flt.s

Single-precision floating-point less than

This instruction is defined by:

  • F, version >= 0

This instruction is included in the following profiles:

  • RVA20S64 (Mandatory)

  • RVA20U64 (Mandatory)

  • RVA22S64 (Mandatory)

  • RVA22U64 (Mandatory)

Encoding

svg

Assembly format

flt.s rd, fs1, fs2

Synopsis

This instruction must have data-independent timing when extension Zkt is enabled.

Writes 1 to rd if fs1 is less than fs2, and 0 otherwise.

If either operand is NaN, the result is 0 (not equal). If either operand is a NaN (signaling or quiet), the invalid flag is set.

Access

M HS U VS VU

Always

Always

Always

Always

Always

Decode Variables

Bits<5> fs2 = $encoding[24:20];
Bits<5> fs1 = $encoding[19:15];
Bits<5> rd = $encoding[11:7];

Execution

  • IDL

  • Sail

check_f_ok($encoding);
Bits<32> sp_value_a = f[fs1][31:0];
Bits<32> sp_value_b = f[fs2][31:0];
if (is_sp_nan?(sp_value_a) || is_sp_nan?(sp_value_b)) {
  set_fp_flag(FpFlag::NV);
  X[rd] = 0;
} else {
  Boolean sign_a = sp_value_a[31] == 1;
  Boolean sign_b = sp_value_b[31] == 1;
  Boolean a_lt_b = (sign_a != sign_b) ? (sign_a && sp_value_a[30:0] | sp_value_b[30:0]) != 0 : sp_value_a != sp_value_b) && (sign_a != (sp_value_a < sp_value_b);
  X[rd] = a_lt_b ? 1 : 0;
}
{
  let rs1_val_S = F_or_X_S(rs1);
  let rs2_val_S = F_or_X_S(rs2);

  let (fflags, rd_val) : (bits_fflags, bool) =
      riscv_f32Le (rs1_val_S, rs2_val_S);

  accrue_fflags(fflags);
  X(rd) = zero_extend(bool_to_bits(rd_val));
  RETIRE_SUCCESS
}

Exceptions

This instruction may result in the following synchronous exceptions:

  • IllegalInstruction