Skip to content
MUSEBOARD

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

Workbench / Online
Model / Gemma 27BUnits / GB (10⁹ B)

A / Model configuration

Google
B

Editing the count switches to a custom model with an estimated architecture.

Precision0.5 B / param
Context length8,192 tokens
Batch sizeConcurrent sequences
Workload
Pinned

Architecture / published config

Layers
46
Hidden
4,608
KV heads
16 × 128
KV / token
368 KB

B / Estimated VRAM

Inference

18.0GB

Serving: weights + KV cache + runtime overhead. Estimate — not a guarantee.

  • Model weights13.6 GB
  • KV cache3.1 GB
  • Runtime overhead1.4 GB

Memory headroom

62.0 GB

Free after estimated load

Utilization

23%

Target ≤ 90% of 80.0 GB

Decode ceiling

~246 tok/s

Theoretical, batch 1, bandwidth-bound

VRAM usage

18.0 GB / 80.0 GB

0 GB80 GB

Pinned configuration

1 × NVIDIA H100 SXM 80GB

Total VRAM
80.0 GB
Topology
Single GPU
Interconnect
NVLink 4 · 900 GB/s
Board power
700 W

Compatibility / 1 × H100 SXM

  • Inference

    18.0 GB · 23% of 1 × H100 SXM

    Ready
  • Fine-tuning

    31.8 GB · 40% of 1 × H100 SXM

    Ready
  • Training

    493.2 GB · Needs 8 × H100 SXM

    Limited

Alternative configurations