511	512	...
res_pair.hi(uint32_t)	289	290	291	292	293	294	295	296	...

max

HVX_Vector Q6_Vb_vmax_VbVb(HVX_Vector Vu,HVX_Vector Vv)

Vd.b[i] = max(Vu.b[i], Vv.b[i])  取两个向量最大值

Vu(int8_t)	0	1	2	3	4	5	6	7	...
Vv(int8_t)	4	4	4	4	4	4	4	4	...
res(int8_t)	4	4	4	4	4	5	6	7	...

Other Instructions:
HVX_Vector Q6_Vh_vmax_VhVh( HVX_Vector Vu,HVX_Vector Vv)
HVX_Vector Q6_Vw_vmax_VwVw(HVX_Vector Vu,HVX_Vector Vv)
HVX_Vector Q6_Vub_vmax_VubVub(HVX_Vector Vu,HVX_Vector Vv)
HVX_Vector Q6_Vuh_vmax_VuhVuh(HVX_Vector Vu,HVX_Vector Vv)
HVX_Vector Q6_Vhf_vmax_VhfVhf(HVX_Vector Vu,HVX_Vector Vv)
HVX_Vector Q6_Vsf_vmax_VsfVsf(HVX_Vector Vu,HVX_Vector Vv)

min

HVX_Vector Q6_Vb_vmin_VbVb(HVX_Vector Vu,HVX_Vector Vv)

Vd.b[i] = min(Vu.b[i], Vv.b[i])  取两个向量最小值

Vu(int8_t)	0	1	2	3	4	5	6	7	...
Vv(int8_t)	4	4	4	4	4	4	4	4	...
res(int8_t)	0	1	2	3	4	4	4	4	...

Other Instructions:
HVX_Vector Q6_Vh_vmin_VhVh(HVX_Vector Vu,HVX_Vector Vv)
HVX_Vector Q6_Vw_vmin_VwVw(HVX_Vector Vu,HVX_Vector Vv)
HVX_Vector Q6_Vub_vmin_VubVub(HVX_Vector Vu,HVX_Vector Vv)
HVX_Vector Q6_Vuh_vmin_VuhVuh(HVX_Vector Vu,HVX_Vector Vv)
HVX_Vector Q6_Vhf_vmin_VhfVhf(HVX_Vector Vu,HVX_Vector Vv)
HVX_Vector Q6_Vsf_vmin_VsfVsf(HVX_Vector Vu,HVX_Vector Vv)

avg

HVX_Vector Q6_Vb_vavg_VbVb(HVX_Vector Vu, HVX_Vector Vv)

Vd.b[i] = (Vu.b[i] + Vv.b[i]) / 2； 有符号数的相加除2； （ 需要注意负数的/2，本质就是右移一位）

Vu(int8_t)	0	1	127	127	-1	-5	-128	-128	...
Vv(int8_t)	2	1	127	126	3	2	-1	-128	...
res(int8_t)	1	1	127	126	1	-2	-65	-128	...

HVX_Vector Q6_Vb_vavg_VbVb_rnd(HVX_Vector Vu, HVX_Vector Vv)

Vd.b[i] = (Vu.b[i] + Vv.b[i] + 1) / 2； 有符号数的相加，rounding后除2

Vu(int8_t)	0	1	127	127	-1	-5	-128	-128	...
Vv(int8_t)	2	1	127	126	3	2	-1	-128	...
res(int8_t)	1	1	127	127	1	-1	-64	-128	...

Other Instructions:

HVX_Vector Q6_Vh_vavg_VhVh(HVX_Vector Vu, HVX_Vector Vv)
HVX_Vector Q6_Vh_vavg_VhVh_rnd(HVX_Vector Vu, HVX_Vector Vv)
HVX_Vector Q6_Vw_vavg_VwVw(HVX_Vector Vu, HVX_Vector Vv)
HVX_Vector Q6_Vw_vavg_VwVw_rnd(HVX_Vector Vu,HVX_Vector Vv

HVX_Vector Q6_Vub_vavg_VubVub(HVX_Vector Vu, HVX_Vector Vv)

Vd.ub[i] = (Vu.ub[i] + Vv.ub[i]) / 2； 无符号数的相加除2

Vu(uint8_t)	0	1	127	127	255	254	128	255	...
Vv(uint8_t)	2	2	127	126	3	5	129	255	...
res(uint8_t)	1	1	127	126	129	129	128	255	...

HVX_Vector Q6_Vub_vavg_VubVub_rnd(HVX_Vector Vu, HVX_Vector Vv)

Vd.ub[i] = (Vu.ub[i] + Vv.ub[i] + 1) / 2； 无符号数的相加，rounding后除2

Vu(uint8_t)	0	1	127	127	255	254	128	255	...
Vv(uint8_t)	2	2	127	126	3	5	129	255	...
res(uint8_t)	1	2	127	127	129	130	129	255	...

Other Instructions:

HVX_Vector Q6_Vuh_vavg_VuhVuh(HVX_Vector Vu, HVX_Vector Vv)
HVX_Vector Q6_Vuh_vavg_VuhVuh_rnd(HVX_Vector Vu, HVX_Vector Vv
HVX_Vector Q6_Vuw_vavg_VuwVuw(HVX_Vector Vu, HVX_Vector Vv)
HVX_Vector Q6_Vuw_vavg_VuwVuw_rnd(HVX_Vector Vu, HVX_Vector Vv

HVX_Vector Q6_Vb_vnavg_VbVb(HVX_Vector Vu, HVX_Vector Vv)

Vd.b[i] = (Vu.b[i] - Vv.b[i])/2 ; 有符号数的相减除2

Vu(int8_t)	0	1	-128	127	127	127	-1	5	...
Vv(int8_t)	1	-128	1	-1	-2	-128	-2	1	...
res(int8_t)	-1	64	-65	64	64	127	0	2	...

HVX_Vector Q6_Vb_vnavg_VubVub(HVX_Vector Vu, HVX_Vector Vv)

Vd.b[i] = (Vu.ub[i] - Vv.ub[i])/2 ; 无符号数的相减除2

Vu(uint8_t)	5	7	255	0	0	1	129	130	...
Vv(uint8_t)	1	3	0	255	1	3	2	2	...
res(int8_t)	2	2	127	-128	-1	-1	63	64	...

Other Instructions:

HVX_Vector Q6_Vh_vnavg_VhVh(HVX_Vector Vu, HVX_Vector Vv)
HVX_Vector Q6_Vw_vnavg_VwVw(HVX_Vector Vu,HVX_Vector Vv)

Condacc

HVX_Vector Q6_Vb_condacc_QnVbVb(HVX_VectorPred Qv, HVX_Vector Vx, HVX_Vector Vu)

Vx.b[i] = !Qv[i] ? Vx.b[i] + Vu.b[i] : Vx.b[i]

HVX_Vector Q6_Vb_condnac_QnVbVb(HVX_VectorPred Qv, HVX_Vector Vx, HVX_Vector Vu)

Vx.b[i] = !Qv[i] ? Vx.b[i] - Vu.b[i] : Vx.b[i]

HVX_Vector Q6_Vb_condacc_QVbVb(HVX_VectorPredQv, HVX_Vector Vx, HVX_Vector Vu)

Vx.b[i] = Qv[i] ? Vx.b[i] + Vu.b[i] : Vx.b[i]

HVX_Vector Q6_Vb_condnac_QVbVb(HVX_VectorPredQv, HVX_Vector Vx, HVX_Vector Vu)

Vx.b[i] = Qv[i] ? Vx.b[i] - Vu.b[i] : Vx.b[i]

Other Instructions:

HVX_Vector Q6_Vh_condacc_QnVhVh(HVX_VectorPredQv, HVX_Vector Vx, HVX_Vector Vu)

HVX_Vector Q6_Vh_condnac_QnVhVh(HVX_VectorPredQv, HVX_Vector Vx, HVX_Vector Vu)

HVX_Vector Q6_Vw_condacc_QnVwVw(HVX_VectorPredQv, HVX_Vector Vx, HVX_Vector Vu)

HVX_Vector Q6_Vw_condnac_QnVwVw(HVX_VectorPredQv, HVX_Vector Vx, HVX_Vector Vu)

HVX_Vector Q6_Vh_condacc_QVhVh(HVX_VectorPredQv, HVX_Vector Vx, HVX_Vector Vu)

HVX_Vector Q6_Vh_condnac_QVhVh(HVX_VectorPredQv, HVX_Vector Vx, HVX_Vector Vu)

HVX_Vector Q6_Vw_condacc_QVwVw(HVX_VectorPredQv, HVX_Vector Vx, HVX_Vector Vu)

HVX_Vector Q6_Vw_condnac_QVwVw(HVX_VectorPredQv, HVX_Vector Vx, HVX_Vector Vu)

mpy

HVX_VectorPair Q6_Wh_vmpy_VbVb(HVX_Vector Vu, HVX_Vector Vv)

两个Vector的b相乘，然后放到VectorPair的h中。 （奇偶分离）
Vdd.v[0].h[i] = (Vu.h[i].b[0] * Vv.h[i].b[0]) ;
Vdd.v[1].h[i] = (Vu.h[i].b[1] * Vv.h[i].b[1]) ;

Vu(int8_t)	0	1	2	127	-1	-2	-128	-128	...
Vv(int8_t)	0	1	2	127	-1	1	127	-128	...
res_pair.lo(int16_t)	0	4	1	-16256	0	0	0	0	...
res_pair.hi(int16_t)	1	16129	-2	16384	0	0	0	0	...

HVX_VectorPair Q6_Wh_vmpy_VubVb(HVX_Vector Vu, HVX_Vector Vv)

两个Vector的ub和b相乘，然后放到VectorPair的h中。 （奇偶分离）
Vdd.v[0].h[i] = (Vu.h[i].ub[0] * Vv.h[i].b[0]) ;
Vdd.v[1].h[i] = (Vu.h[i].ub[1] * Vv.h[i].b[1]) ;

Vu(uint8_t)	0	1	2	127	128	255	255	255	...
Vv(int8_t)	0	1	2	127	-1	1	127	-128	...
res_pair.lo(int16_t)	0	4	-128	32385	0	0	0	0	...
res_pair.hi(int16_t)	1	16129	255	-32640	0	0	0	0	...

HVX_VectorPair Q6_Wuh_vmpy_VubVub(HVX_Vector Vu, HVX_Vector Vv)

两个Vector的ub和ub相乘，然后放到VectorPair的uh中。 （奇偶分离）
Vdd.v[0].uh[i] = (Vu.h[i].ub[0] * Vv.h[i].ub[0]) ;
Vdd.v[1].uh[i] = (Vu.h[i].ub[1] * Vv.h[i].ub[1]) ;

Vu(uint8_t)	0	1	2	3	4	5	6	255	...
Vv(uint8_t)	0	1	2	3	4	5	6	255	...
res_pair.lo(uint16_t)	0	4	16	36	0	0	0	0	...
res_pair.hi(uint16_t)	1	9	25	65025	0	0	0	0	...

Other Instructions:
HVX_VectorPair Q6_Wuw_vmpy_VuhVuh(HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPair Q6_Ww_vmpy_VhVh(HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPair Q6_Ww_vmpy_VhVuh(HVX_Vector Vu, HVX_Vector Vv)

HVX_VectorPair Q6_Wh_vmpyacc_WhVbVb(HVX_VectorPair Vxx, HVX_Vector Vu, HVX_Vector Vv)

Q6_Wh_vmpy_VbVb之后再叠加一个Vxx

Vu(int8_t)	0	1	2	127	-1	-2	-128	-128	...
Vv(int8_t)	0	1	2	127	-1	1	127	-128	...
Vxx.lo(int16_t)	0	1	2	3	4	5	6	7	...
Vxx.hi(int16_t)	64	65	66	67	68	69	70	71	...
res_pair.lo(int16_t)	0	5	3	-16253	4	5	6	7	...
res_pair.hi(int16_t)	65	16194	64	16451	68	69	70	71	...

Other Instructions:

HVX_VectorPair Q6_Wh_vmpyacc_WhVubVb(HVX_VectorPair Vxx, HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPair Q6_Wuh_vmpyacc_WuhVubVub(HVX_VectorPair Vxx, HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPair Q6_Wuw_vmpyacc_WuwVuhVuh(HVX_VectorPair Vxx, HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPair Q6_Ww_vmpyacc_WwVhVh(HVX_VectorPair Vxx, HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPair Q6_Ww_vmpyacc_WwVhVuh(HVX_VectorPair Vxx, HVX_Vector Vu, HVX_Vector Vv)

HVX_VectorPair Q6_Wh_vmpy_VubRb(HVX_Vector Vu, Word32 Rt) Rt = 0x04030201

Vu的ub与Rt对应的b相乘，存入Vdd
Vdd.v[0].h[i] = (Vu.uh[i].ub[0] * Rt.b[(2 * i + 0) % 4]);
Vdd.v[1].h[i] = (Vu.uh[i].ub[1] * Rt.b[(2 * i + 1) % 4]);

Vu(uint8_t)	0	1	2	3	4	5	6	255	...
res_pair.lo(int16_t)	0	6	4	18	0	0	0	0	...
res_pair.hi(int16_t)	2	12	10	1020	0	0	0	0	...

Other Instructions:

HVX_VectorPair Q6_Wuh_vmpy_VubRub(HVX_Vector Vu, Word32 Rt)
HVX_VectorPair Q6_Wuw_vmpy_VuhRuh(HVX_Vector Vu, Word32 Rt)
HVX_VectorPair Q6_Ww_vmpy_VhRh(HVX_Vector Vu, Word32 Rt)

HVX_VectorPair Q6_Wh_vmpyacc_WhVubRb(HVX_VectorPair Vxx, HVX_Vector Vu, Word32 Rt) Rt = 0x04030201

Q6_Wh_vmpy_VubRb之后叠加Vxx

Vu(uint8_t)	0	1	2	3	4	5	6	255	...
Vxx.lo(int16_t)	0	1	2	3	4	5	6	7	...
Vxx.hi(int16_t)	64	65	66	67	68	69	70	71	...
res_pair.lo(int16_t)	0	7	6	21	4	5	6	7	...
res_pair.hi(int16_t)	66	77	76	1087	68	69	70	71	...

Other Instructions:

HVX_VectorPair Q6_Wuh_vmpyacc_WuhVubRub(HVX_VectorPair Vxx, HVX_Vector Vu, Word32 Rt)
HVX_VectorPair Q6_Wuw_vmpyacc_WuwVuhRuh(HVX_VectorPair Vxx, HVX_Vector Vu, Word32 Rt)
HVX_VectorPair Q6_Ww_vmpyacc_WwVhRh(HVX_VectorPair Vxx, HVX_Vector Vu, Word32 Rt)
HVX_VectorPair Q6_Ww_vmpyacc_WwVhRh_sat(HVX_VectorPair Vxx, HVX_Vector Vu, Word32 Rt)

TODO:

HVX_Vector Q6_Vh_vmpy_VhRh_s1_rnd_sat(HVX_Vector Vu, Word32 Rt)
HVX_Vector Q6_Vh_vmpy_VhRh_s1_sat(HVX_Vector Vu, Word32 Rt)
HVX_Vector Q6_Vh_vmpy_VhVh_s1_rnd_sat(HVX_Vector Vu, HVX_Vector Vv)
HVX_Vector Q6_Vqf16_vmpy_VhfVhf(HVX_Vector Vu, HVX_Vector Vv)
HVX_Vector Q6_Vqf16_vmpy_Vqf16Vhf(HVX_Vector Vu, HVX_Vector Vv)
HVX_Vector Q6_Vqf16_vmpy_Vqf16Vqf16(HVX_Vector Vu, HVX_Vector Vv)
HVX_Vector Q6_Vqf32_vmpy_Vqf32Vqf32(HVX_Vector Vu, HVX_Vector Vv)
HVX_Vector Q6_Vqf32_vmpy_VsfVsf(HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPair Q6_Wqf32_vmpy_VhfVhf(HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPair Q6_Wqf32_vmpy_Vqf16Vhf(HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPair Q6_Wqf32_vmpy_Vqf16Vqf16(HVX_Vector Vu, HVX_Vector Vv)

HVX_VectorPair Q6_Ww_v6mpy_WubWbI_h(HVX_VectorPair Vuu, HVX_VectorPair Vvv, Word32 Iu2)

太复杂了, 不想测 :(

PseudoCode:
c00=((  (((Vvv.v[0].uw[i].ub[3] >> (2 * 0)) & 3) << 8)  | Vvv.v[0].uw[i].ub[0]  ) << 6) >> 6;
c01=((  (((Vvv.v[0].uw[i].ub[3] >> (2 * 1)) & 3) << 8)  | Vvv.v[0].uw[i].ub[1]  ) << 6) >> 6;
c02=((  (((Vvv.v[0].uw[i].ub[3] >> (2 * 2)) & 3) << 8)  | Vvv.v[0].uw[i].ub[2]  ) << 6) >> 6;
c10=((  (((Vvv.v[1].uw[i].ub[3] >> (2 * 0)) & 3) << 8)  | Vvv.v[1].uw[i].ub[0]  ) << 6) >> 6;
c11=((  (((Vvv.v[1].uw[i].ub[3] >> (2 * 1)) & 3) << 8)  | Vvv.v[1].uw[i].ub[1]  ) << 6) >> 6;
c12=((  (((Vvv.v[1].uw[i].ub[3] >> (2 * 2)) & 3) << 8)  | Vvv.v[1].uw[i].ub[2]  ) << 6) >> 6;

if Iu3 == 0
    Vdd.v[1].w[i]  = (Vuu.v[1].uw[i].ub[3] * c10);
    Vdd.v[1].w[i] += (Vuu.v[1].uw[i].ub[1] * c11);
    Vdd.v[1].w[i] += (Vuu.v[0].uw[i].ub[3] * c12);
    Vdd.v[1].w[i] += (Vuu.v[1].uw[i].ub[2] * c00);
    Vdd.v[1].w[i] += (Vuu.v[1].uw[i].ub[0] * c01);
    Vdd.v[1].w[i] += (Vuu.v[0].uw[i].ub[2] * c02);
    // w1 = (c00*ub12 + c01*ub10 + c02*ub02 + c10*ub13 + c11*ub11 + c12*ub03)

    Vdd.v[0].w[i]  = (Vuu.v[1].uw[i].ub[2] * c10);
    Vdd.v[0].w[i] += (Vuu.v[1].uw[i].ub[0] * c11);
    Vdd.v[0].w[i] += (Vuu.v[0].uw[i].ub[2] * c12);
    // w0 = (c10*ub12 + c11*ub10 + c12*ub02)

if Iu3 == 1
    Vdd.v[1].w[i]  = (Vuu.v[1].uw[i].ub[3] * c00);
    Vdd.v[1].w[i] += (Vuu.v[1].uw[i].ub[1] * c01);
    Vdd.v[1].w[i] += (Vuu.v[0].uw[i].ub[3] * c02);
    // w1 = (c00*ub13 + c01*ub11 + c02*ub03)

    Vdd.v[0].w[i]  = (Vuu.v[1].uw[i].ub[3] * c10);
    Vdd.v[0].w[i] += (Vuu.v[1].uw[i].ub[1] * c11);
    Vdd.v[0].w[i] += (Vuu.v[0].uw[i].ub[3] * c12);
    Vdd.v[0].w[i] += (Vuu.v[1].uw[i].ub[2] * c00);
    Vdd.v[0].w[i] += (Vuu.v[1].uw[i].ub[0] * c01);
    Vdd.v[0].w[i] += (Vuu.v[0].uw[i].ub[2] * c02);
    // w0 = (c00*ub12 + c01*ub10 + c02*ub02 + c10*ub13 + c11*ub11 + c12*ub03)

if Iu3 == 2
    Vdd.v[1].w[i]  = (Vuu.v[1].uw[i].ub[1] * c10);
    Vdd.v[1].w[i] += (Vuu.v[0].uw[i].ub[3] * c11);
    Vdd.v[1].w[i] += (Vuu.v[0].uw[i].ub[1] * c12);
    Vdd.v[1].w[i] += (Vuu.v[1].uw[i].ub[0] * c00);
    Vdd.v[1].w[i] += (Vuu.v[0].uw[i].ub[2] * c01);
    Vdd.v[1].w[i] += (Vuu.v[0].uw[i].ub[0] * c02);
    // w1 = (c00*ub10 + c01*ub02 + c02*ub00 + c10*ub11 + c11*ub03 + c12*ub01)

    Vdd.v[0].w[i]  = (Vuu.v[1].uw[i].ub[0] * c10);
    Vdd.v[0].w[i] += (Vuu.v[0].uw[i].ub[2] * c11);
    Vdd.v[0].w[i] += (Vuu.v[0].uw[i].ub[0] * c12);
    // w0 = (c10*ub10 + c11*ub02 + c12*ub00)

if Iu3 == 3
    Vdd.v[1].w[i]  = (Vuu.v[1].uw[i].ub[1] * c00);
    Vdd.v[1].w[i] += (Vuu.v[0].uw[i].ub[3] * c01);
    Vdd.v[1].w[i] += (Vuu.v[0].uw[i].ub[1] * c02);
    // w1 = (c00*ub11 + c01*ub03 + c02*ub01)

    Vdd.v[0].w[i]  = (Vuu.v[1].uw[i].ub[1] * c10);
    Vdd.v[0].w[i] += (Vuu.v[0].uw[i].ub[3] * c11);
    Vdd.v[0].w[i] += (Vuu.v[0].uw[i].ub[1] * c12);
    Vdd.v[0].w[i] += (Vuu.v[1].uw[i].ub[0] * c00);
    Vdd.v[0].w[i] += (Vuu.v[0].uw[i].ub[2] * c01);
    Vdd.v[0].w[i] += (Vuu.v[0].uw[i].ub[0] * c02);
    // w0 = (c00*ub10 + c01*ub02 + c02*ub00 + c10*ub11 + c11*ub03 + c12*ub01)

Other Instructions

HVX_VectorPair Q6_Ww_v6mpy_WubWbI_v(HVX_VectorPair Vuu, HVX_VectorPair Vvv, Word32 Iu2)
HVX_VectorPair Q6_Ww_v6mpyacc_WwWubWbI_h(HVX_VectorPair Vxx, HVX_VectorPair Vuu, HVX_VectorPair Vvv, Word32 Iu2)
HVX_VectorPair Q6_Ww_v6mpyacc_WwWubWbI_v(HVX_VectorPair Vxx, HVX_VectorPair Vuu, HVX_VectorPair Vvv, Word32 Iu2)

HVX_Vector Q6_Vw_vmpye_VwVuh(HVX_Vector Vu, HVX_Vector Vv)

res.w[i] = (Vu.w[i] * Vv.w[i].uh[0]) >> 16;

Vu(int32_t)	0x12345678	0x12345678	0x12345678	0x12345678	...
Vv(int32_t)	0x10001	0x100010	0x1000100	0x10001000	...
res(int32_t)	0x1234	0x12345	0x123456	0x1234567	...

