NerdFace is an interactive LLM VRAM, RAM, and KV cache sizing calculator. Enable JavaScript to run real-time hardware fit simulations and query Hugging Face tensor metadata.
☰
✓
NerdFace
🔍
✕
All
★
Best Fit
Context
4K
8K
16K
32K
64K
128K
256K
512K
1M
KV
FP16
Q8_0
Q4_0
NVMe
★
☀️
Light
🌙
Dark
💻
Auto
💾
Save Hardware Preset
Preset Name
Topology:
VRAM / Pool:
Sys RAM:
OS Reserve:
Saved in LocalStorage:
✕
Cancel
Save
Sort:
Throughput
↓
↑
Speed
↓
↑
Fit
↓
↑
RAM
↓
↑
Scale
↓
↑
Name
↓
↑
Memory Bar:
OS
VRAM
RAM
KV
✓
⚠️ deranked
VRAM:
Sys RAM:
Unified:
RAM:
Speed:
Score:
Loading more… (
/
)
Model & Creator
↕
↑
↓
Scale
↕
↑
↓
Hardware Fit
↕
↑
↓
OS
VRAM
RAM
KV
Total RAM (W/KV)
↕
↑
↓
Est. Speed
↕
↑
↓
Throughput Index
↕
↓
↑
Run
✓
⚠️ deranked
VRAM:
Sys RAM:
Unified:
/
Loading more… (
/
)
👻 Oops, nothing found here.
Querying Hugging Face API & Sizing KV Cache Matrices...