faddv
SVE Floating-Point Add Reduction
FADDV <Vd>, <Pg>, <Zn>.<T>
Sums all active floating-point elements into a scalar result.
Details
SVE floating-point addition reduction that sums all active floating-point elements in the source vector under predicate control and writes the scalar result to the destination. The destination is a floating-point scalar register sized according to the element type. Elements where the predicate is false are excluded from the sum. Reductions are performed with strict left-to-right ordering for reproducibility. NZCV flags are not affected.
Pseudocode Operation
result ← 0.0
for i = 0 to VL/element_size-1:
if Pg[i] then
result ← result + Zn[i]
Vd ← result
Example
FADDV v0.4s, p0/m, z1.s.T
Encoding
Binary Layout
01100101
size
000
00
0
001
Pg
Zn
Vd
Operands
-
Vd
Dest Scalar -
Pg
Mask -
Zn
Vector
Reference (Arm A64 ISA)
Instruction Forms
| Encoding | Instruction | ISA | Bit pattern | ||
|---|---|---|---|---|---|
| 0x65002000 | FADDV <V><d>, <Pg>, <Zn>.<T> | A64 | 01100101 | size | 000 | 00 | 0 | 001 | Pg | Zn | Vd |
Description
Floating-point add horizontally over all lanes of a vector using a recursive pairwise reduction, and place the result in the SIMD&FP scalar destination register. Inactive elements in the source vector are treated as +0.0.
Operation
CheckSVEEnabled();
constant integer VL = CurrentVL;
constant integer PL = VL DIV 8;
bits(PL) mask = P[g, PL];
bits(VL) operand = if AnyActiveElement(mask, esize) then Z[n, VL] else Zeros(VL);
bits(esize) identity = FPZero('0', esize);
V[d, esize] = FPReducePredicated(ReduceOp_FADD, operand, mask, identity, FPCR);