HVX_Vector Q6_Vuw_vmpye_VuhRuh(HVX_Vector Vu, Word32 Rt) Rt = 0x12345678

res.uw[i] = (Vu.uw[i].uh[0] * Rt.uh[0])

Vu(uint32_t)	0x10001	0x100010	0x1000100	0x10001000	...
res(uint32_t)	0x5678	0x56780	0x567800	0x5678000	...

HVX_Vector Q6_Vuw_vmpyeacc_VuwVuhRuh(HVX_Vector Vx, HVX_Vector Vu, Word32 Rt) Rt = 0x12345678

res.uw[i] = (Vx.uw[i] + Vu.uw[i].uh[0] * Rt.uh[0])

Vx(uint32_t)	0x1	0x1	0x1	0x1	...
Vu(uint32_t)	0x10001	0x100010	0x1000100	0x10001000	...
res(uint32_t)	0x5679	0x56781	0x567801	0x5678001	...

HVX_VectorPair Q6_W_vmpye_VwVuh(HVX_Vector Vu, HVX_Vector Vv)

prod = Vu.w[i] * Vv.w[i].uh[0];
res_pair.v[0].w[i] = prod << 16; 
res_pair.v[1].w[i] = prod >> 16;

Vu(int32_t)	0x12345678	0x12345678	0x12345678	0x12345678	...
Vv(int32_t)	0x10001	0x100010	0x1000100	0x10001000	...
res_pair.lo(int32_t)	0x56780000	0x67800000	0x78000000	0x80000000	...
res_pair.hi(int32_t)	0x1234	0x12345	0x123456	0x1234567	...

HVX_Vector Q6_Vw_vmpyo_VwVh_s1_sat(HVX_Vector Vu, HVX_Vector Vv)

res.w[i] = sat_32{ (Vu.w[i] * Vv.w[i].h[1]) >> 15 };

Vu(int32_t)	0xffff8000	0xffff7000	0x80000000	0x7fffffff	...
Vv(int32_t)	0x10001	0x10001	0x80008000	0x80008000	...
res(int32_t)	0xffffffff	0xfffffffe	0x7fffffff	0x80000001	...

HVX_Vector Q6_Vw_vmpyoacc_VwVwVh_s1_sat_shift(HVX_Vector Vx, HVX_Vector Vu, HVX_Vector Vv)

res.w[i] = sat_32{ (Vx.w[i] + Vu.w[i] * Vv.w[i].h[1]) >> 15 }; <br>

Vx(int32_t)	0x8000	0x8000	0x8000	0x8000	...
Vu(int32_t)	0xffff8000	0xffff7000	0x80000000	0x7fffffff	...
Vv(int32_t)	0x10001	0x10001	0x80008000	0x80008000	...
res(int32_t)	0x0	0xffffffff	0x7fffffff	0x80000002	...

HVX_Vector Q6_Vw_vmpyo_VwVh_s1_rnd_sat(HVX_Vector Vu, HVX_Vector Vv)

res.w[i] = sat_32{  ( ((Vu.w[i] * Vv.w[i].h[1]) >> 14) + 1 ) >> 1  };

Vu(int32_t)	0xffff8000	0xffff7000	0x80000000	0x7fffffff	...
Vv(int32_t)	0x10001	0x10001	0x80008000	0x80008000	...
res(int32_t)	0xffffffff	0xffffffff	0x7fffffff	0x80000001	...

HVX_Vector Q6_Vw_vmpyoacc_VwVwVh_s1_rnd_sat_shift(HVX_Vector Vx, HVX_Vector Vu, HVX_Vector Vv)

res.w[i] = sat_32{  ( ((Vx.w[i] + Vu.w[i] * Vv.w[i].h[1]) >> 14) + 1 ) >> 1  };

Vx(int32_t)	0x8000	0x8000	0x8000	0x8000	...
Vu(int32_t)	0xffff8000	0xffff7000	0x80000000	0x7fffffff	...
Vv(int32_t)	0x10001	0x10001	0x80008000	0x80008000	...
res(int32_t)	0x0	0x0	0x7fffffff	0x80000002	...

HVX_VectorPair Q6_W_vmpyoacc_WVwVh(HVX_VectorPair Vxx, HVX_Vector Vu, HVX_Vector Vv)

prod = (Vu.w[i] * Vv.w[i].h[1]) + Vxx.v[1].w[i];
Vxx.v[1].w[i] = prod >> 16;
Vxx.v[0].w[i].h[0]=Vxx.v[0].w[i] >> 16;
Vxx.v[0].w[i].h[1]=prod & 0x0000ffff;

HVX_Vector Q6_Vh_vdmpy_VubRb(HVX_Vector Vu, Word32 Rt) Rt = 0x04030201

Vd.h[i] = Vu.ub[2*i] * Rt.b[2*i % 4] + Vu.ub[2*i + 1] * Rt.b[(2*i + 1) % 4]

Vu(uint8_t)	0	1	2	3	4	5	6	7	...
res(int16_t)	2	18	14	46	26	74	38	102	...

HVX_Vector Q6_Vh_vdmpyacc_VhVubRb(HVX_Vector Vx, HVX_Vector Vu, Word32 Rt) Rt = 0x04030201

Vd.h[i] = Vx.h[i] + Vu.ub[2*i] * Rt.b[2*i % 4] + Vu.ub[2*i + 1] * Rt.b[(2*i + 1) % 4]

Vu(uint8_t)	0	1	2	3	4	5	6	7	...
Vx(int16_t)	0	1	2	3	4	5	6	7	...
res(int16_t)	2	19	16	49	30	79	44	109	...

Other Instructions:

HVX_Vector Q6_Vw_vdmpy_VhRb(HVX_Vector Vu, Word32 Rt)
HVX_Vector Q6_Vw_vdmpyacc_VwVhRb(HVX_Vector Vx, HVX_Vector Vu, Word32 Rt)
HVX_Vector Q6_Vw_vdmpy_VhRh_sat(HVX_Vector Vu, Word32 Rt)
HVX_Vector Q6_Vw_vdmpy_VhRuh_sat(HVX_Vector Vu, Word32 Rt)
HVX_Vector Q6_Vw_vdmpyacc_VwVhRh_sat(HVX_Vector Vx, HVX_Vector Vu, Word32 Rt)
HVX_Vector Q6_Vw_vdmpyacc_VwVhRuh_sat(HVX_Vector Vx, HVX_Vector Vu, Word32 Rt)
HVX_Vector Q6_Vw_vdmpy_VhVh_sat(HVX_Vector Vu, HVX_Vector Vv)
HVX_Vector Q6_Vw_vdmpyacc_VwVhVh_sat(HVX_Vector Vx, HVX_Vector Vu, HVX_Vector Vv)
HVX_Vector Q6_Vw_vdmpy_WhRh_sat(HVX_VectorPair Vuu, Word32 Rt)
HVX_Vector Q6_Vw_vdmpyacc_VwWhRh_sat(HVX_Vector Vx, HVX_VectorPair Vuu, Word32 Rt)
HVX_Vector Q6_Vw_vdmpy_WhRuh_sat(HVX_VectorPair Vuu, Word32 Rt)
HVX_Vector Q6_Vw_vdmpyacc_VwWhRuh_sat(HVX_Vector Vx, HVX_VectorPair Vuu, Word32 Rt)
HVX_VectorPair Q6_Wh_vdmpy_WubRb(HVX_VectorPair Vuu, Word32 Rt)
HVX_VectorPair Q6_Wh_vdmpyacc_WhWubRb(HVX_VectorPair Vxx, HVX_VectorPair Vuu, Word32 Rt)
HVX_VectorPair Q6_Ww_vdmpy_WhRb(HVX_VectorPair Vuu, Word32 Rt)
HVX_VectorPair Q6_Ww_vdmpyacc_WwWhRb(HVX_VectorPair Vxx, HVX_VectorPair Vuu, Word32 Rt)

HVX_Vector Q6_Vh_vmpyi_VhVh(HVX_Vector Vu, HVX_Vector Vv)

整数乘,没有饱和操作
res.h[i] = Vu.h[i] * Vv.h[i];

Vu(int16_t)	0	1	2	-32768	-32768	32767	32767	-32768	...
Vv(int16_t)	0	1	3	2	-1	32767	-32768	-32768	...
res(int16_t)	0	1	6	0	-32768	1	-32768	0	...

HVX_Vector Q6_Vh_vmpyiacc_VhVhVh(HVX_Vector Vx, HVX_Vector Vu, HVX_Vector Vv)

Q6_Vh_vmpyi_VhVh的基础上再叠加Vx
res.h[i] = Vx.h[i] + Vu.h[i] * Vv.h[i];

Vx(int16_t)	1	1	1	1	1	1	1	1	...
Vu(int16_t)	0	1	2	-32768	-32768	32767	32767	-32768	...
Vv(int16_t)	0	1	3	2	-1	32767	-32768	-32768	...
res(int16_t)	1	2	7	1	-32767	2	-32767	1	...

HVX_Vector Q6_Vh_vmpyi_VhRb(HVX_Vector Vu, Word32 Rt)

向量和固定值的整数乘法
res.h[i] = Vu.h[i] * Rt.b[i % 4];

HVX_Vector Q6_Vh_vmpyiacc_VhVhRb(HVX_Vector Vx, HVX_Vector Vu, Word32 Rt)

Q6_Vh_vmpyi_VhRb后再叠加Vx
res.h[i] = Vx.h[i] + Vu.h[i] * Rt.b[i % 4];

Other Instructions:

HVX_Vector Q6_Vw_vmpyi_VwRb(HVX_Vector Vu, Word32 Rt)
HVX_Vector Q6_Vw_vmpyiacc_VwVwRb(HVX_Vector Vx, HVX_Vector Vu, Word32 Rt)
HVX_Vector Q6_Vw_vmpyi_VwRub(HVX_Vector Vu, Word32 Rt)
HVX_Vector Q6_Vw_vmpyiacc_VwVwRub(HVX_Vector Vx, HVX_Vector Vu, Word32 Rt)
HVX_Vector Q6_Vw_vmpyi_VwRh(HVX_Vector Vu, Word32 Rt)
HVX_Vector Q6_Vw_vmpyiacc_VwVwRh(HVX_Vector Vx, HVX_Vector Vu, Word32 Rt)

HVX_Vector Q6_Vw_vmpyie_VwVuh(HVX_Vector Vu, HVX_Vector Vv)

Vu的w和Vv的偶数位uh乘法
res.w[i] = (Vu.w[i] * Vv.w[i].uh[0]);

HVX_Vector Q6_Vw_vmpyio_VwVh(HVX_Vector Vu, HVX_Vector Vv)

Vu的w和Vv的奇数位h乘法
res.w[i] = (Vu.w[i] * Vv.w[i].h[1]);

HVX_Vector Q6_Vw_vmpyieacc_VwVwVuh(HVX_Vector Vx, HVX_Vector Vu, HVX_Vector Vv)

Q6_Vw_vmpyie_VwVuh后再叠加Vx
res.w[i] = Vx.w[i] + (Vu.w[i] * Vv.w[i].uh[0]);

HVX_Vector Q6_Vw_vmpyieacc_VwVwVh(HVX_Vector Vx, HVX_Vector Vu, HVX_Vector Vv)

Q6_Vw_vmpyio_VwVh后再叠加Vx
res.w[i] = Vx.w[i] + (Vu.w[i] * Vv.w[i].h[0]);

HVX_Vector Q6_Vw_vmpyieo_VhVh(HVX_Vector Vu, HVX_Vector Vv)

Vu.w[i]的低16位 x Vv.w[i]的高16位, 然后结果左移16位
res.w[i] = (Vu.w[i].h[0] * Vv.w[i].h[1]) << 16;

Vu(int32_t)	1	1	1	1	2	2	2	2	...
Vv(int32_t)	0x10000	0xffffffff	0x80000000	0x7fffffff	0x10000	0xffffffff	0x80000000	0x7fffffff	...
res(int32_t)	0x10000	0xffff0000	0x80000000	0x7fff0000	0x20000	0xfffe0000	0x0	0xfffe0000	...

HVX_Vector Q6_Vuw_vrmpy_VubRub(HVX_Vector Vu, Word32 Rt) Rt = 0x01020304

Vu.w[i]中的ub和常数Rt的ub对应相乘并相加 (无符号乘加)
res.uw[i] = (Vu.uw[i].ub[0] * Rt.ub[0] + 
             Vu.uw[i].ub[1] * Rt.ub[1] + 
             Vu.uw[i].ub[2] * Rt.ub[2] + 
             Vu.uw[i].ub[3] * Rt.ub[3]);

Vu(uint32_t)	0x1020304	0x5060708	0x90a0b0c	0xd0e0f10	0x1020304	0x5060708	0x90a0b0c	0xd0e0f10	...
res(uint32_t)	30	70	110	150	30	70	110	150	...

HVX_Vector Q6_Vuw_vrmpyacc_VuwVubRub(HVX_Vector Vx, HVX_Vector Vu, Word32 Rt) Rt = 0x01020304

Q6_Vuw_vrmpy_VubRub之后再叠加Vx
res.uw[i] = Vx.uw[i] + (Vu.uw[i].ub[0] * Rt.ub[0] + 
                        Vu.uw[i].ub[1] * Rt.ub[1] + 
                        Vu.uw[i].ub[2] * Rt.ub[2] + 
                        Vu.uw[i].ub[3] * Rt.ub[3]);

Vx(uint32_t)	0x1	0x1	0x1	0x1	0x1	0x1	0x1	0x1	...
Vu(uint32_t)	0x1020304	0x5060708	0x90a0b0c	0xd0e0f10	0x1020304	0x5060708	0x90a0b0c	0xd0e0f10	...
res(uint32_t)	31	71	111	151	31	71	111	151	...

HVX_Vector Q6_Vw_vrmpy_VubRb(HVX_Vector Vu, Word32 Rt) Rt = 0xfffefdfc

Vu.w[i]中的ub和常数Rt的b对应相乘并相加  (有符号乘加)
res.w[i] = (Vu.w[i].ub[0] * Rt.b[0] + 
            Vu.w[i].ub[1] * Rt.b[1] + 
            Vu.w[i].ub[2] * Rt.b[2] + 
            Vu.w[i].ub[3] * Rt.b[3]);

Vu(uint32_t)	0x1020304	0x5060708	0x90a0b0c	0xd0e0f10	0x1020304	0x5060708	0x90a0b0c	0xd0e0f10	...
res(int32_t)	-30	-70	-110	-150	-30	-70	-110	-150	...

HVX_Vector Q6_Vw_vrmpyacc_VwVubRb(HVX_Vector Vx, HVX_Vector Vu, Word32 Rt) Rt = 0xfffefdfc

Q6_Vw_vrmpy_VubRb之后再叠加Vx
res.w[i] = Vx.w[i] + (Vu.w[i].ub[0] * Rt.b[0] + 
                      Vu.w[i].ub[1] * Rt.b[1] + 
                      Vu.w[i].ub[2] * Rt.b[2] + 
                      Vu.w[i].ub[3] * Rt.b[3]);

Vx(uint32_t)	0x1	0x1	0x1	0x1	0x1	0x1	0x1	0x1	...
Vu(uint32_t)	0x1020304	0x5060708	0x90a0b0c	0xd0e0f10	0x1020304	0x5060708	0x90a0b0c	0xd0e0f10	...
res(int32_t)	-29	-69	-109	-149	-29	-69	-109	-149	...

HVX_Vector Q6_Vuw_vrmpy_VubVub(HVX_Vector Vu, HVX_Vector Vv)

Vu.w[i]与Vv.w[i]的ub进行点积
res.uw[i] = (Vu.uw[i].ub[0] * Vv.uw[i].ub[0] + 
             Vu.uw[i].ub[1] * Vv.uw[i].ub[1] + 
             Vu.uw[i].ub[2] * Vv.uw[i].ub[2] + 
             Vu.uw[i].ub[3] * Vv.uw[i].ub[3]);

Vu(uint32_t)	0x1020304	0x5060708	0x90a0b0c	0xd0e0f10	0x1020304	0x5060708	0x90a0b0c	0xd0e0f10	...
Vv(uint32_t)	0x1020304	0x5060708	0x90a0b0c	0xd0e0f10	0x1020304	0x5060708	0x90a0b0c	0xd0e0f10	...
res(uint32_t)	30	174	446	846	30	174	446	846	...

HVX_Vector Q6_Vuw_vrmpyacc_VuwVubVub(HVX_Vector Vx, HVX_Vector Vu, HVX_Vector Vv)

Q6_Vuw_vrmpy_VubVub之后再叠加Vx
res.uw[i] = Vx.uw[i] + (Vu.uw[i].ub[0] * Vv.uw[i].ub[0] + 
                        Vu.uw[i].ub[1] * Vv.uw[i].ub[1] + 
                        Vu.uw[i].ub[2] * Vv.uw[i].ub[2] + 
                        Vu.uw[i].ub[3] * Vv.uw[i].ub[3]);

Vx(uint32_t)	0x1	0x1	0x1	0x1	0x1	0x1	0x1	0x1	...
Vu(uint32_t)	0x1020304	0x5060708	0x90a0b0c	0xd0e0f10	0x1020304	0x5060708	0x90a0b0c	0xd0e0f10	...
res(uint32_t)	31	175	447	847	31	175	447	847	...

