c++

The article critiques memory safety absolutism in programming language debates, arguing that new approaches like Fil-C have trade-offs and that dismissing Rust as unsafe ignores practical benefits.

Fil-C is a fully memory-safe implementation of C/C++. It bundles pointer values with boundary information using invisicaps at the LLVM IR stage, achieving high compatibility with a performance penalty of about 4x.

A guide to implementing a fast lock-free queue in modern C++, covering techniques for concurrent data structures without locks.

This paper presents a systems study comparing C++ and PyTorch inference runtimes for reward model scoring in RLHF pipelines, finding that ONNXRuntime provides speedups on CPU while torch.compile leads on GPU, with batching strategy mattering more than language or runtime.

Gecode 6.3.0 and 6.4.0 are released after a long pause, bringing a modernized build system (CMake) and updated MiniZinc integration.

C++20 introduces a new syntax for range-based for loops that allows declaring an initializer in the loop scope, matching the convenience of languages like Python and Lua.

A pull request adds an initial ET backend to llama.cpp, expanding hardware support for LLM inference.

A detailed overview of all C++20 core language features with examples, serving as a cheat sheet.

Article comparing old C++ performance tricks with modern compiler capabilities, showing that compilers can now optimize naive code better than hand-tuned hacks. Includes benchmarks on AMD Zen 5 with Clang 21.

BUSY is a lean, statically typed, cross-platform build system for GCC, Clang, and MSVC toolchains, designed for easy bootstrapping and minimal dependencies.

OpenAI engineers detail the diagnosis of seemingly impossible crashes in Rockset's C++ data infrastructure, revealing both a silent hardware corruption bug on Azure and an 18-year-old race condition in GNU libunwind, resolved through epidemiological analysis of crash data.

This blog post explores an optimization for LLVM's SmallVector::push_back by tail-calling the grow-and-push path, which eliminates callee-saved register spills and improves the fast path performance.

A user shares their success using the MiMo V2.5 quantized model (MiMo-V2.5-GGUF) with llama.cpp to generate and execute a C++ program that calculates compound interest, demonstrating effective model-assisted coding.

The article explores constructing the slowest possible data access pattern for summing integers in an array by exploiting CPU cache behavior, demonstrating that a carefully crafted pattern can be over 30% slower than randomized access.

audio.cpp is a C++/ggml runtime that integrates 12 audio models including Qwen3-TTS, PocketTTS, and VeVo2, achieving TTS up to 5x faster than Python on CUDA.

This draft book chapter provides an infographic and detailed analysis of operation costs in CPU clock cycles for modern C++ CPUs, covering multiplication, division, and RTTI with latency tables for various architectures.

A developer built Luz, a C++20 path tracer from scratch with zero dependencies, featuring Monte Carlo path tracing, global illumination, BVH acceleration, and a Blender-to-Luz exporter.

The article discusses the pros and cons of using C-style void* pointers versus uint8_t* and std::span for passing memory blobs in C++, arguing for the simplicity and readability of void*.

A detailed catalogue of C++ standard library features that have been formally deprecated, informally discouraged, or are effectively broken but cannot be fixed due to ABI constraints, spanning from C++11 to C++26. The article argues a consistent pattern of the C++ committee shipping replacements for its own features, including a benchmark showing 58x P99 latency difference between Rust and C++ standard library containers.

The article discusses a C++/WinRT pattern for caching the result of a Windows Runtime IAsyncOperation, including handling failures, so that multiple coroutines can share the cached result or exception.