Workbench / Online
BUILD YOUR COMPUTE STACK
Start with the model. MuseBoard maps the memory and hardware. Settings are mirrored to the URL, so any configuration can be shared.
Compute workbench
B / Estimated VRAM
Inference4.7GB
Serving: weights + KV cache + runtime overhead. Estimate — not a guarantee.
- Model weights3.8 GB
- KV cache0.47 GB
- Runtime overhead0.38 GB
Memory headroom
Free after estimated load
Utilization
Target ≤ 90% of 80.0 GB
Decode ceiling
Theoretical, batch 1, bandwidth-bound
VRAM usage
4.7 GB / 80.0 GB
Pinned configuration
1 × NVIDIA A100 80GB
- Total VRAM
- 80.0 GB
- Topology
- Single GPU
- Interconnect
- NVLink 3 · 600 GB/s
- Board power
- 400 W
Compatibility / 1 × A100 80GB
- Ready
Inference
4.7 GB · 6% of 1 × A100 80GB
- Ready
Fine-tuning
13.2 GB · 17% of 1 × A100 80GB
- Limited
Training
142.5 GB · Needs 2 × A100 80GB
Alternative configurations
- 01SERVE / READY
Inference
Estimate memory requirements for serving models.
weights + KV cache + overhead
Open in workbench - 02ADAPT / LORA
Fine-tuning
Explore memory requirements for adapting existing models.
frozen base + adapters + activations
Open in workbench - 03TRAIN / ADAM
Training
Understand large-scale compute requirements.
weights + grads + optimizer + activations
Open in workbench
MuseBoard gives first-order estimates for planning conversations. It does not replace profiling, capacity testing or detailed ML infrastructure planning.