** Other Instructions:**

HVX_Vector Q6_Vw_vrmpy_VbVb(HVX_Vector Vu, HVX_Vector Vv)
HVX_Vector Q6_Vw_vrmpyacc_VwVbVb(HVX_Vector Vx, HVX_Vector Vu, HVX_Vector Vv)
HVX_Vector Q6_Vw_vrmpy_VubVb(HVX_Vector Vu, HVX_Vector Vv)
HVX_Vector Q6_Vw_vrmpyacc_VwVubVb(HVX_Vector Vx, HVX_Vector Vu, HVX_Vector Vv)

HVX_VectorPair Q6_Wuw_vrmpy_WubRubI(HVX_VectorPair Vuu, Word32 Rt, Word32 Iu1)

for (i = 0; i < VELEM(32); i++)
{
    Vdd.v[0].uw[i]  = (Vuu.v[#u ? 1:0].uw[i].ub[0] * Rt.ub[(0-#u) & 0x3]);
    Vdd.v[0].uw[i] += (Vuu.v[0       ].uw[i].ub[1] * Rt.ub[(1-#u) & 0x3]);
    Vdd.v[0].uw[i] += (Vuu.v[0       ].uw[i].ub[2] * Rt.ub[(2-#u) & 0x3]);
    Vdd.v[0].uw[i] += (Vuu.v[0       ].uw[i].ub[3] * Rt.ub[(3-#u) & 0x3]);

    Vdd.v[1].uw[i]  = (Vuu.v[1       ].uw[i].ub[0] * Rt.ub[(2-#u) & 0x3]);
    Vdd.v[1].uw[i] += (Vuu.v[1       ].uw[i].ub[1] * Rt.ub[(3-#u) & 0x3]);
    Vdd.v[1].uw[i] += (Vuu.v[#u ? 1:0].uw[i].ub[2] * Rt.ub[(0-#u) & 0x3]);
    Vdd.v[1].uw[i] += (Vuu.v[0       ].uw[i].ub[3] * Rt.ub[(1-#u) & 0x3]);
}

// when Iu1 == 0
for (i = 0; i < VELEM(32); i++)
{
    Vdd.v[0].uw[i]  = (Vuu.v[0].uw[i].ub[0] * Rt.ub[0]);
    Vdd.v[0].uw[i] += (Vuu.v[0].uw[i].ub[1] * Rt.ub[1]);
    Vdd.v[0].uw[i] += (Vuu.v[0].uw[i].ub[2] * Rt.ub[2]);
    Vdd.v[0].uw[i] += (Vuu.v[0].uw[i].ub[3] * Rt.ub[3]);

    Vdd.v[1].uw[i]  = (Vuu.v[1].uw[i].ub[0] * Rt.ub[2]);
    Vdd.v[1].uw[i] += (Vuu.v[1].uw[i].ub[1] * Rt.ub[3]);
    Vdd.v[1].uw[i] += (Vuu.v[0].uw[i].ub[2] * Rt.ub[0]);
    Vdd.v[1].uw[i] += (Vuu.v[0].uw[i].ub[3] * Rt.ub[1]);
}

// when Iu1 == 1
for (i = 0; i < VELEM(32); i++)
{
    Vdd.v[0].uw[i]  = (Vuu.v[1].uw[i].ub[0] * Rt.ub[3]);
    Vdd.v[0].uw[i] += (Vuu.v[0].uw[i].ub[1] * Rt.ub[0]);
    Vdd.v[0].uw[i] += (Vuu.v[0].uw[i].ub[2] * Rt.ub[1]);
    Vdd.v[0].uw[i] += (Vuu.v[0].uw[i].ub[3] * Rt.ub[2]);

    Vdd.v[1].uw[i]  = (Vuu.v[1].uw[i].ub[0] * Rt.ub[1]);
    Vdd.v[1].uw[i] += (Vuu.v[1].uw[i].ub[1] * Rt.ub[2]);
    Vdd.v[1].uw[i] += (Vuu.v[1].uw[i].ub[2] * Rt.ub[3]);
    Vdd.v[1].uw[i] += (Vuu.v[0].uw[i].ub[3] * Rt.ub[0]);
}

Other Instructions:

HVX_VectorPair Q6_Wuw_vrmpyacc_WuwWubRubI(HVX_VectorPair Vxx, HVX_VectorPair Vuu, Word32 Rt, Word32 Iu1)
HVX_VectorPair Q6_Ww_vrmpy_WubRbI(HVX_VectorPair Vuu, Word32 Rt, Word32 Iu1)
HVX_VectorPair Q6_Ww_vrmpyacc_WwWubRbI(HVX_VectorPair Vxx, HVX_VectorPair Vuu, Word32 Rt, Word32 Iu1)

HVX_VectorPair Q6_Wh_vtmpy_WbRb(HVX_VectorPair Vuu, Word32 Rt)

for (i = 0; i < VELEM(16); i++) {
    Vdd.v[0].h[i]  = (Vuu.v[0].h[i].b[0] * Rt.b[(2*i    ) % 4]);
    Vdd.v[0].h[i] += (Vuu.v[0].h[i].b[1] * Rt.b[(2*i + 1) % 4]);
    Vdd.v[0].h[i] +=  Vuu.v[1].h[i].b[0];

    Vdd.v[1].h[i]  = (Vuu.v[0].h[i].b[1] * Rt.b[(2*i    ) % 4]);
    Vdd.v[1].h[i] += (Vuu.v[1].h[i].b[0] * Rt.b[(2*i + 1) % 4]);
    Vdd.v[1].h[i] +=  Vuu.v[1].h[i].b[1];
}

Rt = 0xfeff0201

Vuu.lo(int16_t)	0x201	0x201	0xfeff	0xfeff	...
Vuu.hi(int16_t)	0x403	0x403	0xfcfd	0xfcfd	...
res_pair.lo(int16_t)	8	-2	-8	2	...
res_pair.hi(int16_t)	12	-4	-12	4	...

Other Instructions:

HVX_VectorPair Q6_Wh_vtmpyacc_WhWbRb(HVX_VectorPair Vxx, HVX_VectorPair Vuu, Word32 Rt)

HVX_VectorPair Q6_Wh_vtmpy_WubRb(HVX_VectorPair Vuu, Word32 Rt)
HVX_VectorPair Q6_Wh_vtmpyacc_WhWubRb(HVX_VectorPair Vxx, HVX_VectorPair Vuu, Word32 Rt)

HVX_VectorPair Q6_Ww_vtmpy_WhRb(HVX_VectorPair Vuu, Word32 Rt)
HVX_VectorPair Q6_Ww_vtmpyacc_WwWhRb(HVX_VectorPair Vxx, HVX_VectorPair Vuu, Word32 Rt)

Mpa

HVX_Vector Q6_Vh_vmpa_VhVhVhPh_sat(HVX_Vector Vx, HVX_Vector Vu, Word64 Rtt)

Vx.h[i]= sat_16((  ( (Vx.h[i] * Vu.h[i]) <<  1) + 
                   (Rtt.h[(Vu.h[i] >> 14) & 0x3] << 15)   ) >> 16)

HVX_Vector Q6_Vh_vmpa_VhVhVuhPuh_sat(HVX_Vector Vx, HVX_Vector Vu, Word64 Rtt)

Vx.h[i]= sat_16((  (Vx.h[i] * Vu.uh[i]) +
                   (Rtt.uh[((Vu.uh[i] >> 14) & 0x3)] << 15)  ) >> 16) ;

HVX_Vector Q6_Vh_vmps_VhVhVuhPuh_sat(HVX_Vector Vx, HVX_Vector Vu, Word64 Rtt)

Vx.h[i]= sat_16((  (Vx.h[i] * Vu.uh[i]) -
                   (Rtt.uh[((Vu.uh[i] >> 14) & 0x3)] << 15)  ) >>16)

HVX_VectorPair Q6_Wh_vmpa_WubRb(HVX_VectorPair Vuu, Word32 Rt)

Vdd.v[0].h[i] = (Vuu.v[0].uh[i].ub[0] * Rt.b[0]) + 
                (Vuu.v[1].uh[i].ub[0] * Rt.b[1]);
Vdd.v[1].h[i] = (Vuu.v[0].uh[i].ub[1] * Rt.b[2]) + 
                (Vuu.v[1].uh[i].ub[1] * Rt.b[3]);

HVX_VectorPair Q6_Wh_vmpaacc_WhWubRb(HVX_VectorPair Vxx, HVX_VectorPair Vuu, Word32 Rt)

Q6_Wh_vmpa_WubRb之后再叠加Vxx
Vxx.v[0].h[i] += (Vuu.v[0].uh[i].ub[0] * Rt.b[0]) + 
                 (Vuu.v[1].uh[i].ub[0] * Rt.b[1]);
Vxx.v[1].h[i] += (Vuu.v[0].uh[i].ub[1] * Rt.b[2]) + 
                 (Vuu.v[1].uh[i].ub[1] * Rt.b[3]);

HVX_VectorPair Q6_Wh_vmpa_WubRub(HVX_VectorPair Vuu, Word32 Rt)

Vdd.v[0].uh[i] = (Vuu.v[0].uh[i].ub[0] * Rt.ub[0]) + 
                 (Vuu.v[1].uh[i].ub[0] * Rt.ub[1]);
Vdd.v[1].uh[i] = (Vuu.v[0].uh[i].ub[1] * Rt.ub[2]) + 
                 (Vuu.v[1].uh[i].ub[1] * Rt.ub[3]);

HVX_VectorPair Q6_Wh_vmpaacc_WhWubRub(HVX_VectorPair Vxx, HVX_VectorPair Vuu, Word32 Rt)

Q6_Wh_vmpa_WubRub之后再叠加Vxx
Vxx.v[0].uh[i] += (Vuu.v[0].uh[i].ub[0] * Rt.ub[0]) + 
                  (Vuu.v[1].uh[i].ub[0] * Rt.ub[1]);
Vxx.v[1].uh[i] += (Vuu.v[0].uh[i].ub[1] * Rt.ub[2]) + 
                  (Vuu.v[1].uh[i].ub[1] * Rt.ub[3]);

HVX_VectorPair Q6_Ww_vmpa_WhRb(HVX_VectorPair Vuu, Word32 Rt)

Vdd.v[0].w[i] = (Vuu.v[0].w[i].h[0] * Rt.b[0]) + 
                (Vuu.v[1].w[i].h[0] * Rt.b[1]);
Vdd.v[1].w[i] = (Vuu.v[0].w[i].h[1] * Rt.b[2]) + 
                (Vuu.v[1].w[i].h[1] * Rt.b[3]) ;

HVX_VectorPair Q6_Ww_vmpaacc_WwWhRb(HVX_VectorPair Vxx, HVX_VectorPair Vuu, Word32 Rt)

Vxx.v[0].w[i] += (Vuu.v[0].w[i].h[0] * Rt.b[0]) + 
                 (Vuu.v[1].w[i].h[0] * Rt.b[1]);
Vxx.v[1].w[i] += (Vuu.v[0].w[i].h[1] * Rt.b[2]) + 
                 (Vuu.v[1].w[i].h[1] * Rt.b[3]);

HVX_VectorPair Q6_Ww_vmpa_WuhRb(HVX_VectorPair Vuu, Word32 Rt)

Vdd.v[0].w[i] = (Vuu.v[0].w[i].uh[0] * Rt.b[0]) + 
                (Vuu.v[1].w[i].uh[0] * Rt.b[1]);
Vdd.v[1].w[i] = (Vuu.v[0].w[i].uh[1] * Rt.b[2]) + 
                (Vuu.v[1].w[i].uh[1] * Rt.b[3]);

HVX_VectorPair Q6_Ww_vmpaacc_WwWuhRb(HVX_VectorPair Vxx, HVX_VectorPair Vuu, Word32 Rt)

Q6_Ww_vmpa_WuhRb之后再叠加Vxx
Vxx.v[0].w[i] += (Vuu.v[0].w[i].uh[0] * Rt.b[0]) + 
                 (Vuu.v[1].w[i].uh[0] * Rt.b[1]);
Vxx.v[1].w[i] += (Vuu.v[0].w[i].uh[1] * Rt.b[2]) + 
                 (Vuu.v[1].w[i].uh[1] * Rt.b[3]);

HVX_VectorPair Q6_Wh_vmpa_WubWb(HVX_VectorPair Vuu, HVX_VectorPair Vvv)

Vdd.v[0].h[i] = (Vuu.v[0].uh[i].ub[0] * Vvv.v[0].uh[i].b[0]) + 
                (Vuu.v[1].uh[i].ub[0] * Vvv.v[1].uh[i].b[0]);
Vdd.v[1].h[i] = (Vuu.v[0].uh[i].ub[1] * Vvv.v[0].uh[i].b[1]) + 
                (Vuu.v[1].uh[i].ub[1] * Vvv.v[1].uh[i].b[1]) ;

HVX_VectorPair Q6_Wh_vmpa_WubWub(HVX_VectorPair Vuu, HVX_VectorPair Vvv)

Vdd.v[0].h[i] = (Vuu.v[0].uh[i].ub[0] * Vvv.v[0].uh[i].ub[0]) +
                (Vuu.v[1].uh[i].ub[0] * Vvv.v[1].uh[i].ub[0]);
Vdd.v[1].h[i] = (Vuu.v[0].uh[i].ub[1] * Vvv.v[0].uh[i].ub[1]) + 
                (Vuu.v[1].uh[i].ub[1] * Vvv.v[1].uh[i].ub[1]) ;

shift

HVX_Vector Q6_Vh_vasl_VhR(HVX_Vector Vu, Word32 Rt)

常数项左移 (shift left)
Vd.h[i] = (Vu.h[i] << (Rt & (16-1)));

这里的向左移位就是简单的向左移位并在右侧补零, 操作不当会造成溢出,需要格外注意.

Rt = 1

Vu(int16_t)	1	16383	16384	32767	-1	-16384	-16385	-32768	...
res(int16_t)	2	32766	-32768	-2	-2	-32768	32766	0	...

HVX_Vector Q6_Vh_vaslacc_VhVhR(HVX_Vector Vx, HVX_Vector Vu, Word32 Rt)

Q6_Vh_vasl_VhR之后叠加Vx (需考虑溢出情况)
Vd.h[i] = Vx.h[i] + (Vu.h[i] << (Rt & (16-1)));

Rt = 1

Vx(int16_t)	2	2	2	2	-2	-2	-2	-2	...
Vu(int16_t)	1	16383	16384	32767	-1	-16384	-16385	-32768	...
res(int16_t)	4	-32768	-32766	0	-4	32766	32764	-2	...

HVX_Vector Q6_Vh_vasl_VhVh(HVX_Vector Vu, HVX_Vector Vv)

Vd.h[i] = (sxt(4+1)->16 (Vv.h[i]) > 0) ? (Vu.h[i] << sxt(4+1)->16(Vv.h[i])) :
                                         (Vu.h[i] >> sxt(4+1)->16(Vv.h[i])) ;
sxt(4+1)->16: 是一种运算符, 具体指代不清楚. sxt应该代表有符号扩展 (signed extension)

Vu(int16_t)	2	2	1	1	1	-1	-1	-1	...
Vv(int16_t)	1	-1	15	16	-1	15	16	-1	...
res(int16_t)	4	1	-32768	0	0	-32768	-1	-1	...

Other Instructions:

HVX_Vector Q6_Vw_vasl_VwR(HVX_Vector Vu, Word32 Rt)
HVX_Vector Q6_Vw_vasl_VwVw(HVX_Vector Vu, HVX_Vector Vv)
HVX_Vector Q6_Vw_vaslacc_VwVwR(HVX_Vector Vx, HVX_Vector Vu, Word32 Rt)

HVX_Vector Q6_Vh_vasr_VhR(HVX_Vector Vu, Word32 Rt)

常数项右移 (shift right)
Vd.h[i] = (Vu.h[i] >> (Rt & (16-1)));

Rt = 1

Vu(int16_t)	1	16383	16384	32767	-1	-16384	-16385	-32768	...
res(int16_t)	0	8191	8192	16383	-1	-8192	-8193	-16384	...

HVX_Vector Q6_Vh_vasracc_VhVhR(HVX_Vector Vx, HVX_Vector Vu, Word32 Rt)
Rt = 1

Q6_Vh_vasr_VhR之后再叠加Vx
Vd.h[i] = Vx.h[i] + (Vu.h[i] >> (Rt & (16-1)));

Vx(int16_t)	2	2	2	16385	-2	-2	-2	-16385	...
Vu(int16_t)	1	16383	16384	32767	-1	-16384	-16385	-32768	...
res(int16_t)	2	8193	8194	-32768	-3	-8194	-8195	32767	...

HVX_Vector Q6_Vh_vasr_VhVh(HVX_Vector Vu, HVX_Vector Vv)

Vd.h[i] = (sxt(4+1)->16(Vv.h[i]) > 0) ? (Vu.h[i] >> sxt(4+1)->16(Vv.h[i])):
                                        (Vu.h[i] << sxt(4+1)->16(Vv.h[i])) ;
}

Vu(int16_t)	2	2	1	1	1	-1	-1	-1	...
Vv(int16_t)	1	-1	15	16	-1	15	16	-1	...
res(int16_t)	1	4	0	0	2	-1	0	-2	...

** Other Instructions: **

HVX_Vector Q6_Vw_vasr_VwR(HVX_Vector Vu, Word32 Rt)
HVX_Vector Q6_Vw_vasracc_VwVwR(HVX_Vector Vx, HVX_Vector Vu, Word32 Rt)
HVX_Vector Q6_Vw_vasr_VwVw(HVX_Vector Vu, HVX_Vector Vv)

HVX_Vector Q6_Vh_vlsr_VhVh(HVX_Vector Vu, HVX_Vector Vv)

Vd.uh[i] = (sxt(4+1)->16(Vv.h[i]) > 0) ? (Vu.uh[i] >> sxt(4+1)->16(Vv.h[i])) : 
                                         (Vu.uh[i] << sxt(4+1)->16(Vv.h[i]));

Vu(int16_t)	1	16383	16384	32767	-1	-16384	-16385	-32768	...
Vv(int16_t)	1	-1	15	16	-1	15	16	-1	...
res(int16_t)	0	32766	0	0	-2	1	0	0	...

** Other Instructions: **

HVX_Vector Q6_Vw_vlsr_VwVw(HVX_Vector Vu, HVX_Vector Vv)

HVX_Vector Q6_Vub_vlsr_VubR(HVX_Vector Vu, Word32 Rt)

Vu的ub右移
Vd.b[i] = Vu.ub[i] >> (Rt & 0x7);

Vu(uint8_t)	0	1	2	3	127	128	254	255	...
res(Rt = 1, uint8_t)	0	0	1	1	63	64	127	127	...
res(Rt = -1, uint8_t)	0	0	0	0	0	1	1	1	...

** Other Instructions: **

HVX_Vector Q6_Vuh_vlsr_VuhR(HVX_Vector Vu, Word32 Rt)
HVX_Vector Q6_Vuw_vlsr_VuwR(HVX_Vector Vu, Word32 Rt)

Vector Permutation

更多讨论: HVX Permutation 介绍

valign

align: 对两个向量进行拼接
lalign: 跟algin的方向相反

HVX_Vector Q6_V_valign_VVI(HVX_Vector Vu, HVX_Vector Vv, Word32 Iu3) Iu3: immediate value, 0~7, here we set it as 7

Vu(uint8_t)	128	129	130	131	132	133	134	135	...	248	249	250	251	252	253	254	255
Vv(uint8_t)	0	1	2	3	4	5	6	7	...	120	121	122	123	124	125	126	127
res(uint8_t)	7	8	9	10	11	12	13	14	...	127	128	129	130	131	132	133	134

HVX_Vector Q6_V_valign_VVR(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt) Rt: register value, here, we set as 9

Vu(uint8_t)	128	129	130	131	132	133	134	135	136	137	...	246	247	248	249	250	251	252	253	254	255
Vv(uint8_t)	0	1	2	3	4	5	6	7	8	9	...	118	119	120	121	122	123	124	125	126	127
res(uint8_t)	9	10	11	12	13	14	15	16	17	18	...	127	128	129	130	131	132	133	134	135	136

HVX_Vector Q6_V_vlalign_VVI(HVX_Vector Vu, HVX_Vector Vv, Word32 Iu3) Iu3: immediate value, 0~7, here we set it as 7

Vu(uint8_t)	128	129	130	131	132	133	134	135	...	248	249	250	251	252	253	254	255
Vv(uint8_t)	0	1	2	3	4	5	6	7	...	120	121	122	123	124	125	126	127
res(uint8_t)	121	122	123	124	125	126	127	128	...	241	242	243	244	245	246	247	248

HVX_Vector Q6_V_vlalign_VVR(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt) Rt: register value, here, we set as 9

Vu(uint8_t)	128	129	130	131	132	133	134	135	136	137	...	246	247	248	249	250	251	252	253	254	255
Vv(uint8_t)	0	1	2	3	4	5	6	7	8	9	...	118	119	120	121	122	123	124	125	126	127
res(uint8_t)	119	120	121	122	123	124	125	126	127	128	...	237	238	239	240	241	242	243	244	245	246

rotr

ror: 一个向量进行循环右移 (rotate right)

HVX_Vector Q6_V_vror_VR(HVX_Vector Vu, Word32 Rt) Rt: register value, here, we set as 3

Vu(uint8_t)	0	1	2	3	4	5	...	122	123	124	125	126	127
res(uint8_t)	3	4	5	6	7	8	...	125	126	127	0	1	2

HVX_Vector Q6_Vuw_vrotr_VuwVuw(HVX_Vector Vu, HVX_Vector Vv)

Vu(uint32_t)	ffff0000	ffff	ff00ff00	ff00ff	f0f0f0f	f0f0f0f0	1234567	89abcdef	...
Vv(uint32_t)	4	4	4	4	4	4	4	4	...
res(uint32_t)	ffff000	f0000fff	ff00ff0	f00ff00f	f0f0f0f0	f0f0f0f	70123456	f89abcde	...

shuffle

shuffle: 洗牌指令
shuff: 向量内部的洗牌
shuffe: 向量间的洗牌, 只洗偶数(even)
shuffo: 向量间的洗牌, 只洗奇数(odd)
shuffoe: (全都要)向量间的洗牌, 奇数和偶数都洗, 偶数放在低位, 奇数放在高位
shuff_VVR: 更灵活的洗牌方式, 也比较复杂

HVX_VectorPred Q6_Qb_vshuffe_QhQh(HVX_VectorPred Qs, HVX_VectorPred Qt)

Qs(uint8_t)	0	0	0	0	0	0	0	0	...
Qt(uint8_t)	1	1	1	1	1	1	1	1	...
res_pred(uint8_t)	1	0	1	0	1	0	1	0	...

