Improve current plots

[liamsi]

Add description/title/legend to current plots:

At least:

• some meaningful title
• x-axis: DA sample #
• y-axis: time in ms or sec

Nice to have:

• meta data about the node, e.g. in the title (e.g. digital ocean region - as regions seem to show quite different behavior in terms of latency)

[evan-forbes]

Is this close to what you had in mind?

930cf8a

[liamsi]

Yeah, this is super dope! And the additional per region plot is very useful to understand the latency differences between regions.

Some questions/feedback on the first plot:

• isn't "100 ms" easier to understand than "tenths of seconds"
• the x-axis should be integers, where 1,2,3, ... means the first DA proof, the second DA proof etc
• question: did you run with 20 samples as one DA proof?
• "Sample Grouping" could be changed to: "DA proof (15 sample requests)" (or 20 sample requests or how many you grouped together in the experiment)
• question: do you want to group all results of one region into a single plot? If yes, that isn't really necessary right now but also useful! It would additionally be interesting to have a plot per node (and the plot carries the info of the droplets region in its title) and one similar plot for all nodes

[evan-forbes]

I only ran 15 samples per DA proof, or whatever the default was. However, I plotted with 20 samples per groups by mistake.

Are you looking for something similar to what is in the "errorbar_plotter.py" script for each region, or something that averages the latency for each proof across all regions? Something like this maybe?

[liamsi]

Are you looking for something similar to what is in the "errorbar_plotter.py" script for each region, or something that averages the latency for each proof across all regions? Something like this maybe?

I think a plot per node/droplet, and a plot that averages over all nodes (and optionally a plot per region if much more nodes than different regions were used in an experiment) would be ideal. And yes, your plot looks perfect IMO 👌🏼

[liamsi]

I think all of this was addressed in #9. Closing this and if anything is missing will open smaller followup issues.
Thanks @evan-forbes!