Skip to content

Deriving Data from TaskId

Sending data from serial to parallel is often costly. To optimize performance, we aim to send the smallest possible amount of data. One technique is to leverage taskId to derive other information without transmitting it explicitly.

This page will guide you through deriving various kinds of information with minimal data transmission

X and Y from taskId

Given a grid or matrix, you can calculate the X and Y positions from the taskId and width of the grid.

local function getXYFromTaskId(taskId: number, width: number)
    return taskId % width, taskId // width -- x, y
end
Example usage:
worker.server.luau
Parallelizer.ListenToTask(Actor, 'XY', function(taskId: number, localMemory: SharedTable) 
    local x, y = GetXYFromTaskId(taskId, localMemory[1]) -- Assuming the width is stored in the local memory

    -- ... do something
end)

Info

For this to work, the thread count dispatched needs to be equal to width * height