The most significant factor is the "bar size" setting for chart. At the
"1-hour" setting, the memory usage would be 720 times less compared to the
"5-second" setting. So, set this judiciously. For the long term performance
charts (such as 1-year), I set it to 1-hour bar. If I want to focus on a
particular day, select the "start/end dates" in the backtester, and then
you can go as low s 5-seconds.

On Sat, Dec 15, 2012 at 7:52 PM, Judson Wilson <[email protected]>wrote:

> Just thinking out loud here.... maybe you need to force garbage collection?
>
> On Sat, Dec 15, 2012 at 2:47 PM, Klaus <[email protected]> wrote:
> > I did it with Xmx256M and it does indeed seem to flatten (at about 300
> MB).
> > Also with Xmx150M
> > it flattened at about 260M.
> > So, I will try to investigate further, why and under what conditions it
> does
> > fail..
> >
> > Klaus
> >
> >
> > Am Sonntag, 9. Dezember 2012 13:18:58 UTC-5 schrieb nonlinear:
> >>
> >> The default starting option for JBT is "aggressive heap":
> >> -XX:+AggressiveHeap
> >> This is meant for the use when you are optimizing. When backtesting,
> >> forward-testing, and trading, you may want to use a less aggressive
> option,
> >> for example:
> >> -Xmx=256M
> >>
> >> Can you try that and see if you still observe the same issue as you
> >> reported? Thanks.
> >>
> >> On Sat, Dec 8, 2012 at 9:55 PM, Klaus <[email protected]> wrote:
> >>>
> >>> Well, what I observe is that when I run a strategy with JBT, it (of
> >>> course allocates memory), when next thing I run a different strategy
> more
> >>> memory is allocated.
> >>> I just tried it with the sample data and the distribution JBT. Only
> >>> change was to create 20 copies of the Sample strategy.  It rapidly
> builds up
> >>> from 250 to about 350 MB when more strategies are run.
> >>> There the curve flattens at some point.
> >>> However it is a small data set. I have the problem that if I do the
> same
> >>> with my installation and my dataset  I see a much larger initial
> >>> memory allocation and actually a similar build up of memory allocation.
> >>> Thus, at some point it hits the wall and I have to restart JBT to run
> more
> >>> strategies. This seems to be in line with the aymptotic build up of
> memory
> >>> use.
> >>>
> >>> Klaus
> >>>
> >>>
> >>> Am Montag, 3. Dezember 2012 09:27:58 UTC-5 schrieb nonlinear:
> >>>>
> >>>> Charts can (potentially) use enormous amounts of memory, because all
> of
> >>>> the historical data with regards to prices and indicators is
> preserved. For
> >>>> that reason, in JBT, you can show charts only one at a time. That is,
> the
> >>>> chart becomes available after you run a backtest for a given strategy
> A. If
> >>>> subsequently you run a backtest for strategy B, the chart for B
> becomes
> >>>> available, and the chart for A is no longer in memory.
> >>>>
> >>>> On Sun, Dec 2, 2012 at 8:19 PM, Klaus <[email protected]> wrote:
> >>>>>
> >>>>> It seems that
> >>>>> a) if I start a back test of a different strategy, I can no longer
> >>>>> display an earlier chart.
> >>>>>     (says: run the strategy first)
> >>>>> b) on the other hand, it seems that if I run a number of different
> >>>>> strategy, the memory
> >>>>> is later on not (completely) freed. - I thought this would be the
> >>>>> charting data.
> >>>>>
> >>>>> While both make from their respective point of view sense, both at
> the
> >>>>> same time do not
> >>>>> make much sense. - Or is the loss of memory s.th. different?
> >>>>>
> >>>>> Klaus
> >>>>>
> >>>>> --
> >>>>> You received this message because you are subscribed to the Google
> >>>>> Groups "JBookTrader" group.
> >>>>> To view this discussion on the web visit
> >>>>> https://groups.google.com/d/msg/jbooktrader/-/_UDw8OUOQLQJ.
> >>>>> To post to this group, send email to [email protected].
> >>>>> To unsubscribe from this group, send email to
> >>>>> [email protected].
> >>>>>
> >>>>> For more options, visit this group at
> >>>>> http://groups.google.com/group/jbooktrader?hl=en.
> >>>>
> >>>>
> >>> --
> >>> You received this message because you are subscribed to the Google
> Groups
> >>> "JBookTrader" group.
> >>> To view this discussion on the web visit
> >>> https://groups.google.com/d/msg/jbooktrader/-/BrvC38YGFrkJ.
> >>>
> >>> To post to this group, send email to [email protected].
> >>> To unsubscribe from this group, send email to
> >>> [email protected].
> >>> For more options, visit this group at
> >>> http://groups.google.com/group/jbooktrader?hl=en.
> >>
> >>
> > --
> > You received this message because you are subscribed to the Google Groups
> > "JBookTrader" group.
> > To view this discussion on the web visit
> > https://groups.google.com/d/msg/jbooktrader/-/H6rI2sRGd_wJ.
> >
> > To post to this group, send email to [email protected].
> > To unsubscribe from this group, send email to
> > [email protected].
> > For more options, visit this group at
> > http://groups.google.com/group/jbooktrader?hl=en.
>
> --
> You received this message because you are subscribed to the Google Groups
> "JBookTrader" group.
> To post to this group, send email to [email protected].
> To unsubscribe from this group, send email to
> [email protected].
> For more options, visit this group at
> http://groups.google.com/group/jbooktrader?hl=en.
>
>

-- 
You received this message because you are subscribed to the Google Groups 
"JBookTrader" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/jbooktrader?hl=en.

Reply via email to