Skip to main content

Decimal128 by Miguel

Home
Publications
Benchmarks
Business Case
Q&A
GitHub ↗

Q&A

IEEE 754-2019 compliant decimal128 high-performance software solution created by Miguel.

Frequently asked questions about the multiplatform decimal128 project.

Search the knowledge base

Table of Contents

General / Overview

What is decimal128 and why would I want it?
Why not just use binary floating-point (double)?
How many languages are supported by this decimal128 implementation?
Is this a binding over a single C library?
Which platforms does it run on?
Who is Miguel?

Correctness & Conformance

Is the arithmetic correctly rounded?
Which test suites does it pass?
How many test vectors are there?
How do you know all the ports agree with each other?
How are NaN, infinity, and signed zero handled?

Performance

Why is performance better?
How is performance measured?
How does it compare to Java BigDecimal or Python's decimal?
Do you need decimal hardware to be fast?

Language Support & Ports

Which languages have a native core versus a wrapper?
How is Python supported?
Why is Swift the reference implementation?
Does using it pull in a native/C dependency?

Architecture & Design

What are the primitive, core, and wrapper layers?
Why native ports instead of one shared C engine?
What is UBD and why a custom in-memory format?
How do you keep the ports byte-for-byte identical?

Using the Library

What is the default rounding mode, and can I change it?
Is a decimal128 value thread-safe? Are decimal128 operations thread-safe?

Finance Layer

What does the finance layer provide?
Why does the finance layer live in Core rather than the wrapper?

Project & Roadmap

How is it licensed?
What is planned next?

Technical FAQs

How do you implement decimal on a binary computer?
What are the key performance constraints?
Why is proper rounding important?
Isn't controlled rounding of arithmetic operations computationally expensive?
Why can't I perform operations, then round the result?
Why does Double Rounding fail to produce correct results?

© 2026 Decimal128 Miguel's version. Content is provided for information only.