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
Inference162.6GB
Serving: weights + KV cache + runtime overhead. Estimate — not a guarantee.
- Model weights145.4 GB
- KV cache2.7 GB
- Runtime overhead14.5 GB
Memory headroom
Free after estimated load
Utilization
Target ≤ 90% of 192.0 GB
Decode ceiling
Theoretical, batch 1, bandwidth-bound
VRAM usage
162.6 GB / 192.0 GB
Recommended configuration
1 × NVIDIA B200 192GB
- Total VRAM
- 192.0 GB
- Topology
- Single GPU
- Interconnect
- NVLink 5 · 1.8 TB/s
- Board power
- 1,000 W
Compatibility / 1 × B200
- Ready
Inference
162.6 GB · 85% of 1 × B200
- Limited
Fine-tuning
186.1 GB · Needs 2 × B200
- Limited
Training
1,299 GB · Needs 8 × B200
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.