FLOGNP1A
Operation
Approximates natural log of one plus the source, with minus one as the divide-by-zero boundary.
Assembler Syntax
FLOGNP1A.{S|D}(z) Fn(s), Fn(d)
Privilege
Unprivileged
Required CPUID flags
FP
FPTRANSA
Repeat eligibility
REP eligible
After DAZ, computes an implementation-bounded S- or D-format approximation of the \(\ln(1 + x)\) reference without an intermediate rounded addition under the selected FPTRANSA accuracy contract. The reference domain includes finite \(x \ge -1\) and positive infinity; \(-1\) is the DZ boundary.
The required FPTRANSA_ACCURACY contract is 0x0042.
Special values
-
Source equal to -1: DZ is generated and the result is -infinity.
-
Source below -1: NV is generated and the result is the architectural default quiet NaN.
-
Positive infinity: the result is +infinity.
-
Zero source: the result is the source.
Exact anchors
-
\(\operatorname{log1p}(+0) = +0\)
-
\(\operatorname{log1p}(-0) = -0\)
-
\(\operatorname{log1p}(-1) = -\infty\)
Required properties
- strictly increasing on \((-1, +\infty)\)
2.75in
Encodings
FLOGNP1A.{S|D}(z) Fn(s), Fn(d)
Format: Instruction format for FLOGNP1A.{S|D}(z) Fn(s), Fn(d)