Trying to better understand RAM usage

[rayko]

Hello nicbizz,

What about "Keep best # results" parameter. Default is 10.

Given the number iterations your testing, reducing this parameter
usually makes a dramatic impact on RAM, wall clock processing time.

[NinjaTrader_Jesse]

Hello nicbizz,

That may be due to the code you are using in the strategy. I would suggest trying a simple test like the samplemacrossover and see if you can note a considerable difference. If not the code in your script is likely causing that. you can comment out your code to test different setups to see what is using the memory.

[nicbizz]

I have a strategy where a backtest on 1054 instruments, 60 minute chart from Jan 1, 2016 to Dec 31, 2024 uses up approximately 14GB in RAM, and takes about 2 hours to complete (1054 iterations).

However, when I run an optimization whereby I introduce a single alternate to a single variable (so 2108 iterations), it uses a whopping 800GB in RAM, and takes 10 hours to complete (IsInstantiatedOnEachOptimizationIteration is set to False).

I'm trying to understand why there is such a huge disparity in RAM usage (57x) and time taken (5x) when the number of iterations is only doubled. From a layman point of view, I would expect it to only use twice the amount of resources.

I guess I'm asking this to find out if there's a fundamental deficiency in my code that is causing the optimization to blow up, or this is just the way NT8's engine works, and I should just focus my efforts on how to work with it.

Thanks in advance.