[Feature request] Review graph rendering algorithm

[madcampos]

When loading large graphs, like LLMs, it can be very slow to load such graphs to the point of freezing model-explorer. This issue is an example of what happens: #50

And from investigating slow to load graphs, it appears that the layouting step is what takes the longest.

Recently an article came out from the SpiderMonkey team on exactly this problem: layouting graphs, specially graphs that represent code.

My suggestion then is to evaluate if it is feasible to use a similar strategy from what is proposed on the SpiderMonkey article for layout to improve performance.

[gaikwadrahul8]

Hi @madcampos,
I apologize for the delayed response, Thank you for the thorough research and for surfacing the SpiderMonkey IonGraph article this is exactly the kind of well reasoned performance optimization proposal we value. You're correct that layout computation is a bottleneck for large graphs. Before committing to architectural changes we need to profile whether the issue is layout vs our GPU accelerated rendering, and evaluate how IonGraph's domain specific approach applies to our per layer lazy layout strategy.

I seems like we need to investigate this and provide an update on feasibility. Thank you for your cooperation and patience.