HVX_VectorPred Q6_Qh_vshuffe_QwQw(HVX_VectorPred Qs, HVX_VectorPred Qt)

Qs(uint8_t)	0	0	0	0	0	0	0	0	...
Qt(uint8_t)	1	1	1	1	1	1	1	1	...
res_pred(uint8_t)	1	1	0	0	1	1	0	0	...

HVX_Vector Q6_Vb_vshuff_Vb(HVX_Vector Vu)

Vu(uint8_t)	0	1	2	3	4	5	6	7	...	120	121	122	123	124	125	126	127
res(uint8_t)	0	64	1	65	2	66	3	67	...	60	124	61	125	62	126	63	127

HVX_Vector Q6_Vh_vshuff_Vh(HVX_Vector Vu)

Vu(uint16_t)	256	770	1284	1798	2312	2826	3340	3854	...	29040	29554	30068	30582	31096	31610	32124	32638
res(uint16_t)	256	16704	770	17218	1284	17732	1798	18246	...	14648	31096	15162	31610	15676	32124	16190	32638

HVX_Vector Q6_Vb_vshuffe_VbVb(HVX_Vector Vu, HVX_Vector Vv)

Vu(uint8_t)	128	129	130	131	132	133	134	135	...	248	249	250	251	252	253	254	255
Vv(uint8_t)	0	1	2	3	4	5	6	7	...	120	121	122	123	124	125	126	127
res(uint8_t)	0	128	2	130	4	132	6	134	...	120	248	122	250	124	252	126	254

HVX_Vector Q6_Vb_vshuffo_VbVb(HVX_Vector Vu, HVX_Vector Vv)

Vu(uint8_t)	128	129	130	131	132	133	134	135	...	248	249	250	251	252	253	254	255
Vv(uint8_t)	0	1	2	3	4	5	6	7	...	120	121	122	123	124	125	126	127
res(uint8_t)	1	129	3	131	5	133	7	135	...	121	249	123	251	125	253	127	255

HVX_Vector Q6_Vh_vshuffe_VhVh(HVX_Vector Vu, HVX_Vector Vv)

Vu(uint16_t)	64	65	66	67	68	69	70	71	...	120	121	122	123	124	125	126	127
Vv(uint16_t)	0	1	2	3	4	5	6	7	...	56	57	58	59	60	61	62	63
res(uint16_t)	0	64	2	66	4	68	6	70	...	56	120	58	122	60	124	62	126

HVX_Vector Q6_Vh_vshuffo_VhVh(HVX_Vector Vu, HVX_Vector Vv)

Vu(uint16_t)	64	65	66	67	68	69	70	71	...	120	121	122	123	124	125	126	127
Vv(uint16_t)	0	1	2	3	4	5	6	7	...	56	57	58	59	60	61	62	63
res(uint16_t)	1	65	3	67	5	69	7	71	...	57	121	59	123	61	125	63	127

HVX_VectorPair Q6_Wb_vshuffoe_VbVb(HVX_Vector Vu, HVX_Vector Vv)

Vu(uint8_t)	128	129	130	131	132	133	134	135	...	248	249	250	251	252	253	254	255
Vv(uint8_t)	0	1	2	3	4	5	6	7	...	120	121	122	123	124	125	126	127
res_pair.lo(uint8_t)	0	128	2	130	4	132	6	134	...	120	248	122	250	124	252	126	254
res_pair.hi(uint8_t)	1	129	3	131	5	133	7	135	...	121	249	123	251	125	253	127	255

HVX_VectorPair Q6_Wh_vshuffoe_VhVh(HVX_Vector Vu, HVX_Vector Vv)

Vu(uint16_t)	64	65	66	67	68	69	70	71	...	120	121	122	123	124	125	126	127
Vv(uint16_t)	0	1	2	3	4	5	6	7	...	56	57	58	59	60	61	62	63
res_pair.lo(uint16_t)	0	64	2	66	4	68	6	70	...	56	120	58	122	60	124	62	126
res_pair.hi(uint16_t)	1	65	3	67	5	69	7	71	...	57	121	59	123	61	125	63	127

如上图所示, vshuff_VVR做的大概是上面的操作, 通过来Rt来控制洗牌的粒度. (可以是1B/2B/4B/8B/16B/32B/64B) (当然也可以是128B或更大, 这种情况下相当于做了个无效操作, 此时res.lo = Vv, res.hi = Vu)

当然, 上面展示的也只是Rt为2的n次方时的情况, 我们这里记为 shuff(1), shuff(2), ...

当Rt不是2的n次方时:

shuff(3) = shuff(1) + shuff(2) : 相当于先做了一个shuff(1), 再做了一个shuff(2)

shuff(11) = shuff(1) + shuff(2) + shuff(8) 同理, 先小后大进行拆分

我们详细分析一下下面的情况:
- shuff(127) = shuff(1) + shuff(2) + ... + shuff(64) : 此时完成的是一次完美洗牌
> [ABCDEFGH] + [abcdefgh] -> [AaBbCcDdEeFfGgHh]
- shuff(-1) = shuff(127), 因为内部的处理是通过bit位来进行判断的.

**总结:**
Rt为2的n次方时, (1, 2, 4, 8, ...), 实现的是奇偶分离的洗牌, Rt为粒度;
Rt为负的2的年n次方时(-1, -2, -4), 实现的是完美洗牌, |Rt|为粒度;
其他使用情况可能需要一些理解进行灵活运用.

HVX_VectorPair Q6_W_vshuff_VVR(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt = 1)

Vu(uint8_t)	128	129	130	131	132	133	134	135	136	137	138	139	140	141	142	143	144	145	146	147	148	149	150	151	152	153	154	155	156	157	158	159	160	161	162	163	164	165	166	167	168	169	170	171	172	173	174	175	176	177	178	179	180	181	182	183	184	185	186	187	188	189	190	191	...	192	193	194	195	196	197	198	199	200	201	202	203	204	205	206	207	208	209	210	211	212	213	214	215	216	217	218	219	220	221	222	223	224	225	226	227	228	229	230	231	232	233	234	235	236	237	238	239	240	241	242	243	244	245	246	247	248	249	250	251	252	253	254	255
Vv(uint8_t)	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	...	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127
res_pair.lo(uint8_t)	0	128	2	130	4	132	6	134	8	136	10	138	12	140	14	142	16	144	18	146	20	148	22	150	24	152	26	154	28	156	30	158	32	160	34	162	36	164	38	166	40	168	42	170	44	172	46	174	48	176	50	178	52	180	54	182	56	184	58	186	60	188	62	190	...	64	192	66	194	68	196	70	198	72	200	74	202	76	204	78	206	80	208	82	210	84	212	86	214	88	216	90	218	92	220	94	222	96	224	98	226	100	228	102	230	104	232	106	234	108	236	110	238	112	240	114	242	116	244	118	246	120	248	122	250	124	252	126	254
res_pair.hi(uint8_t)	1	129	3	131	5	133	7	135	9	137	11	139	13	141	15	143	17	145	19	147	21	149	23	151	25	153	27	155	29	157	31	159	33	161	35	163	37	165	39	167	41	169	43	171	45	173	47	175	49	177	51	179	53	181	55	183	57	185	59	187	61	189	63	191	...	65	193	67	195	69	197	71	199	73	201	75	203	77	205	79	207	81	209	83	211	85	213	87	215	89	217	91	219	93	221	95	223	97	225	99	227	101	229	103	231	105	233	107	235	109	237	111	239	113	241	115	243	117	245	119	247	121	249	123	251	125	253	127	255

HVX_VectorPair Q6_W_vshuff_VVR(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt = 2)

Vu(uint8_t)	128	129	130	131	132	133	134	135	136	137	138	139	140	141	142	143	144	145	146	147	148	149	150	151	152	153	154	155	156	157	158	159	160	161	162	163	164	165	166	167	168	169	170	171	172	173	174	175	176	177	178	179	180	181	182	183	184	185	186	187	188	189	190	191	...	192	193	194	195	196	197	198	199	200	201	202	203	204	205	206	207	208	209	210	211	212	213	214	215	216	217	218	219	220	221	222	223	224	225	226	227	228	229	230	231	232	233	234	235	236	237	238	239	240	241	242	243	244	245	246	247	248	249	250	251	252	253	254	255
Vv(uint8_t)	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	...	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127
res_pair.lo(uint8_t)	0	1	128	129	4	5	132	133	8	9	136	137	12	13	140	141	16	17	144	145	20	21	148	149	24	25	152	153	28	29	156	157	32	33	160	161	36	37	164	165	40	41	168	169	44	45	172	173	48	49	176	177	52	53	180	181	56	57	184	185	60	61	188	189	...	64	65	192	193	68	69	196	197	72	73	200	201	76	77	204	205	80	81	208	209	84	85	212	213	88	89	216	217	92	93	220	221	96	97	224	225	100	101	228	229	104	105	232	233	108	109	236	237	112	113	240	241	116	117	244	245	120	121	248	249	124	125	252	253
res_pair.hi(uint8_t)	2	3	130	131	6	7	134	135	10	11	138	139	14	15	142	143	18	19	146	147	22	23	150	151	26	27	154	155	30	31	158	159	34	35	162	163	38	39	166	167	42	43	170	171	46	47	174	175	50	51	178	179	54	55	182	183	58	59	186	187	62	63	190	191	...	66	67	194	195	70	71	198	199	74	75	202	203	78	79	206	207	82	83	210	211	86	87	214	215	90	91	218	219	94	95	222	223	98	99	226	227	102	103	230	231	106	107	234	235	110	111	238	239	114	115	242	243	118	119	246	247	122	123	250	251	126	127	254	255

HVX_VectorPair Q6_W_vshuff_VVR(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt = 3)

Vu(uint8_t)	128	129	130	131	132	133	134	135	136	137	138	139	140	141	142	143	144	145	146	147	148	149	150	151	152	153	154	155	156	157	158	159	160	161	162	163	164	165	166	167	168	169	170	171	172	173	174	175	176	177	178	179	180	181	182	183	184	185	186	187	188	189	190	191	...	192	193	194	195	196	197	198	199	200	201	202	203	204	205	206	207	208	209	210	211	212	213	214	215	216	217	218	219	220	221	222	223	224	225	226	227	228	229	230	231	232	233	234	235	236	237	238	239	240	241	242	243	244	245	246	247	248	249	250	251	252	253	254	255
Vv(uint8_t)	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	...	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127
res_pair.lo(uint8_t)	0	128	1	129	4	132	5	133	8	136	9	137	12	140	13	141	16	144	17	145	20	148	21	149	24	152	25	153	28	156	29	157	32	160	33	161	36	164	37	165	40	168	41	169	44	172	45	173	48	176	49	177	52	180	53	181	56	184	57	185	60	188	61	189	...	64	192	65	193	68	196	69	197	72	200	73	201	76	204	77	205	80	208	81	209	84	212	85	213	88	216	89	217	92	220	93	221	96	224	97	225	100	228	101	229	104	232	105	233	108	236	109	237	112	240	113	241	116	244	117	245	120	248	121	249	124	252	125	253
res_pair.hi(uint8_t)	2	130	3	131	6	134	7	135	10	138	11	139	14	142	15	143	18	146	19	147	22	150	23	151	26	154	27	155	30	158	31	159	34	162	35	163	38	166	39	167	42	170	43	171	46	174	47	175	50	178	51	179	54	182	55	183	58	186	59	187	62	190	63	191	...	66	194	67	195	70	198	71	199	74	202	75	203	78	206	79	207	82	210	83	211	86	214	87	215	90	218	91	219	94	222	95	223	98	226	99	227	102	230	103	231	106	234	107	235	110	238	111	239	114	242	115	243	118	246	119	247	122	250	123	251	126	254	127	255

HVX_VectorPair Q6_W_vshuff_VVR(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt = 32)

Vu(uint8_t)	128	129	130	131	132	133	134	135	136	137	138	139	140	141	142	143	144	145	146	147	148	149	150	151	152	153	154	155	156	157	158	159	160	161	162	163	164	165	166	167	168	169	170	171	172	173	174	175	176	177	178	179	180	181	182	183	184	185	186	187	188	189	190	191	...	192	193	194	195	196	197	198	199	200	201	202	203	204	205	206	207	208	209	210	211	212	213	214	215	216	217	218	219	220	221	222	223	224	225	226	227	228	229	230	231	232	233	234	235	236	237	238	239	240	241	242	243	244	245	246	247	248	249	250	251	252	253	254	255
Vv(uint8_t)	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	...	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127
res_pair.lo(uint8_t)	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	128	129	130	131	132	133	134	135	136	137	138	139	140	141	142	143	144	145	146	147	148	149	150	151	152	153	154	155	156	157	158	159	...	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	192	193	194	195	196	197	198	199	200	201	202	203	204	205	206	207	208	209	210	211	212	213	214	215	216	217	218	219	220	221	222	223
res_pair.hi(uint8_t)	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	160	161	162	163	164	165	166	167	168	169	170	171	172	173	174	175	176	177	178	179	180	181	182	183	184	185	186	187	188	189	190	191	...	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127	224	225	226	227	228	229	230	231	232	233	234	235	236	237	238	239	240	241	242	243	244	245	246	247	248	249	250	251	252	253	254	255

HVX_VectorPair Q6_W_vshuff_VVR(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt = -1)

Vu(uint8_t)	128	129	130	131	132	133	134	135	136	137	138	139	140	141	142	143	144	145	146	147	148	149	150	151	152	153	154	155	156	157	158	159	160	161	162	163	164	165	166	167	168	169	170	171	172	173	174	175	176	177	178	179	180	181	182	183	184	185	186	187	188	189	190	191	...	192	193	194	195	196	197	198	199	200	201	202	203	204	205	206	207	208	209	210	211	212	213	214	215	216	217	218	219	220	221	222	223	224	225	226	227	228	229	230	231	232	233	234	235	236	237	238	239	240	241	242	243	244	245	246	247	248	249	250	251	252	253	254	255
Vv(uint8_t)	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	...	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127
res_pair.lo(uint8_t)	0	128	1	129	2	130	3	131	4	132	5	133	6	134	7	135	8	136	9	137	10	138	11	139	12	140	13	141	14	142	15	143	16	144	17	145	18	146	19	147	20	148	21	149	22	150	23	151	24	152	25	153	26	154	27	155	28	156	29	157	30	158	31	159	...	32	160	33	161	34	162	35	163	36	164	37	165	38	166	39	167	40	168	41	169	42	170	43	171	44	172	45	173	46	174	47	175	48	176	49	177	50	178	51	179	52	180	53	181	54	182	55	183	56	184	57	185	58	186	59	187	60	188	61	189	62	190	63	191
res_pair.hi(uint8_t)	64	192	65	193	66	194	67	195	68	196	69	197	70	198	71	199	72	200	73	201	74	202	75	203	76	204	77	205	78	206	79	207	80	208	81	209	82	210	83	211	84	212	85	213	86	214	87	215	88	216	89	217	90	218	91	219	92	220	93	221	94	222	95	223	...	96	224	97	225	98	226	99	227	100	228	101	229	102	230	103	231	104	232	105	233	106	234	107	235	108	236	109	237	110	238	111	239	112	240	113	241	114	242	115	243	116	244	117	245	118	246	119	247	120	248	121	249	122	250	123	251	124	252	125	253	126	254	127	255

HVX_VectorPair Q6_W_vshuff_VVR(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt = -2)

Vu(uint8_t)	128	129	130	131	132	133	134	135	136	137	138	139	140	141	142	143	144	145	146	147	148	149	150	151	152	153	154	155	156	157	158	159	160	161	162	163	164	165	166	167	168	169	170	171	172	173	174	175	176	177	178	179	180	181	182	183	184	185	186	187	188	189	190	191	...	192	193	194	195	196	197	198	199	200	201	202	203	204	205	206	207	208	209	210	211	212	213	214	215	216	217	218	219	220	221	222	223	224	225	226	227	228	229	230	231	232	233	234	235	236	237	238	239	240	241	242	243	244	245	246	247	248	249	250	251	252	253	254	255
Vv(uint8_t)	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	...	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127
res_pair.lo(uint8_t)	0	1	128	129	2	3	130	131	4	5	132	133	6	7	134	135	8	9	136	137	10	11	138	139	12	13	140	141	14	15	142	143	16	17	144	145	18	19	146	147	20	21	148	149	22	23	150	151	24	25	152	153	26	27	154	155	28	29	156	157	30	31	158	159	...	32	33	160	161	34	35	162	163	36	37	164	165	38	39	166	167	40	41	168	169	42	43	170	171	44	45	172	173	46	47	174	175	48	49	176	177	50	51	178	179	52	53	180	181	54	55	182	183	56	57	184	185	58	59	186	187	60	61	188	189	62	63	190	191
res_pair.hi(uint8_t)	64	65	192	193	66	67	194	195	68	69	196	197	70	71	198	199	72	73	200	201	74	75	202	203	76	77	204	205	78	79	206	207	80	81	208	209	82	83	210	211	84	85	212	213	86	87	214	215	88	89	216	217	90	91	218	219	92	93	220	221	94	95	222	223	...	96	97	224	225	98	99	226	227	100	101	228	229	102	103	230	231	104	105	232	233	106	107	234	235	108	109	236	237	110	111	238	239	112	113	240	241	114	115	242	243	116	117	244	245	118	119	246	247	120	121	248	249	122	123	250	251	124	125	252	253	126	127	254	255

HVX_VectorPair Q6_W_vshuff_VVR(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt = -3)

Vu(uint8_t)	128	129	130	131	132	133	134	135	136	137	138	139	140	141	142	143	144	145	146	147	148	149	150	151	152	153	154	155	156	157	158	159	160	161	162	163	164	165	166	167	168	169	170	171	172	173	174	175	176	177	178	179	180	181	182	183	184	185	186	187	188	189	190	191	...	192	193	194	195	196	197	198	199	200	201	202	203	204	205	206	207	208	209	210	211	212	213	214	215	216	217	218	219	220	221	222	223	224	225	226	227	228	229	230	231	232	233	234	235	236	237	238	239	240	241	242	243	244	245	246	247	248	249	250	251	252	253	254	255
Vv(uint8_t)	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	...	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127
res_pair.lo(uint8_t)	0	128	2	130	1	129	3	131	4	132	6	134	5	133	7	135	8	136	10	138	9	137	11	139	12	140	14	142	13	141	15	143	16	144	18	146	17	145	19	147	20	148	22	150	21	149	23	151	24	152	26	154	25	153	27	155	28	156	30	158	29	157	31	159	...	32	160	34	162	33	161	35	163	36	164	38	166	37	165	39	167	40	168	42	170	41	169	43	171	44	172	46	174	45	173	47	175	48	176	50	178	49	177	51	179	52	180	54	182	53	181	55	183	56	184	58	186	57	185	59	187	60	188	62	190	61	189	63	191
res_pair.hi(uint8_t)	64	192	66	194	65	193	67	195	68	196	70	198	69	197	71	199	72	200	74	202	73	201	75	203	76	204	78	206	77	205	79	207	80	208	82	210	81	209	83	211	84	212	86	214	85	213	87	215	88	216	90	218	89	217	91	219	92	220	94	222	93	221	95	223	...	96	224	98	226	97	225	99	227	100	228	102	230	101	229	103	231	104	232	106	234	105	233	107	235	108	236	110	238	109	237	111	239	112	240	114	242	113	241	115	243	116	244	118	246	117	245	119	247	120	248	122	250	121	249	123	251	124	252	126	254	125	253	127	255

HVX_VectorPair Q6_W_vshuff_VVR(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt = -32)

