Floating-point exponential accelerator
The FEXPA instruction accelerates the polynomial series calculation of the exp(x) function.
The double-precision variant copies the low 52 bits of an entry from a hard-wired table of 64-bit coefficients, indexed by the low 6 bits of each element of the source vector, and prepends to that the next 11 bits of the source element (src<16:6>), setting the sign bit to zero.
The single-precision variant copies the low 23 bits of an entry from hard-wired table of 32-bit coefficients, indexed by the low 6 bits of each element of the source vector, and prepends to that the next 8 bits of the source element (src<13:6>), setting the sign bit to zero.
The half-precision variant copies the low 10 bits of an entry from hard-wired table of 16-bit coefficients, indexed by the low 5 bits of each element of the source vector, and prepends to that the next 5 bits of the source element (src<9:5>), setting the sign bit to zero.
A coefficient table entry with index m holds the floating-point value 2(m/64), or for the half-precision variant 2(m/32). This instruction is unpredicated.
This instruction is illegal when executed in Streaming SVE mode, unless FEAT_SME_FA64 is implemented and enabled.
31 | 30 | 29 | 28 | 27 | 26 | 25 | 24 | 23 | 22 | 21 | 20 | 19 | 18 | 17 | 16 | 15 | 14 | 13 | 12 | 11 | 10 | 9 | 8 | 7 | 6 | 5 | 4 | 3 | 2 | 1 | 0 |
0 | 0 | 0 | 0 | 0 | 1 | 0 | 0 | size | 1 | 0 | 0 | 0 | 0 | 0 | 1 | 0 | 1 | 1 | 1 | 0 | Zn | Zd |
if !IsFeatureImplemented(FEAT_SVE) then UNDEFINED; if size == '00' then UNDEFINED; constant integer esize = 8 << UInt(size); constant integer n = UInt(Zn); constant integer d = UInt(Zd);
<Zd> |
Is the name of the destination scalable vector register, encoded in the "Zd" field. |
<T> |
Is the size specifier,
encoded in
|
<Zn> |
Is the name of the source scalable vector register, encoded in the "Zn" field. |
CheckNonStreamingSVEEnabled(); constant integer VL = CurrentVL; constant integer PL = VL DIV 8; constant integer elements = VL DIV esize; constant bits(VL) operand = Z[n, VL]; bits(VL) result; for e = 0 to elements-1 constant bits(esize) element = Elem[operand, e, esize]; Elem[result, e, esize] = FPExpA(element); Z[d, VL] = result;
Internal version only: aarchmrs v2024-03_relA, pseudocode v2024-03_rel, sve v2024-03_rel ; Build timestamp: 2024-03-26T09:45
Copyright © 2010-2024 Arm Limited or its affiliates. All rights reserved. This document is Non-Confidential.