Перейти к содержимому
TakoAPI
Назад к навыкам

ecocompute

от hongping-zh

Coding Agents & IDEs
526Число загрузок по данным clawskills.sh
0Просмотры
0 лайковВойдите, чтобы поставить лайк навыку
—Ещё нет оценок
Ваша оценка

Кратко

You are an energy efficiency expert for Large Language Model inference.

Описание

Что делает этот навык

An AI advisor for LLM inference energy efficiency, built on 93+ empirical measurements across three NVIDIA GPU architectures (RTX 5090, RTX 4090D, A800). It detects known energy traps where quantization methods like INT8 or NF4 increase rather than reduce power consumption. Provides cost and carbon estimates for deployment scenarios with dollar-denominated recommendations.

It catches counter-intuitive quantization paradoxes, like INT8 default mode using up to 147% more energy than FP16, that most developers only discover after receiving a large electricity bill.

Когда его использовать

  • Checking whether INT8 quantization is inflating your GPU electricity bill
  • Estimating monthly cost and CO2 output before launching a 7B model in production
  • Comparing FP16 vs NF4 energy use to pick a config before committing
  • Diagnosing unexpectedly high power draw on an inference server
  • Auditing bitsandbytes model-loading code for hidden energy waste

Пример сценария

ВВОД

User asks: I'm running Mistral-7B with load_in_8bit=True on an RTX 4090D and my energy usage seems really high

АГЕНТ
  1. 1

    Identifies the INT8 default quantization pattern in the user's config

  2. 2

    Matches config against known INT8 Energy Paradox in the measurement dataset

  3. 3

    Calculates energy overhead vs FP16 baseline (17-147% increase range)

  4. 4

    Provides one-line BitsAndBytesConfig fix to eliminate mixed-precision decomposition overhead

  5. 5

    Outputs before/after cost table with monthly savings and CO2 equivalent

ВЫВОД

Alert showing $188/month in avoidable costs, corrected config snippet, and an energy comparison table for FP16 vs Pure INT8 vs INT8 default

Установка

clawhub install hongping-zh/ecocompute

Выполните команду CLI или вставьте промпт в Claude Code, Cursor, Codex — любой кодинг-агент, которым пользуетесь.

Ссылки