Vu(uint8_t)	128	129	130	131	132	133	134	135	136	137	138	139	140	141	142	143	144	145	146	147	148	149	150	151	152	153	154	155	156	157	158	159	160	161	162	163	164	165	166	167	168	169	170	171	172	173	174	175	176	177	178	179	180	181	182	183	184	185	186	187	188	189	190	191	...	192	193	194	195	196	197	198	199	200	201	202	203	204	205	206	207	208	209	210	211	212	213	214	215	216	217	218	219	220	221	222	223	224	225	226	227	228	229	230	231	232	233	234	235	236	237	238	239	240	241	242	243	244	245	246	247	248	249	250	251	252	253	254	255
Vv(uint8_t)	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	...	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127
res_pair.lo(uint8_t)	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	128	129	130	131	132	133	134	135	136	137	138	139	140	141	142	143	144	145	146	147	148	149	150	151	152	153	154	155	156	157	158	159	...	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	160	161	162	163	164	165	166	167	168	169	170	171	172	173	174	175	176	177	178	179	180	181	182	183	184	185	186	187	188	189	190	191
res_pair.hi(uint8_t)	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	192	193	194	195	196	197	198	199	200	201	202	203	204	205	206	207	208	209	210	211	212	213	214	215	216	217	218	219	220	221	222	223	...	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127	224	225	226	227	228	229	230	231	232	233	234	235	236	237	238	239	240	241	242	243	244	245	246	247	248	249	250	251	252	253	254	255

deal

HVX_Vector Q6_Vb_vdeal_Vb(HVX_Vector Vu)

Vu(uint8_t)	0	1	2	3	4	5	6	7	...	120	121	122	123	124	125	126	127
res(uint8_t)	0	2	4	6	8	10	12	14	...	113	115	117	119	121	123	125	127

HVX_Vector Q6_Vb_vdeale_VbVb(HVX_Vector Vu, HVX_Vector Vv)

Vu(uint8_t)	128	129	130	131	132	133	134	135	136	137	138	139	140	141	142	143	144	145	146	147	148	149	150	151	152	153	154	155	156	157	158	159	160	161	162	163	164	165	166	167	...	216	217	218	219	220	221	222	223	224	225	226	227	228	229	230	231	232	233	234	235	236	237	238	239	240	241	242	243	244	245	246	247	248	249	250	251	252	253	254	255
Vv(uint8_t)	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	...	88	89	90	91	92	93	94	95	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127
res(uint8_t)	0	4	8	12	16	20	24	28	32	36	40	44	48	52	56	60	64	68	72	76	80	84	88	92	96	100	104	108	112	116	120	124	2	6	10	14	18	22	26	30	...	224	228	232	236	240	244	248	252	130	134	138	142	146	150	154	158	162	166	170	174	178	182	186	190	194	198	202	206	210	214	218	222	226	230	234	238	242	246	250	254

HVX_Vector Q6_Vh_vdeal_Vh(HVX_Vector Vu)

Vu(uint16_t)	0	1	2	3	4	5	6	7	...	56	57	58	59	60	61	62	63
res(uint16_t)	0	2	4	6	8	10	12	14	...	49	51	53	55	57	59	61	63

deal: 发牌.
可以想象成一副牌发给两个人的操作.

vdeal: 一副牌均匀发给两个人, 那么一人都是偶数牌, 一人都是奇数牌;
vdeale: 一副牌均匀发给两人人, 但是只发偶数牌(even), 所以一人手里拿到的牌的序列都是4k, 一人手里都是 2*(2k + 1)

当Rt是正的2的n次方时, deal_VVR的指令与deal_VV的指令是等价的, 我们记为 deal(Rt)
当Rt不是正的2的n次方时, 或者为负数的时候, deal_VVR的指令同样可以拆解成多个2的n次方的deal指令, 但是顺序是从大到小的, 与shuff_VVR的指令顺序相反
e.g.

deal(3) = deal(2) + deal(1) : 相当于先做了一个deal(2), 再做了一个deal(1)

deal(11) = deal(8) + deal(2) + deal(1)
因此, 当Rt为负数的时候, 结果也跟shuff_VVR的指令完全不一样. 具体见测试结果
总结:
vdeal_VVR 与 vshuff_VVR互为逆运算, 适用于Rt为任何值的时候.

HVX_VectorPair Q6_W_vdeal_VVR(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt =1)

Vu(uint8_t)	128	129	130	131	132	133	134	135	136	137	138	139	140	141	142	143	144	145	146	147	148	149	150	151	152	153	154	155	156	157	158	159	160	161	162	163	164	165	166	167	168	169	170	171	172	173	174	175	176	177	178	179	180	181	182	183	184	185	186	187	188	189	190	191	...	192	193	194	195	196	197	198	199	200	201	202	203	204	205	206	207	208	209	210	211	212	213	214	215	216	217	218	219	220	221	222	223	224	225	226	227	228	229	230	231	232	233	234	235	236	237	238	239	240	241	242	243	244	245	246	247	248	249	250	251	252	253	254	255
Vv(uint8_t)	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	...	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127
res_pair.lo(uint8_t)	0	128	2	130	4	132	6	134	8	136	10	138	12	140	14	142	16	144	18	146	20	148	22	150	24	152	26	154	28	156	30	158	32	160	34	162	36	164	38	166	40	168	42	170	44	172	46	174	48	176	50	178	52	180	54	182	56	184	58	186	60	188	62	190	...	64	192	66	194	68	196	70	198	72	200	74	202	76	204	78	206	80	208	82	210	84	212	86	214	88	216	90	218	92	220	94	222	96	224	98	226	100	228	102	230	104	232	106	234	108	236	110	238	112	240	114	242	116	244	118	246	120	248	122	250	124	252	126	254
res_pair.hi(uint8_t)	1	129	3	131	5	133	7	135	9	137	11	139	13	141	15	143	17	145	19	147	21	149	23	151	25	153	27	155	29	157	31	159	33	161	35	163	37	165	39	167	41	169	43	171	45	173	47	175	49	177	51	179	53	181	55	183	57	185	59	187	61	189	63	191	...	65	193	67	195	69	197	71	199	73	201	75	203	77	205	79	207	81	209	83	211	85	213	87	215	89	217	91	219	93	221	95	223	97	225	99	227	101	229	103	231	105	233	107	235	109	237	111	239	113	241	115	243	117	245	119	247	121	249	123	251	125	253	127	255

HVX_VectorPair Q6_W_vdeal_VVR(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt =2)

Vu(uint8_t)	128	129	130	131	132	133	134	135	136	137	138	139	140	141	142	143	144	145	146	147	148	149	150	151	152	153	154	155	156	157	158	159	160	161	162	163	164	165	166	167	168	169	170	171	172	173	174	175	176	177	178	179	180	181	182	183	184	185	186	187	188	189	190	191	...	192	193	194	195	196	197	198	199	200	201	202	203	204	205	206	207	208	209	210	211	212	213	214	215	216	217	218	219	220	221	222	223	224	225	226	227	228	229	230	231	232	233	234	235	236	237	238	239	240	241	242	243	244	245	246	247	248	249	250	251	252	253	254	255
Vv(uint8_t)	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	...	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127
res_pair.lo(uint8_t)	0	1	128	129	4	5	132	133	8	9	136	137	12	13	140	141	16	17	144	145	20	21	148	149	24	25	152	153	28	29	156	157	32	33	160	161	36	37	164	165	40	41	168	169	44	45	172	173	48	49	176	177	52	53	180	181	56	57	184	185	60	61	188	189	...	64	65	192	193	68	69	196	197	72	73	200	201	76	77	204	205	80	81	208	209	84	85	212	213	88	89	216	217	92	93	220	221	96	97	224	225	100	101	228	229	104	105	232	233	108	109	236	237	112	113	240	241	116	117	244	245	120	121	248	249	124	125	252	253
res_pair.hi(uint8_t)	2	3	130	131	6	7	134	135	10	11	138	139	14	15	142	143	18	19	146	147	22	23	150	151	26	27	154	155	30	31	158	159	34	35	162	163	38	39	166	167	42	43	170	171	46	47	174	175	50	51	178	179	54	55	182	183	58	59	186	187	62	63	190	191	...	66	67	194	195	70	71	198	199	74	75	202	203	78	79	206	207	82	83	210	211	86	87	214	215	90	91	218	219	94	95	222	223	98	99	226	227	102	103	230	231	106	107	234	235	110	111	238	239	114	115	242	243	118	119	246	247	122	123	250	251	126	127	254	255

HVX_VectorPair Q6_W_vdeal_VVR(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt =3)

Vu(uint8_t)	128	129	130	131	132	133	134	135	136	137	138	139	140	141	142	143	144	145	146	147	148	149	150	151	152	153	154	155	156	157	158	159	160	161	162	163	164	165	166	167	168	169	170	171	172	173	174	175	176	177	178	179	180	181	182	183	184	185	186	187	188	189	190	191	...	192	193	194	195	196	197	198	199	200	201	202	203	204	205	206	207	208	209	210	211	212	213	214	215	216	217	218	219	220	221	222	223	224	225	226	227	228	229	230	231	232	233	234	235	236	237	238	239	240	241	242	243	244	245	246	247	248	249	250	251	252	253	254	255
Vv(uint8_t)	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	...	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127
res_pair.lo(uint8_t)	0	2	128	130	4	6	132	134	8	10	136	138	12	14	140	142	16	18	144	146	20	22	148	150	24	26	152	154	28	30	156	158	32	34	160	162	36	38	164	166	40	42	168	170	44	46	172	174	48	50	176	178	52	54	180	182	56	58	184	186	60	62	188	190	...	64	66	192	194	68	70	196	198	72	74	200	202	76	78	204	206	80	82	208	210	84	86	212	214	88	90	216	218	92	94	220	222	96	98	224	226	100	102	228	230	104	106	232	234	108	110	236	238	112	114	240	242	116	118	244	246	120	122	248	250	124	126	252	254
res_pair.hi(uint8_t)	1	3	129	131	5	7	133	135	9	11	137	139	13	15	141	143	17	19	145	147	21	23	149	151	25	27	153	155	29	31	157	159	33	35	161	163	37	39	165	167	41	43	169	171	45	47	173	175	49	51	177	179	53	55	181	183	57	59	185	187	61	63	189	191	...	65	67	193	195	69	71	197	199	73	75	201	203	77	79	205	207	81	83	209	211	85	87	213	215	89	91	217	219	93	95	221	223	97	99	225	227	101	103	229	231	105	107	233	235	109	111	237	239	113	115	241	243	117	119	245	247	121	123	249	251	125	127	253	255

HVX_VectorPair Q6_W_vdeal_VVR(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt =16)

Vu(uint8_t)	128	129	130	131	132	133	134	135	136	137	138	139	140	141	142	143	144	145	146	147	148	149	150	151	152	153	154	155	156	157	158	159	160	161	162	163	164	165	166	167	168	169	170	171	172	173	174	175	176	177	178	179	180	181	182	183	184	185	186	187	188	189	190	191	...	192	193	194	195	196	197	198	199	200	201	202	203	204	205	206	207	208	209	210	211	212	213	214	215	216	217	218	219	220	221	222	223	224	225	226	227	228	229	230	231	232	233	234	235	236	237	238	239	240	241	242	243	244	245	246	247	248	249	250	251	252	253	254	255
Vv(uint8_t)	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	...	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127
res_pair.lo(uint8_t)	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	128	129	130	131	132	133	134	135	136	137	138	139	140	141	142	143	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	160	161	162	163	164	165	166	167	168	169	170	171	172	173	174	175	...	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	192	193	194	195	196	197	198	199	200	201	202	203	204	205	206	207	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	224	225	226	227	228	229	230	231	232	233	234	235	236	237	238	239
res_pair.hi(uint8_t)	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	144	145	146	147	148	149	150	151	152	153	154	155	156	157	158	159	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	176	177	178	179	180	181	182	183	184	185	186	187	188	189	190	191	...	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	208	209	210	211	212	213	214	215	216	217	218	219	220	221	222	223	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127	240	241	242	243	244	245	246	247	248	249	250	251	252	253	254	255

HVX_VectorPair Q6_W_vdeal_VVR(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt =-1)

Vu(uint8_t)	128	129	130	131	132	133	134	135	136	137	138	139	140	141	142	143	144	145	146	147	148	149	150	151	152	153	154	155	156	157	158	159	160	161	162	163	164	165	166	167	168	169	170	171	172	173	174	175	176	177	178	179	180	181	182	183	184	185	186	187	188	189	190	191	...	192	193	194	195	196	197	198	199	200	201	202	203	204	205	206	207	208	209	210	211	212	213	214	215	216	217	218	219	220	221	222	223	224	225	226	227	228	229	230	231	232	233	234	235	236	237	238	239	240	241	242	243	244	245	246	247	248	249	250	251	252	253	254	255
Vv(uint8_t)	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	...	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127
res_pair.lo(uint8_t)	0	2	4	6	8	10	12	14	16	18	20	22	24	26	28	30	32	34	36	38	40	42	44	46	48	50	52	54	56	58	60	62	64	66	68	70	72	74	76	78	80	82	84	86	88	90	92	94	96	98	100	102	104	106	108	110	112	114	116	118	120	122	124	126	...	128	130	132	134	136	138	140	142	144	146	148	150	152	154	156	158	160	162	164	166	168	170	172	174	176	178	180	182	184	186	188	190	192	194	196	198	200	202	204	206	208	210	212	214	216	218	220	222	224	226	228	230	232	234	236	238	240	242	244	246	248	250	252	254
res_pair.hi(uint8_t)	1	3	5	7	9	11	13	15	17	19	21	23	25	27	29	31	33	35	37	39	41	43	45	47	49	51	53	55	57	59	61	63	65	67	69	71	73	75	77	79	81	83	85	87	89	91	93	95	97	99	101	103	105	107	109	111	113	115	117	119	121	123	125	127	...	129	131	133	135	137	139	141	143	145	147	149	151	153	155	157	159	161	163	165	167	169	171	173	175	177	179	181	183	185	187	189	191	193	195	197	199	201	203	205	207	209	211	213	215	217	219	221	223	225	227	229	231	233	235	237	239	241	243	245	247	249	251	253	255

HVX_VectorPair Q6_W_vdeal_VVR(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt =-2)

Vu(uint8_t)	128	129	130	131	132	133	134	135	136	137	138	139	140	141	142	143	144	145	146	147	148	149	150	151	152	153	154	155	156	157	158	159	160	161	162	163	164	165	166	167	168	169	170	171	172	173	174	175	176	177	178	179	180	181	182	183	184	185	186	187	188	189	190	191	...	192	193	194	195	196	197	198	199	200	201	202	203	204	205	206	207	208	209	210	211	212	213	214	215	216	217	218	219	220	221	222	223	224	225	226	227	228	229	230	231	232	233	234	235	236	237	238	239	240	241	242	243	244	245	246	247	248	249	250	251	252	253	254	255
Vv(uint8_t)	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	...	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127
res_pair.lo(uint8_t)	0	1	4	5	8	9	12	13	16	17	20	21	24	25	28	29	32	33	36	37	40	41	44	45	48	49	52	53	56	57	60	61	64	65	68	69	72	73	76	77	80	81	84	85	88	89	92	93	96	97	100	101	104	105	108	109	112	113	116	117	120	121	124	125	...	128	129	132	133	136	137	140	141	144	145	148	149	152	153	156	157	160	161	164	165	168	169	172	173	176	177	180	181	184	185	188	189	192	193	196	197	200	201	204	205	208	209	212	213	216	217	220	221	224	225	228	229	232	233	236	237	240	241	244	245	248	249	252	253
res_pair.hi(uint8_t)	2	3	6	7	10	11	14	15	18	19	22	23	26	27	30	31	34	35	38	39	42	43	46	47	50	51	54	55	58	59	62	63	66	67	70	71	74	75	78	79	82	83	86	87	90	91	94	95	98	99	102	103	106	107	110	111	114	115	118	119	122	123	126	127	...	130	131	134	135	138	139	142	143	146	147	150	151	154	155	158	159	162	163	166	167	170	171	174	175	178	179	182	183	186	187	190	191	194	195	198	199	202	203	206	207	210	211	214	215	218	219	222	223	226	227	230	231	234	235	238	239	242	243	246	247	250	251	254	255

HVX_VectorPair Q6_W_vdeal_VVR(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt =-3)

Vu(uint8_t)	128	129	130	131	132	133	134	135	136	137	138	139	140	141	142	143	144	145	146	147	148	149	150	151	152	153	154	155	156	157	158	159	160	161	162	163	164	165	166	167	168	169	170	171	172	173	174	175	176	177	178	179	180	181	182	183	184	185	186	187	188	189	190	191	...	192	193	194	195	196	197	198	199	200	201	202	203	204	205	206	207	208	209	210	211	212	213	214	215	216	217	218	219	220	221	222	223	224	225	226	227	228	229	230	231	232	233	234	235	236	237	238	239	240	241	242	243	244	245	246	247	248	249	250	251	252	253	254	255
Vv(uint8_t)	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	...	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127
res_pair.lo(uint8_t)	0	4	2	6	8	12	10	14	16	20	18	22	24	28	26	30	32	36	34	38	40	44	42	46	48	52	50	54	56	60	58	62	64	68	66	70	72	76	74	78	80	84	82	86	88	92	90	94	96	100	98	102	104	108	106	110	112	116	114	118	120	124	122	126	...	128	132	130	134	136	140	138	142	144	148	146	150	152	156	154	158	160	164	162	166	168	172	170	174	176	180	178	182	184	188	186	190	192	196	194	198	200	204	202	206	208	212	210	214	216	220	218	222	224	228	226	230	232	236	234	238	240	244	242	246	248	252	250	254
res_pair.hi(uint8_t)	1	5	3	7	9	13	11	15	17	21	19	23	25	29	27	31	33	37	35	39	41	45	43	47	49	53	51	55	57	61	59	63	65	69	67	71	73	77	75	79	81	85	83	87	89	93	91	95	97	101	99	103	105	109	107	111	113	117	115	119	121	125	123	127	...	129	133	131	135	137	141	139	143	145	149	147	151	153	157	155	159	161	165	163	167	169	173	171	175	177	181	179	183	185	189	187	191	193	197	195	199	201	205	203	207	209	213	211	215	217	221	219	223	225	229	227	231	233	237	235	239	241	245	243	247	249	253	251	255

HVX_VectorPair Q6_W_vdeal_VVR(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt =-16)

Vu(uint8_t)	128	129	130	131	132	133	134	135	136	137	138	139	140	141	142	143	144	145	146	147	148	149	150	151	152	153	154	155	156	157	158	159	160	161	162	163	164	165	166	167	168	169	170	171	172	173	174	175	176	177	178	179	180	181	182	183	184	185	186	187	188	189	190	191	...	192	193	194	195	196	197	198	199	200	201	202	203	204	205	206	207	208	209	210	211	212	213	214	215	216	217	218	219	220	221	222	223	224	225	226	227	228	229	230	231	232	233	234	235	236	237	238	239	240	241	242	243	244	245	246	247	248	249	250	251	252	253	254	255
Vv(uint8_t)	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	...	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127
res_pair.lo(uint8_t)	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	...	128	129	130	131	132	133	134	135	136	137	138	139	140	141	142	143	160	161	162	163	164	165	166	167	168	169	170	171	172	173	174	175	192	193	194	195	196	197	198	199	200	201	202	203	204	205	206	207	224	225	226	227	228	229	230	231	232	233	234	235	236	237	238	239
res_pair.hi(uint8_t)	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127	...	144	145	146	147	148	149	150	151	152	153	154	155	156	157	158	159	176	177	178	179	180	181	182	183	184	185	186	187	188	189	190	191	208	209	210	211	212	213	214	215	216	217	218	219	220	221	222	223	240	241	242	243	244	245	246	247	248	249	250	251	252	253	254	255

pack

这里的pack是指将两个向量合并成一个向量, 但是合并的方式不同;
packe: 将两个向量的低位合并.(有half和word两种单位)
packo: 将两个向量的高位合并.(有half和word两种单位)
pack_sat: 向量的降级. (h -> b | h -> ub | w -> h | w -> uh)

HVX_Vector Q6_Vb_vpacke_VhVh(HVX_Vector Vu, HVX_Vector Vv)

for i = 0 : 64
    Vd.ub[i]      = Vv.uh[i].ub[0];
    Vd.ub[i + 64] = Vu.uh[i].ub[0];

Vu(uint8_t)	128	129	130	131	132	133	134	135	...	248	249	250	251	252	253	254	255
Vv(uint8_t)	0	1	2	3	4	5	6	7	...	120	121	122	123	124	125	126	127
res(uint8_t)	0	2	4	6	8	10	12	14	...	240	242	244	246	248	250	252	254

HVX_Vector Q6_Vb_vpacko_VhVh(HVX_Vector Vu, HVX_Vector Vv)

for i = 0 : 64
    Vd.ub[i]      = Vv.uh[i].ub[1];
    Vd.ub[i + 64] = Vu.uh[i].ub[1];

