1–10 of 30 posts
The B+tree implementation provides significant performance improvements over industry standards for large trees. For some workloads with large trees, we've observed:
- vs Abseil B+tree: 2-5× faster across insert/find/erase operations - vs std::map: 2-5× faster across insert/find/erase operations
High-performance header-only container library for C++23 on x86-64
github.com
It seems the code was written with AI, I hope the author knows what he is doing. Last time I tried to use AI to optimize CPU-heavy C++ code (StackBlur) with SIMD, this failed :/
> History/Motivations This project started as an exploration of using AI agents for software development. Based on experience tuning systems using Abseil's B+tree, I was curious if performance could be improved through SIMD instructions, a customized allocator, and tunable node sizes. Claude proved surprisingly adept at helping implement this quickly, and the resulting B+tree showed compelling performance improvement…
> History/Motivations This project started as an exploration of using AI agents for software development. Based on experience tuning systems using Abseil's B+tree, I was curious if performance could be improved through SIMD instructions, a customized allocator, and tunable node sizes. Claude proved surprisingly adept at helping implement this quickly, and the resulting B+tree showed compelling performance improvement…
Have you tried to do any OpenGL or Vulkan work with it? Very frustrating.
React and HTML, though, pretty awesome.
> History/Motivations This project started as an exploration of using AI agents for software development. Based on experience tuning systems using Abseil's B+tree, I was curious if performance could be improved through SIMD instructions, a customized allocator, and tunable node sizes. Claude proved surprisingly adept at helping implement this quickly, and the resulting B+tree showed compelling performance improvement…
Both Codex/Claude Code are terrible with C++. Not sure why that is, but they just spit out nonsense that creates more work than it helps me. Have you tried to do any OpenGL or Vulkan work with it? Very frustrating. React and HTML, though, pretty awesome.
> History/Motivations This project started as an exploration of using AI agents for software development. Based on experience tuning systems using Abseil's B+tree, I was curious if performance could be improved through SIMD instructions, a customized allocator, and tunable node sizes. Claude proved surprisingly adept at helping implement this quickly, and the resulting B+tree showed compelling performance improvement…
Both Codex/Claude Code are terrible with C++. Not sure why that is, but they just spit out nonsense that creates more work than it helps me. Have you tried to do any OpenGL or Vulkan work with it? Very frustrating. React and HTML, though, pretty awesome.
I suppose part of the problem is that training a model on publicly available C++ isn't going to be great because syntactically broken code gets posted to the web all the time, along with suboptimal solutions. I recall a talk saying that functional languages are better for agents because the code published publicly is formally correct.
> History/Motivations This project started as an exploration of using AI agents for software development. Based on experience tuning systems using Abseil's B+tree, I was curious if performance could be improved through SIMD instructions, a customized allocator, and tunable node sizes. Claude proved surprisingly adept at helping implement this quickly, and the resulting B+tree showed compelling performance improvement…
Both Codex/Claude Code are terrible with C++. Not sure why that is, but they just spit out nonsense that creates more work than it helps me. Have you tried to do any OpenGL or Vulkan work with it? Very frustrating. React and HTML, though, pretty awesome.
It's possible Opus 4.5 and GPT-5.2 are significantly less terrible with C++ than previous models. Those only came out within the past 2 months.
They also have significantly more recent knowledge cut-off dates.
> History/Motivations This project started as an exploration of using AI agents for software development. Based on experience tuning systems using Abseil's B+tree, I was curious if performance could be improved through SIMD instructions, a customized allocator, and tunable node sizes. Claude proved surprisingly adept at helping implement this quickly, and the resulting B+tree showed compelling performance improvement…
Both Codex/Claude Code are terrible with C++. Not sure why that is, but they just spit out nonsense that creates more work than it helps me. Have you tried to do any OpenGL or Vulkan work with it? Very frustrating. React and HTML, though, pretty awesome.
I've started adding this to all of my new conversations and it seems to help:
You are a principal software engineer. I report to you. Do not modify files. Do not write prose. Only provide observations and suggestions so that I can learn from you.
Earlier quoted context omitted.
Both Codex/Claude Code are terrible with C++. Not sure why that is, but they just spit out nonsense that creates more work than it helps me. Have you tried to do any OpenGL or Vulkan work with it? Very frustrating. React and HTML, though, pretty awesome.
I had the same experience. C++ doesn't even compile or I have to tell it all the time "use C++23 features". I tried to learn OpenGL with it. This worked out a bit, since I had to spot the errors :D