Microdeck.tools 100% no navegador

Calculadora de VRAM para LLMs locais

New

Calcule quanta VRAM um modelo de linguagem precisa em FP16, 8 bits ou 4 bits com um comprimento de contexto e veja se cabe na sua GPU.

100% no navegador Sem envio ao servidor Resultados na hora

Esta ferramenta economizou seu tempo? Ajude a manter o Microdeck grátis e atualizado. Opcional · A partir de $3.

Como funciona por dentro

Model weights take parameters × bits per weight ÷ 8 bytes. The KV cache grows with context: 2 × layers × KV heads × head dimension × tokens × batch × bytes per value. The tool adds a runtime overhead of about 10% of the weights plus 0.5 GiB for CUDA context and activations, then compares the total with the VRAM you have.

Passo a passo de uso

1

Pick a preset model or choose custom and enter parameters, layers, KV heads and head dimension.

2

Choose the quantization (FP16, Q8, Q6, Q5, Q4 or Q3 style) and the KV cache precision.

3

Set the context length and batch size you plan to use.

4

Enter the VRAM per GPU and the number of GPUs to see whether the model fits and how much is left.

Principais recursos e vantagens

  • Presets for Llama 3.1, Mistral 7B, Qwen2.5 and Gemma 2 with their real layer and KV head counts
  • Separate lines for weights, KV cache and overhead so you see what to cut first
  • Fit check against common GPU sizes and multi-GPU setups
  • Runs locally; nothing you type is sent anywhere

Metodologia, fórmulas e padrões técnicos

Effective bits per weight for GGUF-style quantizations are averages (Q4_K_M is about 4.85 bits, Q8_0 about 8.5). Real files differ slightly by architecture.

Casos de uso práticos

  • Deciding whether a 24 GB card can run a 32B model at 4-bit with a long context.
  • Sizing a multi-GPU server before renting hardware.
  • Seeing how much memory a longer context window costs.

Exemplos resolvidos e casos especiais

Llama 3.1 8B KV cache

Entrada
Llama 3.1 8B, 8,192 tokens, batch 1, FP16 cache
Saída
1.00 GiB of KV cache

128 KiB per token: 2 × 32 layers × 8 KV heads × 128 dims × 2 bytes.

Weights of a 70B model

Entrada
70B parameters at 16 bits
Saída
140 GB of weights

Far beyond one consumer GPU, which is why 70B is usually run at 4-bit or split across cards.

Limitações e notas de precisão

  • • This is an estimate. Inference engines differ in overhead, and activation memory depends on the backend and batch shape.
  • • It does not model offloading to system RAM, tensor-parallel duplication or training memory (gradients and optimizer state).

Processamento local no navegador e privacidade

Conversores comuns enviam seus documentos, códigos ou vídeos para servidores externos. O Calculadora de VRAM para LLMs locais roda direto na sua aba:

Processado nesta aba

Seus textos e arquivos são processados localmente sem enviar conteúdo para servidores.

Sem registro de conteúdo

O Microdeck não registra nem armazena arquivos ou dados colados nas ferramentas.

Memória temporária da aba

Os resultados ficam apenas na memória do navegador e somem ao fechar ou limpar a aba.

Perguntas frequentes (FAQ)

Why does the KV cache matter?

It grows linearly with context length and batch size. At very long contexts it can rival the weights, so a model that fits at 4K tokens may not fit at 128K.

Does a 4-bit model lose quality?

Usually a little. Q4_K_M and Q5 variants are widely used because the quality loss is small compared with the memory saved, but it depends on the model and task.

Can I run a model that is larger than my VRAM?

Some runtimes offload layers to system RAM, which works but is much slower. This tool only checks whether everything fits on the GPU.

Outras ferramentas relacionadas

Apoiar o Microdeck

Totalmente opcional · A partir de $3 · Sem cadastro

Escolha um valor (USD):Selecionado:$5

O valor mínimo é de $3 USD.

100% opcional · A partir de $3 · Checkout seguro no Gumroad.