Vu(uint8_t)	128	129	130	131	132	133	134	135	...	248	249	250	251	252	253	254	255
Vv(uint8_t)	0	1	2	3	4	5	6	7	...	120	121	122	123	124	125	126	127
res(uint8_t)	1	3	5	7	9	11	13	15	...	241	243	245	247	249	251	253	255

HVX_Vector Q6_Vh_vpacke_VwVw(HVX_Vector Vu, HVX_Vector Vv)

for i = 0 : 32
    Vd.uh[i]      = Vv.uw[i].uh[0];
    Vd.uh[i + 32] = Vu.uw[i].uh[0];

Vu(uint16_t)	64	65	66	67	68	69	70	71	...	120	121	122	123	124	125	126	127
Vv(uint16_t)	0	1	2	3	4	5	6	7	...	56	57	58	59	60	61	62	63
res(uint16_t)	0	2	4	6	8	10	12	14	...	112	114	116	118	120	122	124	126

HVX_Vector Q6_Vh_vpacko_VwVw(HVX_Vector Vu, HVX_Vector Vv)

for i = 0 : 32
    Vd.uh[i]      = Vv.uw[i].uh[1];
    Vd.uh[i + 32] = Vu.uw[i].uh[1];

Vu(uint16_t)	64	65	66	67	68	69	70	71	...	120	121	122	123	124	125	126	127
Vv(uint16_t)	0	1	2	3	4	5	6	7	...	56	57	58	59	60	61	62	63
res(uint16_t)	1	3	5	7	9	11	13	15	...	113	115	117	119	121	123	125	127

HVX_Vector Q6_Vb_vpack_VhVh_sat(HVX_Vector Vu, HVX_Vector Vv)

for i = 0 : 64
    Vd.b[i]      = sat8(Vv.h[i]);
    Vd.b[i + 64] = sat8(Vu.h[i]) ;

Vu(int16_t)	68	69	70	71	72	73	74	75	...	124	125	126	127	128	129	130	131
Vv(int16_t)	-132	-131	-130	-129	-128	-127	-126	-125	...	-76	-75	-74	-73	-72	-71	-70	-69
res(int8_t)	-128	-128	-128	-128	-128	-127	-126	-125	...	124	125	126	127	127	127	127	127

HVX_Vector Q6_Vub_vpack_VhVh_sat(HVX_Vector Vu, HVX_Vector Vv)

for i = 0 : 64
    Vd.ub[i]      = usat8(Vv.h[i]);
    Vd.ub[i + 64] = usat8(Vu.h[i]);

Vu(int16_t)	196	197	198	199	200	201	202	203	...	252	253	254	255	256	257	258	259
Vv(int16_t)	-4	-3	-2	-1	0	1	2	3	...	52	53	54	55	56	57	58	59
res(uint8_t)	0	0	0	0	0	1	2	3	...	252	253	254	255	255	255	255	255

HVX_Vector Q6_Vh_vpack_VwVw_sat(HVX_Vector Vu, HVX_Vector Vv)

for i = 0 : 32
    Vd.h[i]      = sat16(Vv.w[i]);
    Vd.h[i + 32] = sat16(Vu.w[i]) ;

Vu(int32_t)	32740	32741	32742	32743	32744	32745	32746	32747	...	32764	32765	32766	32767	32768	32769	32770	32771
Vv(int32_t)	-32772	-32771	-32770	-32769	-32768	-32767	-32766	-32765	...	-32748	-32747	-32746	-32745	-32744	-32743	-32742	-32741
res(int16_t)	-32768	-32768	-32768	-32768	-32768	-32767	-32766	-32765	...	32764	32765	32766	32767	32767	32767	32767	32767

HVX_Vector Q6_Vuh_vpack_VwVw_sat(HVX_Vector Vu, HVX_Vector Vv)

for i = 0 : 32
    Vd.ub[i]      = usat16(Vv.w[i]);
    Vd.ub[i + 32] = usat16(Vu.w[i]);

Vu(int32_t)	65508	65509	65510	65511	65512	65513	65514	65515	...	65532	65533	65534	65535	65536	65537	65538	65539
Vv(int32_t)	-4	-3	-2	-1	0	1	2	3	...	20	21	22	23	24	25	26	27
res(uint16_t)	0	0	0	0	0	1	2	3	...	65532	65533	65534	65535	65535	65535	65535	65535

unpack

HVX_VectorPair Q6_Wuh_vunpack_Vub(HVX_Vector Vu)

for i = 0 : 128
    Vdd.uh[i] = Vu.ub[i];

Vu(uint8_t)	0	1	2	3	4	5	6	7	...	120	121	122	123	124	125	126	127
res_pair(uint16_t)	0	1	2	3	4	5	6	7	...	120	121	122	123	124	125	126	127

other instructions:

HVX_VectorPair Q6_Wh_vunpack_Vb(HVX_Vector Vu)
HVX_VectorPair Q6_Wuw_vunpack_Vuh(HVX_Vector Vu)
HVX_VectorPair Q6_Ww_vunpack_Vh(HVX_Vector Vu)

HVX_VectorPair Q6_Wh_vunpackoor_WhVb(HVX_VectorPair Vxx, HVX_Vector Vu)

for i = 0 : 128
    Vdd.uh[i] = Vxx.uh[i] | (Vu.ub[i] << 8);

Vxx(uint16_t)	1	1	1	1	1	1	1	1	...	7fff	7fff	7fff	7fff	7fff	7fff	7fff	7fff
Vu(uint8_t)	ff	0	ff	0	ff	0	ff	0	...	ff	0	ff	0	ff	0	ff	0
res_pair(uint16_t)	ff01	1	ff01	1	ff01	1	ff01	1	...	ffff	7fff	ffff	7fff	ffff	7fff	ffff	7fff

Other Instruction:
HVX_VectorPair Q6_Ww_vunpackoor_WwVh(HVX_VectorPair Vxx, HVX_Vector Vu)

Other

HVX_VectorPair Q6_W_vcombine_VV(HVX_Vector Vu, HVX_Vector Vv)

Vu(uint8_t)	128	129	130	131	132	133	134	135	...	248	249	250	251	252	253	254	255
Vv(uint8_t)	0	1	2	3	4	5	6	7	...	120	121	122	123	124	125	126	127
res_pair.lo	0	1	2	3	4	5	6	7	...	120	121	122	123	124	125	126	127
res_pair.hi	128	129	130	131	132	133	134	135	...	248	249	250	251	252	253	254	255

HVX_Vector Q6_V_vdelta_VV(HVX_Vector Vu, HVX_Vector Vv)

//PseudoCode:
for (offset = VWIDTH/2; (offset>>=1) > 0; )
{
    for (k = 0; k < VWIDTH; k++)
    {
        Vd.ub[k] = (Vv.ub[k] & offset) ? Vu.ub[k ^ offset] : Vu.ub[k]; 
    }
    // copy Vd to Vu 
    for (k = 0; k < VWIDTH; k++)
    {
        Vu.ub[k] = Vd.ub[k];
    }
}

HVX_Vector Q6_V_vrdelta_VV(HVX_Vector Vu, HVX_Vector Vv)

//PseudoCode:
for (offset = 1; offset < VWIDTH; offset <<= 1){
    for (k = 0; k < VWIDTH; k++) {
        Vd.ub[k] = (Vv.ub[k]&offset) ? Vu.ub[k ^ offset] : Vu.ub[k];
    }
    for (k = 0; k < VWIDTH; k++) {
        Vu.ub[k] = Vd.ub[k];
    }
}

说明:
"

Vu是输入向量，Vv是控制向量，Vd是输出向量

vdelta和vrdelta的区别是, offset循环的方向不同个, delta是从大到小, rdelta是从小到大

核心代码是: Vd.ub[k] = (Vv.ub[k] & offset) ? Vu.ub[k ^ offset] : Vu.ub[k];

我们这里仅讨论Vu的第一个元素 k=0 的情况, offset的值可能是1/2/4/8/16/32/64
当offset=4时, 0 ^ offset = 4, 4 ^ offest = 0. 此时idx0, 4是互相联系的, 同理(1,5)(2,6)(3,7)(8,12)...也都是互相联系的
此时对应控制向量Vv.ub[0]的第三个bit, 如果是1, 则取Vu.ub[4], 否则取Vu.ub[0]
所以, Vv的每个ub控制对应的Vu的ub的取值, 而每个bit对应不同offset的取值行为

Vv长度为VWIDTH(128), 每个ub中有7个bit可以进行有效控制(最高位128超出了VWIDTH, 无效), 所以, 共有128*2^7=16384种控制行为(存在多个组合导出同样的结果)

delta指令和rdelta指令是互逆的, 前提是没有发生数据的拷贝, 只有数据的交换

下面我们来看一些可能的使用场景(以长度为8做例子来讲解):

1. 向量反转

vector: [0123 4567] control_mask: [0b000, 0b000, 0b000, 0b000, 0b000, 0b000, 0b000, 0b000] // 初期状态
vector: [4567 0123] control_mask: [0b100, 0b100, 0b100, 0b100, 0b100, 0b100, 0b100, 0b100] // offset=4时, 全部元素都启动交换
vector: [6745 2301] control_mask: [0b110, 0b110, 0b110, 0b110, 0b110, 0b110, 0b110, 0b110] // offset=2时, 全部元素都启动交换
vector: [7654 3210] control_mask: [0b111, 0b111, 0b111, 0b111, 0b111, 0b111, 0b111, 0b111] // offset=1时, 全部元素都启动交换
注: 我们这里的offset是从大到小, 所以应该调用delta指令, 当然用rdelta指令也是可以的.
而且, 可以用类似的方法进行局部反转或者全局反转

1. 复制元素

vector: [0123 4567] control_mask: [0b000, 0b000, 0b000, 0b000, 0b000, 0b000, 0b000, 0b000] // 初期状态
vector: [0023 4567] control_mask: [0b000, 0b001, 0b000, 0b000, 0b000, 0b000, 0b000, 0b000] // offset=1时, 第2位元素启动复制
vector: [0000 4567] control_mask: [0b000, 0b001, 0b010, 0b010, 0b000, 0b000, 0b000, 0b000] // offset=2时, 3,4位元素启动复制
vector: [0000 0000] control_mask: [0b000, 0b001, 0b010, 0b010, 0b100, 0b100, 0b100, 0b100] // offset=4时, 5,6,7,8位元素启动复制
注: 我们这里的offest是从小到大, 所以应该调用rdelta指令, 当然用delta指令也是可以的, 只不过顺序要调整一下.
而且, 类似方法可以复制任何一个或任意一段元素

```
int ctrl_reverse[32] __attribute__((aligned(128))) = {
    0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff,
    0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff,
    0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff,
    0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff,
}
Vv = *(HVX_Vector*)ctrl_reverse;
HVX_Vector res = Q6_V_vdelta_VV(Vu, Vv);
```

Vu(uint8_t)	128	129	130	131	132	133	134	135	...	248	249	250	251	252	253	254	255
Vv(uint8_t)	255	255	255	127	255	255	255	127	...	255	255	255	127	255	255	255	127
res(uint8_t)	255	254	253	252	251	250	249	248	...	135	134	133	132	131	130	129	128

```
int ctrl_replicate[32] __attribute__((aligned(128))) = {
    0x00010202, 0x04040404, 0x08080808, 0x08080808, 0x10101010, 0x10101010, 0x10101010, 0x10101010,
    0x20202020, 0x20202020, 0x20202020, 0x20202020, 0x20202020, 0x20202020, 0x20202020, 0x20202020,
    0x40404040, 0x40404040, 0x40404040, 0x40404040, 0x40404040, 0x40404040, 0x40404040, 0x40404040,
    0x40404040, 0x40404040, 0x40404040, 0x40404040, 0x40404040, 0x40404040, 0x40404040, 0x40404040,
}
Vv = *(HVX_Vector*)ctrl_replicate;
HVX_Vector res = Q6_V_vrdelta_VV(Vu, Vv);
```

Vu(uint8_t)	128	129	130	131	132	133	134	135	...	248	249	250	251	252	253	254	255
Vv(uint8_t)	2	2	1	0	4	4	4	4	...	64	64	64	64	64	64	64	64
res(uint8_t)	131	131	131	131	131	131	131	131	...	131	131	131	131	131	131	131	131

HVX_Vector Q6_Vw_vinsert_VwR(HVX_Vector Vu, Word32 Rt)

replace first word into Rt
here, we set Rt to 0x7fffffff

Vu(uint32_t)	0	1	2	3	4	5	6	7	...
res(uint32_t)	2147483647	1	2	3	4	5	6	7	...

HVX_VectorPair Q6_W_vswap_QVV(HVX_VectorPred Qt, HVX_Vector Vu, HVX_Vector Vv)

swap the elements of Vu and Vv according to Qt, put Vv into res_pair.lo, put Vu into res_pair.hi

Vu(uint8_t)	128	129	130	131	132	133	134	135	...
Vv(uint8_t)	0	1	2	3	4	5	6	7	...
Qt(uint8_t)	1	0	1	0	1	0	1	0	...
res_pair.lo	128	1	130	3	132	5	134	7	...
res_pair.hi	0	129	2	131	4	133	6	135	...

HVX_VectorPair Q6_Ww_vasrinto_WwVwVw(HVX_VectorPair Vxx, HVX_Vector Vu, HVX_Vector Vv)

不是非常理解这条指令的用途.

// Pseudocode
for (i = 0; i < VELEM(32); i++) {
    shift  = (Vu.w[i] << 32);
    mask   = (((Vxx.v[0].w[i]) << 32) | Vxx.v[0].w[i]);
    lomask = (((1) << 32) - 1);
    count  = -(0x40 & Vv.w[i]) + (Vv.w[i] & 0x3f);
    result = (count == -0x40) ? 0 : (((count < 0) ? ((shift << -(count)) | (mask & (lomask << - (count)))) : 
                                                    ((shift >>   count)  | (mask & (lomask >>    count)))));
    Vxx.v[1].w[i] = ((result >> 32) & 0xffffffff);
    Vxx.v[0].w[i] = (result & 0xffffffff) ;
}

Vxx.lo(uint32_t)	f0f0f0f0	f0f0f0f0	f0f0f0f0	f0f0f0f0	f0f0f0f0	f0f0f0f0	f0f0f0f0	f0f0f0f0	...
Vu(uint32_t)	ff00ff00	ff00ff00	ff00ff00	ff00ff00	ff00ff00	ff0ff00	ff00ff00	ff00ff00	...
Vv(int32_t)	0	-4	-16	-32	4	16	32	63	...
res_pair.lo(uint32_t)	f0f0f0f0	f0f0f0f0	f0f00000	0	f0f0f0	ff00f0f0	ff00ff00	ffffffff	...
res_pair.hi(uint32_t)	ff00ff00	f00ff000	ff00f0f0	f0f0f0f0	fff00ff0	ff0	ffffffff	ffffffff	...

Logic Operation

compare

HVX VectorPred Q6_Q_vcmp_eq_VbVb(HVX_Vector Vu,HVX_Vector Vv)

Compare Equal
Vd.b[i] = (Vu.b[i] == Vv.b[i]) ? 1 : 0;

Vu(int8_t)	-3	-2	-1	0	1	2	3	4	...
Vv(int8_t)	-2	-1	0	0	1	1	2	3	...
res(int8_t)	0	0	0	1	1	0	0	0	...

HVX VectorPred Q6_Q_vcmp_gt_VbVb(HVX_Vector Vu,HVX_Vector Vv)

Compare Greater Than
Qd.bit[i] = (Vu.b[i] > Vv.b[i])

Vu(int8_t)	-3	-2	-1	0	1	2	3	4	...
Vv(int8_t)	-2	-1	0	0	1	1	2	3	...
res(int8_t)	0	0	0	0	0	1	1	1	...

Other Instruction:

HVX VectorPred Q6_Q_vcmp_eq_VhVh(HVX_Vector Vu,HVX_Vector Vv)
HVX VectorPred Q6_Q_vcmp_eq_VwVw(HVX_Vector Vu,HVX_Vector Vv)
HVX VectorPred Q6_Q_vcmp_gt_VhVh(HVX_Vector Vu,HVX_Vector Vv)
HVX VectorPred Q6_Q_vcmp_gt_VwVw(HVX_Vector Vu,HVX_Vector Vv)
HVX VectorPred Q6_Q_vcmp_gt_VubVub(HVX_VectorVu, HVX_Vector Vv)
HVX VectorPred Q6_Q_vcmp_gt_VuhVuh(HVX_VectorVu, HVX_Vector Vv)
HVX VectorPred Q6_Q_vcmp_gt_VuwVuw(HVX_VectorVu, HVX_Vector Vv)
HVX VectorPred Q6_Q_vcmp_gt_VhfVhf(HVX_VectorVu, HVX_Vector Vv)
HVX VectorPred Q6_Q_vcmp_gt_VsfVsf(HVX_VectorVu, HVX_Vector Vv)

HVX VectorPred Q6_Q_vcmp_eqand_QVbVb(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)

Qd.bit[i] = Qx.bit[i] & (Vu.b[i] == Vv.b[i])

Qx(int8_t)	1	0	1	1	0	1	0	...
Vu(int8_t)	-3	-2	-1	1	2	3	4	...
Vv(int8_t)	-2	-1	0	1	1	2	3	...
res(int8_t)	0	0	0	1	0	0	0	...

HVX VectorPred Q6_Q_vcmp_eqor_QVbVb(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)

Qd.bit[i] = Qx.bit[i] | (Vu.b[i] == Vv.b[i])

Qx(int8_t)	1	0	1	0	1	0	1	0	...
Vu(int8_t)	-3	-2	-1	0	1	2	3	4	...
Vv(int8_t)	-2	-1	0	0	1	1	2	3	...
res(int8_t)	1	0	1	1	1	0	1	0	...

HVX VectorPred Q6_Q_vcmp_eqxacc_QVbVb(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)

Qd.bit[i] = Qx.bit[i] ^ (Vu.b[i] == Vv.b[i])

Qx(int8_t)	1	0	1	0	1	0	1	0	...
Vu(int8_t)	-3	-2	-1	0	1	2	3	4	...
Vv(int8_t)	-2	-1	0	0	1	1	2	3	...
res(int8_t)	1	0	1	1	0	0	1	0	...

Other Instructions:

HVX_VectorPred Q6_Q_vcmp_eqand_QVhVh(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPred Q6_Q_vcmp_eqand_QVwVw(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPred Q6_Q_vcmp_eqor_QVhVh(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPred Q6_Q_vcmp_eqor_QVwVw(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPred Q6_Q_vcmp_eqxacc_QVhVh(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPred Q6_Q_vcmp_eqxacc_QVwVw(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)

HVX VectorPred Q6_Q_vcmp_gtand_QVbVb(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)

Qd.bit[i] = Qx.bit[i] & (Vu.b[i] > Vv.b[i])

Qx(int8_t)	1	0	1	1	0	1	0	...
Vu(int8_t)	-3	-2	-1	1	2	3	4	...
Vv(int8_t)	-2	-1	0	1	1	2	3	...
res(int8_t)	0	0	0	0	0	1	0	...

HVX VectorPred Q6_Q_vcmp_gtor_QVbVb(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)

Qd.bit[i] = Qx.bit[i] | (Vu.b[i] > Vv.b[i])

Qx(int8_t)	1	0	1	1	0	1	0	...
Vu(int8_t)	-3	-2	-1	1	2	3	4	...
Vv(int8_t)	-2	-1	0	1	1	2	3	...
res(int8_t)	1	0	1	1	1	1	1	...

HVX VectorPred Q6_Q_vcmp_gtxacc_QVbVb(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)

Qd.bit[i] = Qx.bit[i] ^ (Vu.b[i] > Vv.b[i])

Qx(int8_t)	1	0	1	1	0	1	0	...
Vu(int8_t)	-3	-2	-1	1	2	3	4	...
Vv(int8_t)	-2	-1	0	1	1	2	3	...
res(int8_t)	1	0	1	1	1	0	1	...

