About memory usage
Author: akuzn
Creation Date: 8/18/2012 6:58 AM
profile picture

akuzn

#1
1.Seems WealthLab can work with dll's compiled by Visual Studio with FrameWork v.3 and v 3.5 option set.
Does it give anything in memory usage?

2. I ve put all DataSeries and metods in special class, compiled in librabry using 2.0 Framework . Strategy creates needed quantity of objects for each symbol i use. All signals are analyzed in main trading cycle by calling metods of this class. I have never before used any destructor or Bars.Cache.Clear. Everything was working normally without it.
I declared in this class different scales of Bars and DataSeries.
Seems 8-10 Strategies loaded are executed and optimized without memory problems. But if i ll use more strategies i ll have this problem.

Do i need to create destructor to free memory, or (and) put there Bars.Cache.Clear? Will it free 5-15% of memory, can it speed up execution and optimization of 8 strategies by 5-8%?
profile picture

Cone

#2
Patience for another month. WLD 6.4, compiled on .NET 4.0 is slated for a mid-Sept. release. Take another look after upgrading.

Generally, my experience with .NET is that manual attempts to free memory do not have any effect on the GC.
profile picture

akuzn

#3
Thank You. At least i know i did all i could.
profile picture

Eugene

#4
(I heard that the GC performance was considerably improved in .NET 4.5, especially for large objects. However, support for the just released 4.5 is not on the radar.)
profile picture

akuzn

#5
During last month i ve recompiled my library many times
So results of memory usage are like zebra.
One time full memory usage, next time 50%. Changes - for example i change only 2 calls, add 2 methods or smthg like that,
WealthLab uses only 55 % of memory).
Setup is the same - 12 strategies loaded, 2-3 optmizing.
I was thinking it was depending on FrameWork version - 2.0 or 3.5. In reality - not. Seems hidden logic of dll assembly depnding of +/- additional phrases.
This website uses cookies to improve your experience. We'll assume you're ok with that, but you can opt-out if you wish (Read more).