Granite 3.1 1B A400m Base — Hardware Requirements & GPU Compatibility
Chat
1.7K downloads
10 likesSpecifications
- Publisher
- IBM
- Parameters
- 1.3B
- Architecture
- GraniteMoeForCausalLM
- Context Length
- 131,072 tokens
- Vocabulary Size
- 49,152
- Release Date
- 2025-01-31
- License
- Apache 2.0
Get Started
HuggingFace
How Much VRAM Does Granite 3.1 1B A400m Base Need?
Select a quantization to see compatible GPUs below.
| Quantization | Bits | VRAM | + Context | File Size | Quality |
|---|---|---|---|---|---|
| BF16 | 16.00 | 3.1 GB | 9.4 GB | 2.67 GB | Brain floating point 16 — preferred for training |
Which GPUs Can Run Granite 3.1 1B A400m Base?
BF16 · 3.1 GBGranite 3.1 1B A400m Base (BF16) requires 3.1 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 4+ GB is recommended. Using the full 131K context window can add up to 6.3 GB, bringing total usage to 9.4 GB. 35 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti.
Runs great
— Plenty of headroomWhich Devices Can Run Granite 3.1 1B A400m Base?
BF16 · 3.1 GB33 devices with unified memory can run Granite 3.1 1B A400m Base, including NVIDIA DGX H100, NVIDIA DGX A100 640GB.
Runs great
— Plenty of headroomRelated Models
Frequently Asked Questions
- How much VRAM does Granite 3.1 1B A400m Base need?
Granite 3.1 1B A400m Base requires 3.1 GB of VRAM at BF16. Full 131K context adds up to 6.3 GB (9.4 GB total).
VRAM = Weights + KV Cache + Overhead
Weights = 1.3B × 16 bits ÷ 8 = 2.7 GB
KV Cache + Overhead ≈ 0.4 GB (at 2K context + ~0.3 GB framework)
KV Cache + Overhead ≈ 6.7 GB (at full 131K context)
VRAM usage by quantization
BF163.1 GBBF16 + full context9.4 GB- Can I run Granite 3.1 1B A400m Base on a Mac?
Granite 3.1 1B A400m Base requires at least 3.1 GB at BF16, which exceeds the unified memory of most consumer Macs. You would need a Mac Studio or Mac Pro with a high-memory configuration.
- Can I run Granite 3.1 1B A400m Base locally?
Yes — Granite 3.1 1B A400m Base can run locally on consumer hardware. At BF16 quantization it needs 3.1 GB of VRAM. Popular tools include Ollama, LM Studio, and llama.cpp.
- How fast is Granite 3.1 1B A400m Base?
At BF16, Granite 3.1 1B A400m Base can reach ~950 tok/s on AMD Instinct MI300X. On NVIDIA GeForce RTX 4090: ~213 tok/s. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.
tok/s = (bandwidth GB/s ÷ model GB) × efficiency
Example: AMD Instinct MI300X → 5300 ÷ 3.1 × 0.55 = ~950 tok/s
Estimated speed at BF16 (3.1 GB)
AMD Instinct MI300X~950 tok/sNVIDIA GeForce RTX 4090~213 tok/sNVIDIA H100 SXM~710 tok/sAMD Instinct MI250X~587 tok/sReal-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
- What's the download size of Granite 3.1 1B A400m Base?
At BF16, the download is about 2.67 GB.
- Which GPUs can run Granite 3.1 1B A400m Base?
35 consumer GPUs can run Granite 3.1 1B A400m Base at BF16 (3.1 GB). Top options include AMD Radeon RX 6700 XT, AMD Radeon RX 6800, AMD Radeon RX 6800 XT. 35 GPUs have plenty of headroom for comfortable inference.
- Which devices can run Granite 3.1 1B A400m Base?
33 devices with unified memory can run Granite 3.1 1B A400m Base at BF16 (3.1 GB), including Mac Mini M4 (16 GB), Mac Mini M4 (32 GB), Mac Mini M4 Pro (24 GB), Mac Mini M4 Pro (48 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.