Other Instructions:

HVX_VectorPred Q6_Q_vcmp_gtand_QVhVh(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPred Q6_Q_vcmp_gtand_QVhfVhf(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPred Q6_Q_vcmp_gtand_QVsfVsf(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPred Q6_Q_vcmp_gtand_QVubVub(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPred Q6_Q_vcmp_gtand_QVuhVuh(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPred Q6_Q_vcmp_gtand_QVuwVuw(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPred Q6_Q_vcmp_gtand_QVwVw(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPred Q6_Q_vcmp_gtor_QVhVh(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPred Q6_Q_vcmp_gtor_QVhfVhf(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPred Q6_Q_vcmp_gtor_QVsfVsf(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPred Q6_Q_vcmp_gtor_QVubVub(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPred Q6_Q_vcmp_gtor_QVuhVuh(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPred Q6_Q_vcmp_gtor_QVuwVuw(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPred Q6_Q_vcmp_gtor_QVwVw(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPred Q6_Q_vcmp_gtxacc_QVbVb(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPred Q6_Q_vcmp_gtxacc_QVhVh(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPred Q6_Q_vcmp_gtxacc_QVhfVhf(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPred Q6_Q_vcmp_gtxacc_QVsfVsf(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPred Q6_Q_vcmp_gtxacc_QVubVub(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPred Q6_Q_vcmp_gtxacc_QVuhVuh(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPred Q6_Q_vcmp_gtxacc_QVuwVuw(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)
HVX_VectorPred Q6_Q_vcmp_gtxacc_QVwVw(HVX_VectorPred Qx,HVX_Vector Vu, HVX_Vector Vv)

Bit Operation

pred & pred

HVX_VectorPred Q6_Q_not_Q(HVX_VectorPred Qs)

!Qs 逻辑非

Qs(int8_t)	0	1	0	1	0	1	0	1	...
res_pred(int8_t)	1	0	1	0	1	0	1	0	...

HVX_VectorPred Q6_Q_and_QQn(HVX_VectorPred Qs,HVX_VectorPred Qt)

Qs & (!Qt) 逻辑与非

Qs(int8_t)	0	1	0	1	0	1	0	1	...
Qt(int8_t)	1	1	0	0	1	1	0	0	...
res_pred(int8_t)	0	0	0	1	0	0	0	1	...

HVX_VectorPred Q6_Q_and_QQ(HVX_VectorPred Qs,HVX_VectorPred Qt)

Qs & Qt 逻辑与

Qs(int8_t)	0	1	0	1	0	1	0	1	...
Qt(int8_t)	1	1	0	0	1	1	0	0	...
res_pred(int8_t)	0	1	0	0	0	1	0	0	...

HVX_VectorPred Q6_Q_or_QQn(HVX_VectorPred Qs,HVX_VectorPred Qt)

Qs | (!Qt) 逻辑或非

Qs(int8_t)	0	1	0	1	0	1	0	1	...
Qt(int8_t)	1	1	0	0	1	1	0	0	...
res_pred(int8_t)	0	1	1	1	0	1	1	1	...

HVX_VectorPred Q6_Q_or_QQ(HVX_VectorPred Qs,HVX_VectorPred Qt)

Qs | Qt 逻辑或

Qs(int8_t)	0	1	0	1	0	1	0	1	...
Qt(int8_t)	1	1	0	0	1	1	0	0	...
res_pred(int8_t)	1	1	0	1	1	1	0	1	...

HVX_VectorPred Q6_Q_xor_QQ(HVX_VectorPred Qs,HVX_VectorPred Qt)

Qs ^ Qt 逻辑异或

Qs(int8_t)	0	1	0	1	0	1	0	1	...
Qt(int8_t)	1	1	0	0	1	1	0	0	...
res_pred(int8_t)	1	0	0	1	1	0	0	1	...

pred & vec

HVX_Vector Q6_V_vand_QnV(HVX_VectorPred Qt, HVX_Vector Vu)

Qt为0时，res_vec.b[i] = Vu.b[i]，否则res_vec.b[i] = 0

Qt(uint8_t)	1	1	0	0	1	1	0	0	...
Vu(uint8_t)	0	1	2	3	4	5	6	7	...
res_vec(uint8_t)	0	0	2	3	0	0	6	7	...

HVX_Vector Q6_V_vand_QV(HVX_VectorPred Qt, HVX_Vector Vu)

Qt为0时，res_vec.b[i] = 0，否则res_vec.b[i] = Vu.b[i]

Qt(uint8_t)	1	1	0	0	1	1	0	0	...
Vu(uint8_t)	0	1	2	3	4	5	6	7	...
res_vec(uint8_t)	0	1	0	0	4	5	0	0	...

HVX_Vector Q6_V_vmux_QVV(HVX_VectorPred Qt, HVX_Vector Vu, HVX_Vector Vv)

Qt为0时，res_vec.b[i] = Vv.b[i]，否则res_vec.b[i] = Vu.b[i]

Qt(uint8_t)	1	1	0	0	1	1	0	0	...
Vu(uint8_t)	0	1	2	3	4	5	6	7	...
Vv(uint8_t)	128	129	130	131	132	133	134	135	...
res_vec(uint8_t)	0	1	130	131	4	5	134	135	...

HVX_Vector Q6_V_vand_QnR(HVX_VectorPred Qt, Word32 Rt)

Qt为0时，res_vec.b[i] = Rt.b[i & 0x3]，否则res_vec.b[i] = 0

Rt: 0x04030201

Qt(uint8_t)	1	1	0	0	1	1	0	0	...
res_vec(uint8_t)	0	0	3	4	0	0	3	4	...

HVX_Vector Q6_V_vand_QR(HVX_VectorPred Qt, Word32 Rt)

Qt为0时，res_vec.b[i] = 0，否则res_vec.b[i] = Rt.b[i & 0x3]

Rt: 0x04030201

Qt(uint8_t)	1	1	0	0	1	1	0	0	...
res_vec(uint8_t)	1	2	0	0	1	2	0	0	...

HVX_Vector Q6_V_vandor_VQnR(HVX_Vector Vu, HVX_VectorPred Qt, Word32 Rt)

Qt为0时，res_vec.b[i] = Vu.b[i] | Rt.b[i & 0x3], 否则res_vec.b[i] = 0

Rt: 0x00f00fff

Qt(uint8_t)	1	1	0	0	1	1	0	0	...
Vu(uint8_t)	0	1	2	3	4	5	6	7	...
res_vec(uint8_t)	0	1	242	3	4	5	246	7	...

HVX_Vector Q6_V_vandor_VQR(HVX_Vector Vu, HVX_VectorPred Qt, Word32 Rt)

Qt为1时，res_vec.b[i] = Vu.b[i] | Rt.b[i & 0x3], 否则res_vec.b[i] = 0

Rt: 0x00f00fff

Qt(uint8_t)	1	1	0	0	1	1	0	0	...
Vu(uint8_t)	0	1	2	3	4	5	6	7	...
res_vec(uint8_t)	255	15	2	3	255	15	6	7	...

HVX_VectorPred Q6_Q_vand_VR(HVX_Vector Vu, Word32 Rt)

Vv.b[i] & Rt.b[i & 0x3] != 0时，res_pred.b[i] = 1，否则res_pred.b[i] = 0

Rt = Rt = 0xff0ff000

Vu(uint8_t)	ff	f	f0	0	0	f	f0	ff	...
res_pred(uint8_t)	0	0	0	0	0	0	0	1	...

HVX_VectorPred Q6_Q_vandor_QVR(HVX_VectorPred Qx, HVX_Vector Vu, Word32 Rt)

Vv.b[i] & Rt.b[i & 0x3] != 0时或者对应Qx不为0，res_pred.b[i] = 1，否则res_pred.b[i] = 0

Rt = Rt = 0xff0ff000

Qt(uint8_t)	1	0	1	0	1	0	1	0	...
Vu(uint8_t)	ff	f	f0	0	0	f	f0	ff	...
res_pred(uint8_t)	1	0	1	0	1	0	1	1	...

vec & vec -> vec

HVX_Vector Q6_V_vand_VV(HVX_Vector Vu, HVX_Vector Vv)

向量与操作
Vd.b[i] = Vu.b[i] & Vv.b[i]

Vu(uint8_t)	ff	f0	f	0	ff	f0	f	0	...
Vv(uint8_t)	ff	ff	ff	ff	0	0	0	0	...
res_vec(uint8_t)	ff	f0	f	0	0	0	0	0	...

HVX_Vector Q6_V_vnot_V(HVX_Vector Vu)

向量非操作
Vd.b[i] = !Vu.b[i]

Vu(uint8_t)	ff	f0	f	0	ff	f0	f	0	...
res_vec(uint8_t)	0	f	f0	ff	0	f	f0	ff	...

HVX_Vector Q6_V_vor_VV(HVX_Vector Vu, HVX_Vector Vv)

向量或操作
Vd.b[i] = Vu.b[i] | Vv.b[i]

Vu(uint8_t)	ff	f0	f	0	ff	f0	f	0	...
Vv(uint8_t)	ff	ff	ff	ff	0	0	0	0	...
res_vec(uint8_t)	ff	ff	ff	ff	ff	f0	f	0	...

HVX_Vector Q6_V_vxor_VV(HVX_Vector Vu, HVX_Vector Vv)

向量异或操作
Vd.b[i] = Vu.b[i] ^ Vv.b[i]

Vu(uint8_t)	ff	f0	f	0	ff	f0	f	0	...
Vv(uint8_t)	ff	ff	ff	ff	0	0	0	0	...
res_vec(uint8_t)	0	f	f0	ff	ff	f0	f	0	...

HVX_Vector Q6_Vh_vadd_vclb_VhVh(HVX_Vector Vu, HVX_Vector Vv)

clb: Count Leading Bit (0 or 1)
Vd.h[i] = Vv.h[i] + clb(Vu.h[i])

Vu(int16_t)	7fff	7fff	ffff	fff	f000	ff00	0	fffe	...
Vv(int16_t)	-32768	32767	-3	-2	-1	0	1	2	...
res_vec(int16_t)	-32767	-32768	13	2	3	8	17	17	...

HVX_Vector Q6_Vh_vnormamt_Vh(HVX_Vector Vu)

clb(Vu.h[i]) - 1

Vu(int16_t)	7fff	7fff	ffff	fff	f000	ff00	0	fffe	...
res_vec(int16_t)	0	0	15	3	3	7	15	14	...

HVX_Vector Q6_Vh_vpopcount_Vh(HVX_Vector Vu)

count number of bit one

Vu(int16_t)	7fff	7fff	ffff	fff	f000	ff00	0	fffe	...
res_vec(int16_t)	15	15	16	12	4	8	0	15	...

HVX_Vector Q6_Vuh_vcl0_Vuh(HVX_Vector Vu)

cl0: Count Leading Zero

Vu(uint16_t)	7fff	7fff	ffff	fff	f000	ff00	0	fffe	...
res_vec(uint16_t)	1	1	0	4	0	0	16	0	...

Other Instructions:

HVX_Vector Q6_Vuw_vcl0_Vuw(HVX_Vector Vu)
HVX_Vector Q6_Vw_vadd_vclb_VwVw(HVX_Vector Vu, HVX_Vector Vv)
HVX_Vector Q6_Vw_vnormamt_Vw(HVX_Vector Vu)

HVX_Vector Q6_V_equals_V(HVX_Vector Vu)

向量拷贝

Vu(int8_t)	0	1	2	3	4	5	6	7	...
res(int8_t)	0	1	2	3	4	5	6	7	...

**Other Instructions: **

HVX_Vector Q6_Vhf_equals_Vqf16(HVX_Vector Vu)
HVX_Vector Q6_Vhf_equals_Wqf32(HVX_VectorPair Vuu)
HVX_Vector Q6_Vsf_equals_Vqf32(HVX_Vector Vu)

Vector Transfer

data type upgrade

HVX_VectorPair Q6_Wh_vsxt_Vb(HVX_Vector Vu)

sign extend: 在高位补符号位

Vu(int8_t)	-3	-2	-1	0	1	2	3	4	...
res_pair.lo(int16_t)	-3	-1	1	3	8	10	12	14	...
res_pair.hi(int16_t)	-2	0	2	4	9	11	13	15	...

HVX_VectorPair Q6_Wuh_vzxt_Vub(HVX_Vector Vu)

zero extend: 在高位补0

Vu(int8_t)	0	1	2	3	127	128	254	255	...
res_pair.lo(int16_t)	0	2	127	254	9	11	13	15	...
res_pair.hi(int16_t)	1	3	128	255	0	0	0	0	...

**Other Instructions: **

HVX_VectorPair Q6_Wuw_vzxt_Vuh(HVX_Vector Vu)
HVX_VectorPair Q6_Ww_vsxt_Vh(HVX_Vector Vu)

shift right and downgrade
右移Rt位, 且数据类型降级;
- [u]h -> [u]b: Rt范围是 0 ~ 7 (内部通过 Rt & 0x7的方式实现)
- [u]w -> [u]h: Rt范围是 0 ~ 15 (内部通过 Rt & 0xf的方式实现) rnd模式: 需要先 + 1 << (Rt - 1), 然后再右移Rt位
在下面的测试中, 设置 Rt = 2.

HVX_Vector Q6_Vb_vasr_VhVhR_sat(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt)

Vu(int16_t)	-6	-8	-9	-1024	0	0	0	0	...
Vv(int16_t)	6	8	9	512	0	0	0	0	...
res(int8_t)	1	-2	2	-2	2	-3	127	-128	...

HVX_Vector Q6_Vb_vasr_VhVhR_rnd_sat(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt)

Vu(int16_t)	-6	-8	-9	-1024	0	0	0	0	...
Vv(int16_t)	6	8	9	512	0	0	0	0	...
res(int8_t)	2	-1	2	-2	2	-2	127	-128	...

HVX_Vector Q6_Vub_vasr_VhVhR_sat(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt)

Vu(int16_t)	-6	-8	-9	-1024	0	0	0	0	...
Vv(int16_t)	6	8	9	512	0	0	0	0	...
res(uint8_t)	1	0	2	0	2	0	128	0	...

HVX_Vector Q6_Vub_vasr_VhVhR_rnd_sat(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt)

Vu(int16_t)	-6	-8	-9	-1024	0	0	0	0	...
Vv(int16_t)	6	8	9	512	0	0	0	0	...
res(uint8_t)	2	0	2	0	2	0	128	0	...

HVX_Vector Q6_Vub_vasr_VuhVuhR_sat(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt)

Vu(uint16_t)	3	4	5	6	0	0	0	0	...
Vv(uint16_t)	1017	1018	1019	1024	0	0	0	0	...
res(uint8_t)	254	0	254	1	254	1	255	1	...

HVX_Vector Q6_Vub_vasr_VuhVuhR_rnd_sat(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt)

Vu(uint16_t)	3	4	5	6	0	0	0	0	...
Vv(uint16_t)	1017	1018	1019	1024	0	0	0	0	...
res(uint8_t)	254	1	255	1	255	1	255	2	...

Other Instructions:

HVX_Vector Q6_Vh_vasr_VwVwR_rnd_sat(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt)
HVX_Vector Q6_Vuh_vasr_VwVwR_rnd_sat(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt)
HVX_Vector Q6_Vuh_vasr_VuwVuwR_rnd_sat(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt)
HVX_Vector Q6_Vh_vasr_VwVwR(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt)
HVX_Vector Q6_Vh_vasr_VwVwR_sat(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt)
HVX_Vector Q6_Vuh_vasr_VuwVuwR_sat(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt)
HVX_Vector Q6_Vuh_vasr_VwVwR_sat(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt)

-round sat and donwgrade

HVX_Vector Q6_Vb_vround_VhVh_sat(HVX_Vector Vu, HVX_Vector Vv)

将Vv.h[i]和Vu.h[i]分别做roudning(先+128, 然后>>8) + sat , 然后交替放在结果向量中

Vu(int16_t)	-383	-384	-640	-641	0	0	0	0	...
Vv(int16_t)	127	128	639	640	0	0	0	0	...
res(int8_t)	0	-1	1	-1	2	-2	3	-3	...

HVX_Vector Q6_Vub_vround_VhVh_sat(HVX_Vector Vu, HVX_Vector Vv)

将Vv.h[i]和Vu.h[i]分别做roudning(先+128, 然后>>8) + sat(负数饱和成0) , 然后交替放在结果向量中

Vu(int16_t)	-383	-384	-640	-641	0	0	0	0	...
Vv(int16_t)	127	128	639	640	0	0	0	0	...
res(uint8_t)	0	0	1	0	2	0	3	0	...

HVX_Vector Q6_Vub_vround_VuhVuh_sat(HVX_Vector Vu, HVX_Vector Vv)

将Vv.uh[i]和Vu.uh[i]分别做roudning(先+128, 然后>>8) + sat , 然后交替放在结果向量中

Vu(uint16_t)	127	128	383	384	0	0	0	0	...
Vv(uint16_t)	65150	65151	65152	65535	0	0	0	0	...
res(uint8_t)	254	0	254	1	255	1	255	2	...

**Other Instructions: **

HVX_Vector Q6_Vh_vround_VwVw_sat(HVX_Vector Vu, HVX_Vector Vv)
HVX_Vector Q6_Vuh_vround_VuwVuw_sat(HVX_Vector Vu, HVX_Vector Vv)
HVX_Vector Q6_Vuh_vround_VwVw_sat(HVX_Vector Vu, HVX_Vector Vv)

sat

HVX_Vector Q6_Vh_vsat_VwVw(HVX_Vector Vu, HVX_Vector Vv)

int32饱和成int16

Vu(int32_t)	32772	32773	32774	32775	32776	32777	32778	32779	...
Vv(int32_t)	32739	32740	32741	32742	32743	32744	32745	32746	...
res(int16_t)	32739	32767	32740	32767	32741	32767	32742	32767	...

HVX_Vector Q6_Vub_vsat_VhVh(HVX_Vector Vu, HVX_Vector Vv)

int16饱和成uint8

Vu(int16_t)	-32768	-129	-128	-1	0	0	0	0	...
Vv(int16_t)	32767	129	128	127	0	0	0	0	...
res(uint8_t)	255	0	129	0	128	0	127	0	...

HVX_Vector Q6_Vuh_vsat_VuwVuw(HVX_Vector Vu, HVX_Vector Vv)

uint32饱和成uint16

Vu(uint32_t)	0	1	2	3	4	5	6	7	...
Vv(uint32_t)	65533	65534	65535	65536	0	0	0	0	...
res(uint16_t)	65533	0	65534	1	65535	2	65535	3	...

HVX_Vector Q6_Vw_vsatdw_VwVw(HVX_Vector Vu, HVX_Vector Vv)

把Vu.w[i]作为高位, Vv.w[i]作为低位, 生成一个int64, 然后进行饱和

Vu(int32_t)	0	0	0	-1	-1	-1	-1	-2	...
Vv(int32_t)	1	2147483647	-2147483648	-1	-2147483648	2147483647	0	0	...
res(int32_t)	1	2147483647	2147483647	-1	-2147483648	-2147483648	-2147483648	-2147483648	...

Scatter & Gather

gather

void Q6_vgather_AQRMWw(HVX_Vector* A, HVX_VectorPred Qs, HVX_Vector* Rb, Word32 Mu, HVX_VectorPair Vvv)

void Q6_vgather_ARMWw(HVX_Vector* A, HVX_Vector* Rb, Word32 Mu, HVX_VectorPair Vvv)

void Q6_vgather_AQRMVh(HVX_Vector* A, HVX_VectorPred Qs, HVX_Vector* Rb, Word32 Mu, HVX_Vector Vv)

void Q6_vgather_AQRMVw(HVX_Vector* A, HVX_VectorPred Qs, HVX_Vector* Rb, Word32 Mu, HVX_Vector Vv)

void Q6_vgather_ARMVh(HVX_Vector* A, HVX_Vector* Rb, Word32 Mu, HVX_Vector Vv)

void Q6_vgather_ARMVw(HVX_Vector* A, HVX_Vector* Rb, Word32 Mu, HVX_Vector Vv)

scatter

void Q6_vscatter_QRMWwV(HVX_VectorPred Qs, HVX_Vector* Rb, Word32 Mu, HVX_VectorPair Vvv, HVX_Vector Vw)

void Q6_vscatteracc_RMWwV(HVX_Vector* Rb, Word32 Mu, HVX_VectorPair Vvv, HVX_Vector Vw)

void Q6_vscatter_RMWwV(HVX_Vector* Rb, Word32 Mu, HVX_VectorPair Vvv, HVX_Vector Vw)

void Q6_vscatter_QRMVhV(HVX_VectorPred Qs, HVX_Vector* Rb, Word32 Mu, HVX_Vector Vv, HVX_Vector Vw)

void Q6_vscatter_QRMVwV(HVX_VectorPred Qs, HVX_Vector* Rb, Word32 Mu, HVX_Vector Vv, HVX_Vector Vw)

void Q6_vscatteracc_RMVhV(HVX_Vector* Rb, Word32 Mu, HVX_Vector Vv, HVX_Vector Vw)

void Q6_vscatter_RMVhV(HVX_Vector* Rb, Word32 Mu, HVX_Vector Vv, HVX_Vector Vw)

void Q6_vscatteracc_RMVwV(HVX_Vector* Rb, Word32 Mu, HVX_Vector Vv, HVX_Vector Vw)

void Q6_vscatter_RMVwV(HVX_Vector* Rb, Word32 Mu, HVX_Vector Vv, HVX_Vector Vw)

Lut

HVX_Vector Q6_Vb_vlut32_VbVbI(HVX_Vector Vu, HVX_Vector Vv, Word32 Iu3)

指令解释:

Vu: idx向量(以ub作为单位), Vv: 数据向量(以Byte作为单位), Iu3: 模式向量(0~7)
我们把 Iu3 写为 ABC, 每个字母代表1个bit, 从高位到低位
把Vu的某个idx(Vu.ub[i]), 写为 abcdefgh, 每个字母代表1个bit, 从高位到低位
向量Vv的长度为128B, 但是打表的时候是以h(2B)为单位来访问的, 然后再选择低位或是高位Byte
idx对应的访问数据为: Vu.uh[cdefgh].ub[b]
这样我们能完全访问向量的128B的每个Byte了, 但是指令为什么叫lut32呢?

