Multi-vector BFloat16 floating-point clamp to minimum/maximum number

Clamp each BFloat16 element in the two or four destination vectors to between the BFloat16 minimum value in the corresponding element of the first source vector and the BFloat16 maximum value in the corresponding element of the second source vector and destructively place the clamped results in the corresponding elements of the two or four destination vectors.

Regardless of the value of FPCR.AH, the behavior is as follows for each minimum number and maximum number operation:

- Negative zero compares less than positive zero.
- If one value is numeric and the other is a quiet NaN, the result is the numeric value.
- When FPCR.DN is 0, if either value is a signaling NaN or if both values are NaNs, the result is a quiet NaN.
- When FPCR.DN is 1, if either value is a signaling NaN or if both values are NaNs, the result is Default NaN.

This instruction follows SME2 non-widening BFloat16 numerical behaviors corresponding to instructions that place their results in two or four SVE Z vectors.

This instruction is unpredicated.

ID_AA64SMFR0_EL1.B16B16 indicates whether this instruction is implemented.

It has encodings from 2 classes: Two registers and Four registers

(FEAT_SVE_B16B16)

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 |

1 | 1 | 0 | 0 | 0 | 0 | 0 | 1 | 0 | 0 | 1 | Zm | 1 | 1 | 0 | 0 | 0 | 0 | Zn | Zd | 0 | |||||||||||

size<1> | size<0> |

if !IsFeatureImplemented(FEAT_SME2) || !IsFeatureImplemented(FEAT_SVE_B16B16) then UNDEFINED; constant integer n = UInt(Zn); constant integer m = UInt(Zm); constant integer d = UInt(Zd:'0'); constant integer nreg = 2;

(FEAT_SVE_B16B16)

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 |

1 | 1 | 0 | 0 | 0 | 0 | 0 | 1 | 0 | 0 | 1 | Zm | 1 | 1 | 0 | 0 | 1 | 0 | Zn | Zd | 0 | 0 | ||||||||||

size<1> | size<0> |

if !IsFeatureImplemented(FEAT_SME2) || !IsFeatureImplemented(FEAT_SVE_B16B16) then UNDEFINED; constant integer n = UInt(Zn); constant integer m = UInt(Zm); constant integer d = UInt(Zd:'00'); constant integer nreg = 4;

<Zd4> |
Is the name of the fourth scalable vector register of the destination multi-vector group, encoded as "Zd" times 4 plus 3. |

<Zd2> |
Is the name of the second scalable vector register of the destination multi-vector group, encoded as "Zd" times 2 plus 1. |

<Zn> |
Is the name of the first source scalable vector register, encoded in the "Zn" field. |

<Zm> |
Is the name of the second source scalable vector register, encoded in the "Zm" field. |

CheckStreamingSVEEnabled(); constant integer VL = CurrentVL; constant integer elements = VL DIV 16; array [0..3] of bits(VL) results; for r = 0 to nreg-1 constant bits(VL) operand1 = Z[n, VL]; constant bits(VL) operand2 = Z[m, VL]; constant bits(VL) operand3 = Z[d+r, VL]; for e = 0 to elements-1 constant bits(16) element1 = Elem[operand1, e, 16]; constant bits(16) element2 = Elem[operand2, e, 16]; constant bits(16) element3 = Elem[operand3, e, 16]; Elem[results[r], e, 16] = BFMinNum(BFMaxNum(element1, element3, FPCR), element2, FPCR); for r = 0 to nreg-1 Z[d+r, VL] = results[r];

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.