eelgaev /
power-cpu-llm
POWER9 CPU inference engine in Rust for Gemma 4 and DiffusionGemma GGUF models, with an OpenAI-compatible server
A new way for AI to search for data
1GitHub stars
What XERJ indexed
Selected work
“Faster Qwen prefill: per-layer DeltaNet scan, hand-written f32 dot tile”. 6 files, 557 lines in src/kernels, src/engine, src/cli.
Licence
The repository
LGPL-2.1
Checked by a person at the pinned commit.
The XERJ corpus
Approach only
The author agreed by email on 2026-10-23. Agents read it for how it works, not to copy it. Answers name the repo, the commit and the file:line.
in licence reviewnot in the hub yetsuggested use: approach only
Example questions
- What does src/kernels do in power-cpu-llm?
- What changed in power-cpu-llm for “Faster Qwen prefill: per-layer DeltaNet scan”?
The prompt
Paste it into your AI agent. It clones eelgaev/power-cpu-llm at the pinned commit and indexes it on your machine.
Search the GitHub repository eelgaev/power-cpu-llm with XERJ (https://hub.xerj.org/corpus/eelgaev--power-cpu-llm). Run these in a shell, in order, and check each result before the next: 1. xerj --help (success: it prints usage). If xerj is missing: curl -fsSL https://xerj.org/get | sh 2. git clone https://github.com/eelgaev/power-cpu-llm power-cpu-llm && git -C power-cpu-llm checkout 0b2fe9bdac6ea6b9b0bc96f6f9e0d1c20c45c13f 3. Start a local node and leave it running: xerj --insecure --data-dir ./xerj-data 4. xerj autoindex ./power-cpu-llm --url http://127.0.0.1:9200 --prefix power-cpu-llm --progress plain 5. xerj autoindex map --url http://127.0.0.1:9200 Then answer my question from the indexed files and cite the ax_path of every hit. Query syntax: https://xerj.org/llms.txt Licence: LGPL-2.1. Read it for the approach only, do not copy code.
Related
Eli ElgaevAI, prompts and agents corpora
Selected work and consent: the XERJ registry.