I know this is a (very) old topic, but I can confirm it is definitely the GRAPH function causing this. This concerns a so-called memory leak.
Amount of memory approaching the limit.
(verdi55 there talked about it already)
Today (or at least the 6 months or so ago in V11.1) this problem is still there.
What is harder to grasp for us guys, is that this relates to the amount of graphing itself. Thus indeed the more “lines” graphed, the more memory is consumed (and not freed). Also, lines (graphs) may be conditional (like a line showing that you have a position, vs not showing because no position).
In PRT/IG only 5 graphs are allowed in parallel for a reason : their smaller server. So with IG you just can not use as much memory as you can in native PRT. Maybe someone notice the 10 times or so slower IG server services PRT than native PRT ? it just is so … (IG is incredibly slow in comparison).
Confusing : Despite what I just said, notice that this graph data is done/maintained locally, BUT it is steered server side.
Since V11 more Graphs are allowed in native PRT, just as that more historical data is allowed (at least with a Premium account this is so). But it is useless, because you can only use it a few times before the memory message kicks in again. Btw, closing down PRT and restarting it, helps. But it obviously does not help if you consume too much in the first place (like using too much historical data).
Over time I learned to live with it by means of having groups of Graphs active and using them based on variables denoting those groups.
If GraphGroup1Active then
Graph Gain(…)
Graph BrokerFee(…)
Endif
Using the commands sparsely will let me backtest for hours, but eventually the message will be there again and it will be time to restart PRT.
Notice that once this message appeared a first time, your backtesting will run significantly slower (there’s a lack of memory indeed). If you continue (always crossing away the message), finally PRT will hang.
Peter
