Read BitCal like a redesign review, not like a landing page.

Whitepaper reading contract ​

This site is for readers who want to answer six concrete questions before trusting a low-level library:

1. What is the public model? BitCal is converging on bit_block<Bits>, bit_view, const_bit_view, and free algorithms.
2. What remains stable? <bitcal/bitcal.hpp> is the only stable include seam.
3. Where does implementation freedom begin? Dispatch and kernels live below the public contract.
4. How much performance is actually proven? Only the retained baseline and methodology are claimed.
5. What can I treat as reference material? Contract-oriented pages explain roles and semantics, not every internal header.
6. Which external work informs the design? Research pages cite the instruction references, papers, and adjacent systems that shape the posture.

Public model at a glance ​

Owning role

bit_block<Bits>

Borrowing role

bit_view / const_bit_view

Behavioral center

Free algorithms

Implementation boundary

Dispatch + kernel layer

Not part of the stable user identity.

The home page does not attempt to replace the deeper sections. It gives a map:

• the Guide tells you how to read the project now;
• the Whitepaper states the architecture thesis;
• the Performance section separates baseline from method;
• the Reference pages define the contract surface;
• the Research pages show nearby systems and source material;
• the Status page defines the release and support boundary.

Evidence posture ​

BitCal prefers a narrower but more defensible story:

• C++23 baseline is a design decision, not a future aspiration.
• x86-64-first posture means optimization and benchmark interpretation are concentrated where the project retains active evidence.
• Header-only delivery does not remove the need for explicit verification, benchmark discipline, or support limits.
• Whitepaper-first documentation means architecture, methodology, and references stay connected instead of being split between marketing and source comments.

Research footing ​

1. Intel Intrinsics Guide

   Instruction reference

   Grounds x86 SIMD discussion in an authoritative instruction-level source.

2. Agner Fog optimization manuals

   Microarchitecture manuals

   Useful when explaining why latency, throughput, and dispatch policy matter to low-level claims.

3. Sebastiano Vigna, Broadword Implementation of Rank/Select Queries

   Broadword paper

   Provides a conceptual bridge between bit-level semantics and word-parallel implementation techniques.