Benchmark Scoreboard

Important

All benchmarks on this page include the following:

• 50 players online
• Only the test code shown below is used on the server, no other scripts
• Plugins on the server include Skript, SkBee, Spark and StressTestBots (used to create fake players to test)

Periodical Scoreboard:

This scoreboard updates all players every tick.
This is generally not a good way to do it for multiple reasons:

• Every tick is a bit (ok a LOT) overkill
• We're setting lines of a scoreboard that aren't even changing

Code

on join:
	# this is just to setup our fake scores
	set {-var::one::%uuid of player%} to 1
	set {-var::two::%uuid of player%} to 2
	set {-var::three::%uuid of player%} to 3
	set {-var::four::%uuid of player%} to 4

every tick:
	loop all players:
		set title of scoreboard of loop-player to "this is a title"
		set line 15 of scoreboard of loop-player to "&7Player:"
		set line 14 of scoreboard of loop-player to name of loop-player
		set line 13 of scoreboard of loop-player to "Points:"
		set line 12 of scoreboard of loop-player to "&7- &b%{-var::one::%uuid of loop-player%}%"
		set line 11 of scoreboard of loop-player to "Score:"
		set line 10 of scoreboard of loop-player to "&7- &b%{-var::two::%uuid of loop-player%}%"
		set line 9 of scoreboard of loop-player to "KDR:"
		set line 8 of scoreboard of loop-player to "&7- &b%{-var::three::%uuid of loop-player%}%"
		set line 7 of scoreboard of loop-player to "Blerp:"
		set line 6 of scoreboard of loop-player to "&7- &b%{-var::four::%uuid of loop-player%}%"
		set line 5 of scoreboard of loop-player to "IP: play.myserver.com"

Spark Profile:

Here is how that code affects the server. Not too bad, but could be better.

While Loop Scoreboard:

This way to do scoreboards spreads out the work load a bit.

Code

on join:
	while player is online:
		set title of scoreboard of player to "this is a title"
		set line 15 of scoreboard of player to "&7Player:"
		set line 14 of scoreboard of player to name of player
		set line 13 of scoreboard of player to "Points:"
		set line 12 of scoreboard of player to "&7- &b%{-var::points::%uuid of player%}%"
		set line 11 of scoreboard of player to "Score:"
		set line 10 of scoreboard of player to "&7- &b%{-var::score::%uuid of player%}%"
		set line 9 of scoreboard of player to "KDR:"
		set line 8 of scoreboard of player to "&7- &b%{-var::kdr::%uuid of player%}%"
		set line 7 of scoreboard of player to "Blerp:"
		set line 6 of scoreboard of player to "&7- &b%{-var::blerp::%uuid of player%}%"
		set line 5 of scoreboard of player to "IP: play.myserver.com"
		wait 1 tick

Spark Profile:

Ouchie momma! This one is even higher than the previous one, nearly twice as stressful.
This would be caused due to a massive amount of waits (internally these are scheduled tasks).

On The Fly Scoreboard:

In this version, we have created a scoreboard function and only update it when something on the scoreboard actually needs updating.

Code

function updateBoard(p: player):
	set title of scoreboard of {_p} to "this is a title"
	set line 15 of scoreboard of {_p} to "&7Player:"
	set line 14 of scoreboard of {_p} to name of {_p}
	set line 13 of scoreboard of {_p} to "Points:"
	set line 12 of scoreboard of {_p} to "&7- &b%{-var::points::%uuid of {_p}%}%"
	set line 11 of scoreboard of {_p} to "Score:"
	set line 10 of scoreboard of {_p} to "&7- &b%{-var::score::%uuid of {_p}%}%"
	set line 9 of scoreboard of {_p} to "KDR:"
	set line 8 of scoreboard of {_p} to "&7- &b%{-var::kdr::%uuid of {_p}%}%"
	set line 7 of scoreboard of {_p} to "Blerp:"
	set line 6 of scoreboard of {_p} to "&7- &b%{-var::blerp::%uuid of {_p}%}%"
	set line 5 of scoreboard of {_p} to "IP: play.myserver.com"

on join:
	updateBoard(player)

on break:
	add 1 to {-var::blerp::%uuid of player%}
	updateBoard(player)

on place:
	add 1 to {-var::points::%uuid of player%}
	updateBoard(player)

on death of player:
	if attacker is a player:
		add 1 to {-var::kdr::%uuid of victim%}
		add 1 to {-var::score::%uuid of attacker%}
		updateBoard(attacker)
		updateBoard(victim)

Spark Profile:

As you can see in this image, there are significant performance improvements.
(The redacted part is just spark, not significant for this benchmark)