Some suggestions for the graphs

[PierBover]

As a serverless fan I love these benchmarks!

I have a couple of suggestions.

1) Make the single graph much bigger

As it is, it's impossible to see much in there.

2) Use the same scale on all individual graphs

When comparing those graphs one gets the wrong impression that all services have similar max values. Of course when reading the actual millisecond values at the top we see that this is not the case but it's not evident unless you look for that.

3) Use a thousands separator on the millisecond values

Without a separator it's harder to distinguish between say 1000 and 10000 specially when there are so many numbers crammed in a little space.

Would you accept a PR with these changes?

[Strernd]

To 1 and 2) These were discussed before. The problem is that the numbers are so wildly different, that you can't get anything from the graphs if you use the same scale. (See the single graph).
There is no good solution for that. At least I have no idea how to make this readable.

[PierBover]

What about adding an extra bar chart using the data from the millisecond values?

That would give a better representation of the data.