# RahulSChand/gpu_poor

Calculate token/s & GPU memory requirement for any LLM.  Supports llama.cpp/ggml/bnb/QLoRA quantization

Repository: https://github.com/RahulSChand/gpu_poor
Canonical: https://ross.abutalabs.com/products/gpu_poor
Homepage: https://rahulschand.github.io/gpu_poor/
Language: JavaScript
License Family: other
Topics: ggml, huggingface, llm, quantization, gpu, language-model, pytorch, llama, llama2, llamacpp
Last push: 2024-12-03T20:52:37+00:00

## Health v2 (maintenance only)
Score: 18/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 8, longevity 77
- inputs: {"age_days": 1086, "days_push": 638, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1405, forks 88 (observed 2026-08-28T04:04:38.089041+00:00)

## What it is
A web-based calculator that estimates GPU memory requirements and inference/finetuning throughput (token/s) for any LLM. It supports quantization schemes like GGML, bitsandbytes, and QLoRA, and inference frameworks such as vLLM, llama.cpp, and HuggingFace.

## Use cases
- check if my gpu can run this llm
- estimate vram needed for a language model
- calculate tokens per second for llm inference
- estimate finetuning time for llama model
- which quantization fits on my gpu
- max context length and batch size my gpu can handle
- breakdown of gpu memory usage during llm training

## When to choose
- planning hardware purchases or cloud GPU selection for LLM inference or finetuning
- comparing quantization options (GGML, bnb, QLoRA) for memory fit
- estimating token/s and training iteration time before running experiments

## When to avoid
- you need exact benchmarked performance numbers rather than estimates
- you need a CLI/API for automated pipelines - it is an interactive web tool
- models or frameworks outside its supported set (vLLM, llama.cpp, HF)

## Facets
- artifact type: application
- maturity: active
- function: gpu-computing, llm-inference, llm-training, machine-learning, developer-tools
- domain: large-language-models, gpu-computing, machine-learning, developer-tools
- platform: browser
- tags: gpu-memory-calculator, vram-estimator, tokens-per-second, quantization, llama-cpp, qlora, bitsandbytes, vllm, huggingface, finetuning-estimator, web, javascript

## Member repositories
- RahulSChand/gpu_poor (main) score 18

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:38.089041+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:38:45.920266+00:00, confidence not recorded.
  - readme: https://github.com/RahulSChand/gpu_poor (fetched 2026-08-28T04:04:38.089041+00:00, sha d864cf3f77a2)
  - homepage: https://rahulschand.github.io/gpu_poor/ (fetched 2026-08-29T11:52:42.239679+00:00, sha 56f35c116423)
- Data as of 2026-08-30T08:39:29.467469+00:00.