There is no reason to use system-wide lock that you know from thread-based processing. System wide locks slows down whole system, see for reference:
In Tickator is global tick the locking mechanism. Because of rules there is no way two ticklets may contend for the same data:
There are situations when it is necessary to use locking in Tickator. For exaple you create complex ticklet component with many ticklets inside and you want to assure that no new input handling will be started unless previous is finished. This situation may be always solved by local lock composed of ticklets.
Because Tickator obeys all fundamental rules, it may be executed by any number of units - threads, CPU cores, processes, separate CPUs, … There is no chance two units will contend for the same data. So if you have single core CPU in your computer - fine. If you have 32 CPU beast - fine.
In far future we may reach optimal state - there will be 1:1 mapping between CPU core and ticklet. But there is a long way to have it. Imagine searching for maximal value that uses for N values (N/2)*(N/4) ticklets in shape of triangle, each comparing two values. That would be fastest algorithm possible - O(log2(N)).
In our case will be max value found in log2(8)=3 ticks. Table below depicts activity of ticklets during time.
Ticklets are executed on change. So if there is no change in input there is no need to recompute anything. That can save a lot of performance. You probably know the TwoHardThings sentence about cache. This is not true in Tickator - caching is for free! Problem with naming things unfortunatelly persists.
For example take ticklet Add, that sum two numbers from output A and B. Then stores output to RES. Every time any of inputs change, there is new result in next tick. This result is propagated to connected ticklets. But only on change. If nothing is changed, nothing is executed.
In thread-based parallelism there is no determinism. Every time you execute your program will thread process data in different order and timing. In Tickator will computation behave exactly the same - if initial conditions are same. You will ever know that for some input there should be output in exactly 124 ticks. This property can be used for testing and security purposes.
There is no need to explictly mark parallelism. Just place two ticklets next to each other, connect them to inputs and they will work in parallel.
If you need to do sequential task it is easy - just connect outputs of one ticklet to inputs of another… And you have sequence.
I am afraid this goal will not be reach immediatelly after start of Tickator project. But in future programmer should just create components, connect them to other in diagram, etc. No need to write text source code, compile it, etc. Imagine touch screen where you just compose your program from prepared components.
Ticlet component has strictly defined inputs and outputs and there is no other way to influence its behavior. Therefore you can replace component with another one with the same inputs and outputs.