Please notice :

There is no need to sort out the memory issues. They all have been solved (but are awaiting a final test with my larger account (40+ charts / lists) which will probably happen today). However, it looks like the PRT executable(s) have been handed to IG 6 months ago, and that never received an update. Again, I just recognize all the things I created tickets for long gone, and which have been solved for PRT-IB (which is PRT itself and a different entity than PRT IG (different servers etc.).

If you want to use PRT-IG for some longer than 5 minutes when the USA stock market is open, then apply at least the button you see below. This will most probably be insufficient, so apply the bottom button. That will keep on working throughout.  However, that would be for a moderately small environment (like 10 charts / lists).

So the leaking is in the data stream. Lower that and less leaks occur.

What you see in the 2nd attachment is normal memory usage. The most important is that it stays like that and not grows and grows.
The phenomena which affect it are (the loading of) indicators, the amount of candle data (thus, the more charts and lists, the more data has to be pushed through).

When V12 was out a couple of months (say May 2023) it was already possible to backtest during the weekends (no data then); right from the start that already was not possible. When you’d backtest today and during the day time (markets open), I guess that not more than one run can be done before your memory will be eaten (200K bars assumed). Notice that here too I just assume an old version, so when that is refreshed by PRT all could be good instantly.

You should always have TaskManager open, ready to kill the obtrusive PRT process, because the memory usage is so harsh, that once it crosses a threshold your system will be totally unresponsive. A click with the mouse on the Java process followed by End Task, may still take a minute (you won’t see anything happen), but will work in the end. Then your system will recover.

Important : How it really turns out depends on your amount of installed memory in combination with the setting of how much one Java instance may consume. De default is 6GB, but PRT is out of control (in this V12 version for IG) and it will go beyond that. At 8GB things start to be bad.
I had that 6GB increased to 10GB for all my accounts, which is a BAD thing to do with the 32GB of internal memory I have. Obviously now each account can use 10GB, which again goes bananas and will go up to 14GB. For the PRT instance(s) this is too bad (the instance itself will stall), but with more accounts now your installed physical memory will be totally gone and all dies. And point is : you don’t see *that* coming. That’s why the TaskManager has to be ready (you won’t be able to start it when not).

image_2023-12-01_083139356.png image_2023-12-01_083552642.png

That’s an other question;
I am quite confident it has been solved to a certain degree (as told, I was to do the final test for the larger account, and for example, I can set back that normal Push mode already in the smaller account, but whether the large account will load in a reasonable time, I can only know when I start testing that.
Please notice I can’t test that at this moment yet, because of that “reset” thing which needs to happen (the “one time” copying of settings from V11 to V12).

So that pipe line is there, but we also need to interest people over at  PRT-(I think mainly IG), to implement it and for IG, re-TEST it. They may not like to do this right away again.