因为我们对idx还做了相关的限制, 只有当ABC == abc的时候, 才能正常访问, 否则返回0
Iu3 = 0 (0b000)时, idx只能访问 Vv.uh[0 ~ 31].ub[0]
Iu3 = 1 (0b001)时, idx只能访问 Vv.uh[32 ~ 63].ub[0]
Iu3 = 2 (0b010)时, idx只能访问 Vv.uh[0 ~ 31].ub[1]
Iu3 = 3 (0b011)时, idx只能访问 Vv.uh[32 ~ 63].ub[1]

Vv(uint8_t)	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	...	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127
Vu(uint8_t)	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	...	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127
Iu3 = 0(uint8_t)	2	4	6	8	10	12	14	16	18	20	22	24	26	28	30	32	34	36	38	40	42	44	46	48	50	52	54	56	58	60	62	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	...	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
Iu3 = 1(uint8_t)	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	64	66	68	70	72	74	76	78	80	82	84	86	88	90	92	94	96	98	100	102	104	106	108	110	112	114	116	118	120	122	124	126	...	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
Iu3 = 2(uint8_t)	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	...	1	3	5	7	9	11	13	15	17	19	21	23	25	27	29	31	33	35	37	39	41	43	45	47	49	51	53	55	57	59	61	63	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
Iu3 = 3(uint8_t)	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	...	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	65	67	69	71	73	75	77	79	81	83	85	87	89	91	93	95	97	99	101	103	105	107	109	111	113	115	117	119	121	123	125	127
Iu3 = 4(uint8_t)	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	...	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0

在Iu3 = 4(0b100)的时候, 最高位为1, Vu中的idx的最高位都是0, 不匹配, 所以结果全为0

由于idx的解析并不是向量中对应的Byte的idx, 如果想要得到对应的结果, 需要对Vv进行一次转换
Vv = Q6_Vb_vshuff_Vb(Vv)

Vv(uint8_t)	64	1	65	2	66	3	67	4	68	5	69	6	70	7	71	8	72	9	73	10	74	11	75	12	76	13	77	14	78	15	79	16	80	17	81	18	82	19	83	20	84	21	85	22	86	23	87	24	88	25	89	26	90	27	91	28	92	29	93	30	94	31	95	...	32	96	33	97	34	98	35	99	36	100	37	101	38	102	39	103	40	104	41	105	42	106	43	107	44	108	45	109	46	110	47	111	48	112	49	113	50	114	51	115	52	116	53	117	54	118	55	119	56	120	57	121	58	122	59	123	60	124	61	125	62	126	63	127
Vu(uint8_t)	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	...	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127
Iu3 = 0(uint8_t)	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	...	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
Iu3 = 1(uint8_t)	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	...	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
Iu3 = 2(uint8_t)	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	...	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
Iu3 = 3(uint8_t)	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	...	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127

HVX_Vector Q6_Vb_vlut32_VbVbR(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt)
当Rt在0 ~ 7范围内时, 结果与Q6_Vb_vlut32_VbVbI相同;
当Rt > 7时, 内部实际上做了一个 Rt & 0x7 的操作来剪裁 (见下面结果, Rt = 0 与 Rt = 8 的结果是相同的)

Vv(uint8_t)	64	1	65	2	66	3	67	4	68	5	69	6	70	7	71	8	72	9	73	10	74	11	75	12	76	13	77	14	78	15	79	16	80	17	81	18	82	19	83	20	84	21	85	22	86	23	87	24	88	25	89	26	90	27	91	28	92	29	93	30	94	31	95	...	32	96	33	97	34	98	35	99	36	100	37	101	38	102	39	103	40	104	41	105	42	106	43	107	44	108	45	109	46	110	47	111	48	112	49	113	50	114	51	115	52	116	53	117	54	118	55	119	56	120	57	121	58	122	59	123	60	124	61	125	62	126	63	127
Vu(uint8_t)	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	...	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127
Rt = 0(uint8_t)	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	...	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
Rt = 8(uint8_t)	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	...	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0

HVX_Vector Q6_Vb_vlut32_VbVbR_nomatch(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt)

Vv(uint8_t)	0	64	1	65	2	66	3	67	4	68	5	69	6	70	7	71	8	72	9	73	10	74	11	75	12	76	13	77	14	78	15	79	16	80	17	81	18	82	19	83	20	84	21	85	22	86	23	87	24	88	25	89	26	90	27	91	28	92	29	93	30	94	31	95	...	32	96	33	97	34	98	35	99	36	100	37	101	38	102	39	103	40	104	41	105	42	106	43	107	44	108	45	109	46	110	47	111	48	112	49	113	50	114	51	115	52	116	53	117	54	118	55	119	56	120	57	121	58	122	59	123	60	124	61	125	62	126	63	127
Vu(uint8_t)	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	...	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127
Rt = 0(uint8_t)	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	...	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31
Rt = 1(uint8_t)	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	...	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63
Rt = 2(uint8_t)	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	...	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95
Rt = 3(uint8_t)	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127	...	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127
Rt = 4(uint8_t)	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	...	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31
Rt = 5(uint8_t)	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	...	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63
Rt = 6(uint8_t)	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	...	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95
Rt = 7(uint8_t)	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127	...	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127

HVX_Vector Q6_Vb_vlut32or_VbVbVbI(HVX_Vector Vx, HVX_Vector Vu, HVX_Vector Vv, Word32 Iu3)

由于前面的指令都只是对32个idx范围进行操作, 如果idx的范围比较广, 需要连续取值, 那就需要用到下面的指令了

    HVX_Vector res_or  = Q6_Vb_vlut32_VbVbI(Vu, Vv, 0);
    res_or = Q6_Vb_vlut32or_VbVbVbI(res_or, Vu, Vv, 1);
    res_or = Q6_Vb_vlut32or_VbVbVbI(res_or, Vu, Vv, 2);
    res_or = Q6_Vb_vlut32or_VbVbVbI(res_or, Vu, Vv, 3);

Vv(uint8_t)	0	64	1	65	2	66	3	67	4	68	5	69	6	70	7	71	8	72	9	73	10	74	11	75	12	76	13	77	14	78	15	79	16	80	17	81	18	82	19	83	20	84	21	85	22	86	23	87	24	88	25	89	26	90	27	91	28	92	29	93	30	94	31	95	...	32	96	33	97	34	98	35	99	36	100	37	101	38	102	39	103	40	104	41	105	42	106	43	107	44	108	45	109	46	110	47	111	48	112	49	113	50	114	51	115	52	116	53	117	54	118	55	119	56	120	57	121	58	122	59	123	60	124	61	125	62	126	63	127
Vu(uint8_t)	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	...	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127
res_or(uint8_t)	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63	...	64	65	66	67	68	69	70	71	72	73	74	75	76	77	78	79	80	81	82	83	84	85	86	87	88	89	90	91	92	93	94	95	96	97	98	99	100	101	102	103	104	105	106	107	108	109	110	111	112	113	114	115	116	117	118	119	120	121	122	123	124	125	126	127

HVX_VectorPair Q6_Wh_vlut16_VbVhI(HVX_Vector Vu, HVX_Vector Vv, Word32 Iu3)

上面的vlut32是对ub(1B)进行操作的, 这里的vlut16是对uh(2B)进行操作的, 长度基本是一样的
不同的是, 输出是pair向量, 奇数的idx(Vu.uh[i].ub[0])放在pair的低位, 偶数的idx(Vu.uh[i].ub[1])放在pair的高位
如果 idx = [abcdefgh], Iu3 = [BCD], 输出结果为 ([bcd] == [BCD]) ? Vv.w[defgh].h[c] : 0

Vu(uint8_t)	0	1	1	2	2	3	3	4	4	5	5	6	6	7	7	8	8	9	9	10	10	11	11	12	12	13	13	14	14	15	15	...	48	48	49	49	50	50	51	51	52	52	53	53	54	54	55	55	56	56	57	57	58	58	59	59	60	60	61	61	62	62	63	63
Vv(uint16_t)	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	...	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63
res.lo(uint16_t, Rt = 0)	2	4	6	8	10	12	14	16	18	20	22	24	26	28	30	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	...	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
res.hi(uint16_t, Rt = 0)	2	4	6	8	10	12	14	16	18	20	22	24	26	28	30	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	...	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
res.lo(uint16_t, Rt = 1)	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	32	34	36	38	40	42	44	46	48	50	52	54	56	58	60	62	...	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
res.hi(uint16_t, Rt = 1)	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	32	34	36	38	40	42	44	46	48	50	52	54	56	58	60	62	...	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
res.lo(uint16_t, Rt = 2)	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	...	1	3	5	7	9	11	13	15	17	19	21	23	25	27	29	31	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
res.hi(uint16_t, Rt = 2)	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	...	1	3	5	7	9	11	13	15	17	19	21	23	25	27	29	31	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
res.lo(uint16_t, Rt = 3)	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	...	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	33	35	37	39	41	43	45	47	49	51	53	55	57	59	61	63
res.hi(uint16_t, Rt = 3)	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	...	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	33	35	37	39	41	43	45	47	49	51	53	55	57	59	61	63
res.lo(uint16_t, Rt = 4)	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	...	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
res.hi(uint16_t, Rt = 4)	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	...	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0

当Rt = 4时, 因为idx与Iu3不匹配, 导致所有的结果为0

HVX_VectorPair Q6_Wh_vlut16_VbVhR(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt) 与上面的指令基本相同

HVX_VectorPair Q6_Wh_vlut16_VbVhR_nomatch(HVX_Vector Vu, HVX_Vector Vv, Word32 Rt)

Vu(uint8_t)	0	0	1	1	2	2	3	3	4	4	5	5	6	6	7	7	8	8	9	9	10	10	11	11	12	12	13	13	14	14	15	15	...	48	48	49	49	50	50	51	51	52	52	53	53	54	54	55	55	56	56	57	57	58	58	59	59	60	60	61	61	62	62	63	63
Vv(uint16_t)	0	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	...	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63
res.lo(uint16_t, Rt = 0)	0	2	4	6	8	10	12	14	16	18	20	22	24	26	28	30	0	2	4	6	8	10	12	14	16	18	20	22	24	26	28	30	...	0	2	4	6	8	10	12	14	16	18	20	22	24	26	28	30	0	2	4	6	8	10	12	14	16	18	20	22	24	26	28	30
res.hi(uint16_t, Rt = 0)	0	2	4	6	8	10	12	14	16	18	20	22	24	26	28	30	0	2	4	6	8	10	12	14	16	18	20	22	24	26	28	30	...	0	2	4	6	8	10	12	14	16	18	20	22	24	26	28	30	0	2	4	6	8	10	12	14	16	18	20	22	24	26	28	30
res.lo(uint16_t, Rt = 1)	32	34	36	38	40	42	44	46	48	50	52	54	56	58	60	62	32	34	36	38	40	42	44	46	48	50	52	54	56	58	60	62	...	32	34	36	38	40	42	44	46	48	50	52	54	56	58	60	62	32	34	36	38	40	42	44	46	48	50	52	54	56	58	60	62
res.hi(uint16_t, Rt = 1)	32	34	36	38	40	42	44	46	48	50	52	54	56	58	60	62	32	34	36	38	40	42	44	46	48	50	52	54	56	58	60	62	...	32	34	36	38	40	42	44	46	48	50	52	54	56	58	60	62	32	34	36	38	40	42	44	46	48	50	52	54	56	58	60	62
res.lo(uint16_t, Rt = 2)	1	3	5	7	9	11	13	15	17	19	21	23	25	27	29	31	1	3	5	7	9	11	13	15	17	19	21	23	25	27	29	31	...	1	3	5	7	9	11	13	15	17	19	21	23	25	27	29	31	1	3	5	7	9	11	13	15	17	19	21	23	25	27	29	31
res.hi(uint16_t, Rt = 2)	1	3	5	7	9	11	13	15	17	19	21	23	25	27	29	31	1	3	5	7	9	11	13	15	17	19	21	23	25	27	29	31	...	1	3	5	7	9	11	13	15	17	19	21	23	25	27	29	31	1	3	5	7	9	11	13	15	17	19	21	23	25	27	29	31
res.lo(uint16_t, Rt = 3)	33	35	37	39	41	43	45	47	49	51	53	55	57	59	61	63	33	35	37	39	41	43	45	47	49	51	53	55	57	59	61	63	...	33	35	37	39	41	43	45	47	49	51	53	55	57	59	61	63	33	35	37	39	41	43	45	47	49	51	53	55	57	59	61	63
res.hi(uint16_t, Rt = 3)	33	35	37	39	41	43	45	47	49	51	53	55	57	59	61	63	33	35	37	39	41	43	45	47	49	51	53	55	57	59	61	63	...	33	35	37	39	41	43	45	47	49	51	53	55	57	59	61	63	33	35	37	39	41	43	45	47	49	51	53	55	57	59	61	63
res.lo(uint16_t, Rt = 4)	0	2	4	6	8	10	12	14	16	18	20	22	24	26	28	30	0	2	4	6	8	10	12	14	16	18	20	22	24	26	28	30	...	0	2	4	6	8	10	12	14	16	18	20	22	24	26	28	30	0	2	4	6	8	10	12	14	16	18	20	22	24	26	28	30
res.hi(uint16_t, Rt = 4)	0	2	4	6	8	10	12	14	16	18	20	22	24	26	28	30	0	2	4	6	8	10	12	14	16	18	20	22	24	26	28	30	...	0	2	4	6	8	10	12	14	16	18	20	22	24	26	28	30	0	2	4	6	8	10	12	14	16	18	20	22	24	26	28	30

Rt = 0 与 Rt = 4的结果是一样的

HVX_VectorPair Q6_Wh_vlut16or_WhVbVhI(HVX_VectorPair Vxx, HVX_Vector Vu, HVX_Vector Vv, Word32 Iu3)

    Vv = Q6_Vh_vshuff_Vh(Vv);
    res_pair0 = Q6_Wh_vlut16_VbVhI(Vu, Vv, 0);
    res_pair1 = Q6_Wh_vlut16or_WhVbVhI(res_pair0, Vu, Vv, 1);
    res_pair2 = Q6_Wh_vlut16or_WhVbVhI(res_pair1, Vu, Vv, 2);
    res_pair3 = Q6_Wh_vlut16or_WhVbVhI(res_pair2, Vu, Vv, 3);

Vu(uint8_t)	0	1	1	2	2	3	3	4	4	5	5	6	6	7	7	8	8	9	9	10	10	11	11	12	12	13	13	14	14	15	15	...	48	48	49	49	50	50	51	51	52	52	53	53	54	54	55	55	56	56	57	57	58	58	59	59	60	60	61	61	62	62	63	63
Vv(uint16_t)	32	1	33	2	34	3	35	4	36	5	37	6	38	7	39	8	40	9	41	10	42	11	43	12	44	13	45	14	46	15	47	...	16	48	17	49	18	50	19	51	20	52	21	53	22	54	23	55	24	56	25	57	26	58	27	59	28	60	29	61	30	62	31	63
res_pair0.lo(uint16_t)	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	...	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
res_pair0.hi(uint16_t)	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	...	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
res_pair1.lo(uint16_t)	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	...	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
res_pair1.hi(uint16_t)	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	...	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
res_pair2.lo(uint16_t)	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	...	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
res_pair2.hi(uint16_t)	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	...	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0	0
res_pair3.lo(uint16_t)	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	...	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63
res_pair3.hi(uint16_t)	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25	26	27	28	29	30	31	...	32	33	34	35	36	37	38	39	40	41	42	43	44	45	46	47	48	49	50	51	52	53	54	55	56	57	58	59	60	61	62	63

HVX_VectorPair Q6_Wh_vlut16or_VbVhI(HVX_Vector Vu, HVX_Vector Vv, Word32 Iu3)

same as Q6_Wh_vlut16or_WhVbVhI

Other

store

void Q6_vmem_QnRIV_nt(HVX_VectorPred Qv, HVX_Vector* A, HVX_Vector Vs)

void Q6_vmem_QnRIV(HVX_VectorPred Qv, HVX_Vector* A, HVX_Vector Vs)

void Q6_vmem_QRIV_nt(HVX_VectorPred Qv, HVX_Vector* A, HVX_Vector Vs)

void Q6_vmem_QRIV(HVX_VectorPred Qv, HVX_Vector* A, HVX_Vector Vs)

splat

HVX_Vector Q6_V_vsplat_R(Word32 Rt)
Rt = 0x12345678

Vd.w[i] = Rt;

res(uint8_t)	78	56	34	12	78	56	34	12	...

HVX_Vector Q6_Vb_vsplat_R(Word32 Rt)
Rt = 0x12

Vd.b[i] = Rt & 0xff;

res(uint8_t)	12	12	12	12	12	12	12	12	...

HVX_Vector Q6_Vh_vsplat_R(Word32 Rt)
Rt = 0x1234

Vd.h[i] = Rt & 0xffff;

res(uint8_t)	34	12	34	12	34	12	34	12	...

setq

HVX_VectorPred Q6_Q_vsetq_R(Word32 Rt) (Rt = 0 ~ 127)

前Rt个bit为1，后面的bit为0. (最少0个1, 最多127个1)
Rt = 3

res_pred(uint8_t)	1	1	1	0	0	0	0	0	...	0	0	0	0	0	0	0	0

HVX_VectorPred Q6_Q_vsetq2_R(Word32 Rt) (Rt = 1 ~ 128)

前Rt个bit为1，后面的bit为0 (最少1个1, 最多128个1)
Rt = 3

res_pred(uint8_t)	1	1	1	0	0	0	0	0	...	0	0	0	0	0	0	0	0

prefixsum

Init HVX_VectorPred Qv of all ones.

Qv(uint8_t)	1	1	1	1	1	1	1	1	...	1	1	1	1	1	1	1	1

HVX_Vector Q6_Vb_prefixsum_Q(HVX_VectorPred Qv)

从左到右，每个char的值为前面bit的和.

res(uint8_t)	1	2	3	4	5	6	7	8	...	121	122	123	124	125	126	127	128

HVX_Vector Q6_Vh_prefixsum_Q(HVX_VectorPred Qv)

从左到右，每个short的值为前面bit的和.

res(uint16_t)	2	4	6	8	10	12	14	16	...	114	116	118	120	122	124	126	128

HVX_Vector Q6_Vw_prefixsum_Q(HVX_VectorPred Qv)

从左到右，每个int的值为前面bit的和.

res(uint32_t)	4	8	12	16	20	24	28	32	...	100	104	108	112	116	120	124	128

Word32 Q6_R_vextract_VR(HVX_Vector Vu, Word32 Rs)

从Vu中取出第((Rs&127) / 4)个word.

Vu(uint32_t)	0	1	2	3	4	5	6	7	...	24	25	26	27	28	29	30	31

Q6_R_vextract_VR(Vu, 0) = 0
Q6_R_vextract_VR(Vu, 3) = 0
Q6_R_vextract_VR(Vu, 4) = 1
Q6_R_vextract_VR(Vu, 124) = 31
Q6_R_vextract_VR(Vu, 127) = 31
Q6_R_vextract_VR(Vu, 128) = 0

Name		Name	Last commit message	Last commit date
Latest commit History 3 Commits
README.md		README.md

Folders and files

Latest commit

History

Repository files navigation

HVX INTRINSIC EXAMPLES(V68)

Table

ALU

add

sub

abs

abs diff

sad （Sum Absolute Diffence）

max

min

avg

Condacc

mpy

Mpa

shift

Vector Permutation

valign

rotr

shuffle

deal

pack

unpack

Other

Logic Operation

compare

Bit Operation

Vector Transfer

data type upgrade

sat

Scatter & Gather

gather

scatter

Lut

Other

store

splat

setq

prefixsum

About

Resources

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Contributors

Uh oh!

Packages