Provide LDBC benchmark for SF 1000

[whitewum]

Is your feature request related to a problem? Please describe.

I saw that the LDBC benchmark data set is small and generator.scaleFactor = 30.

According to LDBC snb datagen, the largest data set can be SF1000 (and 3000+ WIP)

Describe the solution you'd like

Can you provide the benchmark results on SF1000?

Describe alternatives you've considered
A clear and concise description of any alternative solutions or features you've considered.

Additional context
Add any other context or screenshots about the feature request here.

[zhengpingq]

Hi, @whitewum

Thanks a lot for raising this issue!

We have only included the results using SF30 so that our benchmark results are easy to reproduce--it is faster to load a relatively small dataset.

Having said that, we have done extensive evaluation¹ for the interactive engine in GraphScope, reported in Section 6 of this paper (preprint), which includes performance analysis in terms of query latency and scalability, as well as microbenchmarks for key system mechanisms supporting parallel graph-traversal computation.

Please let us know if further information is needed. Thank you!

¹. The tests have been performed using internal development branch, which will be released as open source in the coming year.