Non-Confidential | ![]() | ARM DUI0472J | ||
| ||||
Home > ARMv6 SIMD Instruction Intrinsics > __smuadx intrinsic |
This intrinsic inserts an SMUADX
instruction into the instruction stream generated by the compiler.
It enables you to exchange the halfwords of the second operand, perform two 16-bit signed integer multiplications, and add the products together. Exchanging the halfwords of the second operand produces top × bottom and bottom × top multiplication. The Q flag is set if the addition overflows. The multiplications cannot overflow.
unsigned int __smuadx(unsigned int
val1
,
unsigned int val2
)
Where:
val1
holds the first halfword operands for each multiplication
val2
holds the second halfword operands for each multiplication.
The __smuadx
intrinsic returns the products
of the two 16-bit signed multiplications.
unsigned int exchange_dual_multiply_prods(unsigned int val1, unsigned int val2) { unsigned int res; res = __smuadx(val1,val2); /* val2[31:16][15:0] = val2[15:0][31:16] p1 = val1[15:0] × val2[15:0] p2 = val1[31:16] × val2[31:16] res[31:0] = p1 + p2 */ return res; }