Computation Settings: Performance and Optimisation

The cloud compute limits

Every Spherene computation runs on Spherene's cloud infrastructure. Two hard limits apply:

Maximum execution time: 45 minutes. Any computation that exceeds this is automatically halted.
Memory ceiling: fixed per computation. Very low DRT values on large envelopes will hit this limit.

These limits apply across all plugins (Rhino, Grasshopper, Fusion, nTop). Planning your DRT and output type strategy around these limits is the key to efficient computation.

DRT: the most important performance lever

DRT has a larger impact on compute time and memory than any other setting. The relationship is non-linear:

DRT change	Compute time impact	Memory impact
Halved (e.g. 0.4 to 0.2 mm)	Up to 8x longer	Up to 6x more
Quartered (e.g. 0.4 to 0.1 mm)	Up to 64x longer	Up to 36x more

The recommended strategy is to work at 0.4 mm or higher during all design iteration, and only reduce DRT for the final production export.

Output type and quality

The output type you request has a significant impact on compute time:

Output type / mode	Relative compute cost	When to use
Single Surface only	Lowest	All design iteration. Fast feedback on density and geometry.
Solid Surface Draft	Moderate	Reviewing wall thickness and boundary transitions during refinement.
Solid Surface Production (High Quality)	High	Final export only. Do not use during iteration.
Labyrinths or Halfspaces	Higher	Only when specifically needed for CFD or downstream analysis.
Single Surface + Solid Surface Production	Highest	Never use during iteration. Final export only.

Random Seed

The Random Seed controls the stochastic variation in the ADMS geometry. At the same density, DRT, and field settings, different seed values produce geometries that are equivalent in performance but visually different at the cell level.

Leave at the default (1) for most work. Change it when you want to explore visual variation at fixed parameters, for example to find the most visually appealing version of a design for a render or presentation.

Maximum execution time

This setting in the compute dialog sets the maximum allowed runtime before the cloud engine halts the job. The ceiling is 45 minutes regardless of what you set.

Setting a lower value (5 to 10 minutes) is useful during parameter exploration: if the job is going to fail due to memory or complexity, it fails fast rather than waiting 45 minutes. Once you have a stable parameter set, increase the limit for the final production run.

Mesh reduction

Most plugins offer a mesh reduction option that reduces the polygon count of the output mesh by a defined percentage. This does not affect the ADMS geometry quality; it simplifies the mesh for downstream processing.

Use mesh reduction when:

The output mesh is very large and slowing down your slicer or FEA preprocessor.
You are exporting for visualisation rather than printing.
You are doing a quick FEA preview and do not need full resolution.

Start with 10 to 20% reduction and increase if the quality remains acceptable. For final print exports, leave reduction at 0%.

Practical workflow strategy

Phase	DRT	Output type	Quality mode	Max time
Exploration	0.6 to 1.0 mm	Single Surface	N/A	5 to 10 min
Design refinement	0.4 mm	Single Surface + Solid Surface	Draft	15 to 20 min
Final production export	0.2 to 0.4 mm	Solid Surface	High Quality	30 to 45 min
CFD preparation	0.4 mm	Labyrinths	Standard	20 to 30 min

Tip: Always start a new project at a high DRT and work downward. It is much faster to reduce DRT at the end when the design is final than to iterate at high resolution throughout.