Blueprint Overview

Source: docs/blueprint/README.md in the main repository.

Low-Latency C++ Stack Blueprint

This document family is an architecture audit and expansion plan for a cutting-edge low-latency C++26 systems repository. It maps industry expectations onto what this project ships today, what is documented, and what remains intentionally out of scope (or next-module work).

Layer index

#	Layer	Guide	Code modules	Tests / examples
1	Hardware & OS tuning	01-hardware-os.md	ll/affinity.hpp, hwloc	[ll][affinity], [industry][hwloc]
2	Network ingress & encoding	02-network-ingress.md	Asio/Beast, ll/sbe_style, FlatBuffers	industry suite, example_sbe_style
3	Memory & cache locality	03-memory.md	ll/arena, ll/pmr_arena, ll/cache_line, mimalloc	[industry][pmr], example_pmr
4	Concurrency & lock-free	04-concurrency.md	ll/spsc_queue, Boost.Lockfree, moodycamel	industry queues example + stress tests
5	Compiler & language	05-compiler.md	ll/branch.hpp	[ll][branch]
6	Benchmarking & telemetry	06-telemetry.md	ll/tsc_clock, ll/hdr_histogram, Google Bench	example_hdr, bench_queues
—	Industry library map	07-industry-libraries.md	CMake STACK_WITH_*	test_industry_stack

Master audit: AUDIT.md · Narrative: LOW_LATENCY_STACK.md ·
Hands-on tutorial: ../tutorials/industry-stack.md

Where this repository focuses hardest

flowchart LR
  subgraph Focus["Primary focus today"]
    L3[Memory / pmr / arenas / cache lines]
    L4[SPSC · Boost LF · moodycamel · atomics]
    L2a[Library mesh · Asio Beast TBB Taskflow stdexec]
    L6[HDR percentiles · TSC · Google Benchmark]
  end
  subgraph Secondary["Documented / optional"]
    L1[hwloc · OS isolation · NUMA]
    L5[constexpr · CRTP · flags]
    L2s[SBE-style · FlatBuffers]
  end
  subgraph Future["Referenced, not productized"]
    L2b[DPDK · Onload · HdrHistogram_c deep]
  end
  Focus --> Secondary --> Future

Loading

Short answer to “which layer is the center of gravity?”
Concurrency + memory architecture + the C++26 library mesh, now wired to industry standards (Boost.Lockfree, moodycamel, std::pmr, hwloc, FlatBuffers, Google Benchmark, mimalloc) — with kernel bypass kept as a documented envelope.

Roadmap (SBE · numa/uring · HdrHistogram_c · bypass lab): complete — see AUDIT.md and docs/tutorials/.

Related public work

Repository	Relationship
This repo (cpp26-systems-stack)	Ecosystem + portable ll::* + industry optional libs
hft-asm-l2-conflator	End-to-end HFT-style conflator (AArch64 kernels)